diff options
Diffstat (limited to 'lib')
102 files changed, 15547 insertions, 1521 deletions
diff --git a/lib/Makefile.in b/lib/Makefile.in index a907254cd46..ac32c7070f3 100644 --- a/lib/Makefile.in +++ b/lib/Makefile.in @@ -79,9 +79,15 @@ endif Makefile: ../config.status $(srcdir)/Makefile.in $(MAKE) -C .. src/$@ -libgnu_a_OBJECTS = $(gl_LIBOBJS) \ +# Object modules that need not be built for Emacs. +# Emacs does not need e-regex.o (it has its own regex-emacs.c), +# and building it would just waste time. +not_emacs_OBJECTS = regex.o + +libgnu_a_OBJECTS = fingerprint.o $(gl_LIBOBJS) \ $(patsubst %.c,%.o,$(filter %.c,$(libgnu_a_SOURCES))) -libegnu_a_OBJECTS = $(patsubst %.o,e-%.o,$(libgnu_a_OBJECTS)) +for_emacs_OBJECTS = $(filter-out $(not_emacs_OBJECTS),$(libgnu_a_OBJECTS)) +libegnu_a_OBJECTS = $(patsubst %.o,e-%.o,$(for_emacs_OBJECTS)) $(libegnu_a_OBJECTS) $(libgnu_a_OBJECTS): $(BUILT_SOURCES) @@ -112,7 +118,7 @@ TAGS: $(ETAGS) $(tagsfiles) .PHONY: $(ETAGS) tags clean: - rm -f *.[ao] *-t \#* $(DEPDIR)/* + rm -f ./*.[ao] ./*-t \#* $(DEPDIR)/* mostlyclean: clean rm -f $(filter-out %-t,$(MOSTLYCLEANFILES)) distclean bootstrap-clean: mostlyclean diff --git a/lib/_Noreturn.h b/lib/_Noreturn.h index c44ad89b7c0..db9b45554c5 100644 --- a/lib/_Noreturn.h +++ b/lib/_Noreturn.h @@ -1,8 +1,31 @@ -#if !defined _Noreturn && __STDC_VERSION__ < 201112 -# if (3 <= __GNUC__ || (__GNUC__ == 2 && 8 <= __GNUC_MINOR__) \ - || 0x5110 <= __SUNPRO_C) +/* A C macro for declaring that a function does not return. + Copyright (C) 2011-2019 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify it + under the terms of the GNU General Public License as published + by the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <https://www.gnu.org/licenses/>. */ + +#ifndef _Noreturn +# if (defined __cplusplus \ + && ((201103 <= __cplusplus && !(__GNUC__ == 4 && __GNUC_MINOR__ == 7)) \ + || (defined _MSC_VER && 1900 <= _MSC_VER))) +# define _Noreturn [[noreturn]] +# elif ((!defined __cplusplus || defined __clang__) \ + && (201112 <= (defined __STDC_VERSION__ ? __STDC_VERSION__ : 0) \ + || 4 < __GNUC__ + (7 <= __GNUC_MINOR__))) + /* _Noreturn works as-is. */ +# elif 2 < __GNUC__ + (8 <= __GNUC_MINOR__) || 0x5110 <= __SUNPRO_C # define _Noreturn __attribute__ ((__noreturn__)) -# elif 1200 <= _MSC_VER +# elif 1200 <= (defined _MSC_VER ? _MSC_VER : 0) # define _Noreturn __declspec (noreturn) # else # define _Noreturn diff --git a/lib/acl-internal.c b/lib/acl-internal.c index 99ebe811051..cc42183f443 100644 --- a/lib/acl-internal.c +++ b/lib/acl-internal.c @@ -23,7 +23,7 @@ #include "acl-internal.h" -#if USE_ACL && HAVE_ACL_GET_FILE +#if USE_ACL && HAVE_ACL_GET_FILE /* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ # if HAVE_ACL_TYPE_EXTENDED /* Mac OS X */ @@ -37,7 +37,7 @@ acl_extended_nontrivial (acl_t acl) return (acl_entries (acl) > 0); } -# else /* Linux, FreeBSD, IRIX, Tru64 */ +# else /* Linux, FreeBSD, IRIX, Tru64, Cygwin >= 2.5 */ /* ACL is an ACL, from a file, stored as type ACL_TYPE_ACCESS. Return 1 if the given ACL is non-trivial. @@ -51,7 +51,7 @@ acl_access_nontrivial (acl_t acl) at least, allowing us to write return (3 < acl_entries (acl)); but the following code is more robust. */ -# if HAVE_ACL_FIRST_ENTRY /* Linux, FreeBSD */ +# if HAVE_ACL_FIRST_ENTRY /* Linux, FreeBSD, Cygwin >= 2.5 */ acl_entry_t ace; int got_one; @@ -124,7 +124,7 @@ acl_default_nontrivial (acl_t acl) # endif -#elif USE_ACL && HAVE_FACL && defined GETACL /* Solaris, Cygwin, not HP-UX */ +#elif USE_ACL && HAVE_FACL && defined GETACL /* Solaris, Cygwin < 2.5, not HP-UX */ /* Test an ACL retrieved with GETACL. Return 1 if the given ACL, consisting of COUNT entries, is non-trivial. @@ -355,7 +355,7 @@ acl_nontrivial (int count, struct acl_entry *entries) struct acl_entry *ace = &entries[i]; if (ace->uid != ACL_NSUSER && ace->gid != ACL_NSGROUP) - return 1; + return 1; } return 0; } @@ -479,7 +479,7 @@ void free_permission_context (struct permission_context *ctx) { #if USE_ACL -# if HAVE_ACL_GET_FILE /* Linux, FreeBSD, Mac OS X, IRIX, Tru64 */ +# if HAVE_ACL_GET_FILE /* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ if (ctx->acl) acl_free (ctx->acl); # if !HAVE_ACL_TYPE_EXTENDED @@ -487,7 +487,7 @@ free_permission_context (struct permission_context *ctx) acl_free (ctx->default_acl); # endif -# elif defined GETACL /* Solaris, Cygwin */ +# elif defined GETACL /* Solaris, Cygwin < 2.5 */ free (ctx->entries); # ifdef ACE_GETACL free (ctx->ace_entries); diff --git a/lib/acl-internal.h b/lib/acl-internal.h index 883f06a3986..05833efaa89 100644 --- a/lib/acl-internal.h +++ b/lib/acl-internal.h @@ -30,7 +30,8 @@ # define GETACLCNT ACL_CNT #endif -/* On Linux, additional ACL related API is available in <acl/libacl.h>. */ +/* On Linux and Cygwin >= 2.5, additional ACL related API is available in + <acl/libacl.h>. */ #ifdef HAVE_ACL_LIBACL_H # include <acl/libacl.h> #endif @@ -72,7 +73,7 @@ _GL_INLINE_HEADER_BEGIN # if HAVE_ACL_GET_FILE /* POSIX 1003.1e (draft 17 -- abandoned) specific version. */ -/* Linux, FreeBSD, Mac OS X, IRIX, Tru64 */ +/* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ # ifndef MIN_ACL_ENTRIES # define MIN_ACL_ENTRIES 4 @@ -122,7 +123,10 @@ rpl_acl_set_fd (int fd, acl_t acl) # endif /* Linux-specific */ -# ifndef HAVE_ACL_EXTENDED_FILE +/* Cygwin >= 2.5 implements this function, but it returns 1 for all + directories, thus is unusable. */ +# if !defined HAVE_ACL_EXTENDED_FILE || defined __CYGWIN__ +# undef HAVE_ACL_EXTENDED_FILE # define HAVE_ACL_EXTENDED_FILE false # define acl_extended_file(name) (-1) # endif @@ -163,7 +167,7 @@ extern int acl_access_nontrivial (acl_t); extern int acl_default_nontrivial (acl_t); # endif -# elif HAVE_FACL && defined GETACL /* Solaris, Cygwin, not HP-UX */ +# elif HAVE_FACL && defined GETACL /* Solaris, Cygwin < 2.5, not HP-UX */ /* Set to 0 if a file's mode is stored independently from the ACL. */ # if defined __CYGWIN__ /* Cygwin */ @@ -256,14 +260,14 @@ extern int acl_nontrivial (int count, struct acl *entries); struct permission_context { mode_t mode; #if USE_ACL -# if HAVE_ACL_GET_FILE /* Linux, FreeBSD, Mac OS X, IRIX, Tru64 */ +# if HAVE_ACL_GET_FILE /* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ acl_t acl; # if !HAVE_ACL_TYPE_EXTENDED acl_t default_acl; # endif bool acls_not_supported; -# elif defined GETACL /* Solaris, Cygwin */ +# elif defined GETACL /* Solaris, Cygwin < 2.5 */ int count; aclent_t *entries; # ifdef ACE_GETACL @@ -293,10 +297,6 @@ struct permission_context { int get_permissions (const char *, int, mode_t, struct permission_context *); int set_permissions (struct permission_context *, const char *, int); -void free_permission_context (struct permission_context *) -#if ! (defined USE_ACL && (HAVE_ACL_GET_FILE || defined GETACL)) - _GL_ATTRIBUTE_CONST -#endif - ; +void free_permission_context (struct permission_context *); _GL_INLINE_HEADER_END diff --git a/lib/acl_entries.c b/lib/acl_entries.c index b58b4db0ee7..19ac3c3bd9f 100644 --- a/lib/acl_entries.c +++ b/lib/acl_entries.c @@ -22,7 +22,7 @@ #include "acl-internal.h" /* This file assumes POSIX-draft like ACLs - (Linux, FreeBSD, Mac OS X, IRIX, Tru64). */ + (Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5). */ /* Return the number of entries in ACL. Return -1 and set errno upon failure to determine it. */ @@ -34,7 +34,7 @@ acl_entries (acl_t acl) if (acl != NULL) { -#if HAVE_ACL_FIRST_ENTRY /* Linux, FreeBSD, Mac OS X */ +#if HAVE_ACL_FIRST_ENTRY /* Linux, FreeBSD, Mac OS X, Cygwin >= 2.5 */ # if HAVE_ACL_TYPE_EXTENDED /* Mac OS X */ /* acl_get_entry returns 0 when it successfully fetches an entry, and -1/EINVAL at the end. */ @@ -45,7 +45,7 @@ acl_entries (acl_t acl) got_one >= 0; got_one = acl_get_entry (acl, ACL_NEXT_ENTRY, &ace)) count++; -# else /* Linux, FreeBSD */ +# else /* Linux, FreeBSD, Cygwin >= 2.5 */ /* acl_get_entry returns 1 when it successfully fetches an entry, and 0 at the end. */ acl_entry_t ace; diff --git a/lib/alloca.in.h b/lib/alloca.in.h index 0e3622f1b96..a581d58f834 100644 --- a/lib/alloca.in.h +++ b/lib/alloca.in.h @@ -1,7 +1,7 @@ /* Memory allocation on the stack. - Copyright (C) 1995, 1999, 2001-2004, 2006-2019 Free Software - Foundation, Inc. + Copyright (C) 1995, 1999, 2001-2004, 2006-2019 Free Software Foundation, + Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published @@ -36,6 +36,12 @@ #ifndef alloca # ifdef __GNUC__ + /* Some version of mingw have an <alloca.h> that causes trouble when + included after 'alloca' gets defined as a macro. As a workaround, include + this <alloca.h> first and define 'alloca' as a macro afterwards. */ +# if (defined _WIN32 && ! defined __CYGWIN__) && @HAVE_ALLOCA_H@ +# include_next <alloca.h> +# endif # define alloca __builtin_alloca # elif defined _AIX # define alloca __alloca diff --git a/lib/binary-io.h b/lib/binary-io.h index 763eddba56a..720b08c7551 100644 --- a/lib/binary-io.h +++ b/lib/binary-io.h @@ -1,6 +1,5 @@ /* Binary mode I/O. - Copyright (C) 2001, 2003, 2005, 2008-2019 Free Software Foundation, - Inc. + Copyright (C) 2001, 2003, 2005, 2008-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -48,10 +47,8 @@ _GL_INLINE_HEADER_BEGIN /* Use a function rather than a macro, to avoid gcc warnings "warning: statement with no effect". */ BINARY_IO_INLINE int -__gl_setmode (int fd, int mode) +__gl_setmode (int fd _GL_UNUSED, int mode _GL_UNUSED) { - (void) fd; - (void) mode; return O_BINARY; } #endif @@ -60,7 +57,7 @@ __gl_setmode (int fd, int mode) extern int __gl_setmode_check (int); #else BINARY_IO_INLINE int -__gl_setmode_check (int fd) { return 0; } +__gl_setmode_check (int fd _GL_UNUSED) { return 0; } #endif /* Set FD's mode to MODE, which should be either O_TEXT or O_BINARY. diff --git a/lib/c-ctype.h b/lib/c-ctype.h index d207bd01a18..4d521763843 100644 --- a/lib/c-ctype.h +++ b/lib/c-ctype.h @@ -5,8 +5,7 @@ <ctype.h> functions' behaviour depends on the current locale set via setlocale. - Copyright (C) 2000-2003, 2006, 2008-2019 Free Software Foundation, - Inc. + Copyright (C) 2000-2003, 2006, 2008-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/c-strcasecmp.c b/lib/c-strcasecmp.c index 620f87566a8..ec50f1abe76 100644 --- a/lib/c-strcasecmp.c +++ b/lib/c-strcasecmp.c @@ -1,6 +1,5 @@ /* c-strcasecmp.c -- case insensitive string comparator in C locale - Copyright (C) 1998-1999, 2005-2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1998-1999, 2005-2006, 2009-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/c-strncasecmp.c b/lib/c-strncasecmp.c index 1d284e62366..513c353f6c2 100644 --- a/lib/c-strncasecmp.c +++ b/lib/c-strncasecmp.c @@ -1,6 +1,5 @@ /* c-strncasecmp.c -- case insensitive string comparator in C locale - Copyright (C) 1998-1999, 2005-2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1998-1999, 2005-2006, 2009-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/careadlinkat.c b/lib/careadlinkat.c index ad75bad3083..e56d5030856 100644 --- a/lib/careadlinkat.c +++ b/lib/careadlinkat.c @@ -1,7 +1,7 @@ /* Read symbolic links into a buffer without size limitation, relative to fd. - Copyright (C) 2001, 2003-2004, 2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2001, 2003-2004, 2007, 2009-2019 Free Software Foundation, + Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/cdefs.h b/lib/cdefs.h new file mode 100644 index 00000000000..96d26164199 --- /dev/null +++ b/lib/cdefs.h @@ -0,0 +1,514 @@ +/* Copyright (C) 1992-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#ifndef _SYS_CDEFS_H +#define _SYS_CDEFS_H 1 + +/* We are almost always included from features.h. */ +#ifndef _FEATURES_H +# include <features.h> +#endif + +/* The GNU libc does not support any K&R compilers or the traditional mode + of ISO C compilers anymore. Check for some of the combinations not + anymore supported. */ +#if defined __GNUC__ && !defined __STDC__ +# error "You need a ISO C conforming compiler to use the glibc headers" +#endif + +/* Some user header file might have defined this before. */ +#undef __P +#undef __PMT + +#ifdef __GNUC__ + +/* All functions, except those with callbacks or those that + synchronize memory, are leaf functions. */ +# if __GNUC_PREREQ (4, 6) && !defined _LIBC +# define __LEAF , __leaf__ +# define __LEAF_ATTR __attribute__ ((__leaf__)) +# else +# define __LEAF +# define __LEAF_ATTR +# endif + +/* GCC can always grok prototypes. For C++ programs we add throw() + to help it optimize the function calls. But this works only with + gcc 2.8.x and egcs. For gcc 3.2 and up we even mark C functions + as non-throwing using a function attribute since programs can use + the -fexceptions options for C code as well. */ +# if !defined __cplusplus && __GNUC_PREREQ (3, 3) +# define __THROW __attribute__ ((__nothrow__ __LEAF)) +# define __THROWNL __attribute__ ((__nothrow__)) +# define __NTH(fct) __attribute__ ((__nothrow__ __LEAF)) fct +# define __NTHNL(fct) __attribute__ ((__nothrow__)) fct +# else +# if defined __cplusplus && __GNUC_PREREQ (2,8) +# define __THROW throw () +# define __THROWNL throw () +# define __NTH(fct) __LEAF_ATTR fct throw () +# define __NTHNL(fct) fct throw () +# else +# define __THROW +# define __THROWNL +# define __NTH(fct) fct +# define __NTHNL(fct) fct +# endif +# endif + +#else /* Not GCC. */ + +# if (defined __cplusplus \ + || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)) +# define __inline inline +# else +# define __inline /* No inline functions. */ +# endif + +# define __THROW +# define __THROWNL +# define __NTH(fct) fct + +#endif /* GCC. */ + +/* Compilers that are not clang may object to + #if defined __clang__ && __has_extension(...) + even though they do not need to evaluate the right-hand side of the &&. */ +#if defined __clang__ && defined __has_extension +# define __glibc_clang_has_extension(ext) __has_extension (ext) +#else +# define __glibc_clang_has_extension(ext) 0 +#endif + +/* These two macros are not used in glibc anymore. They are kept here + only because some other projects expect the macros to be defined. */ +#define __P(args) args +#define __PMT(args) args + +/* For these things, GCC behaves the ANSI way normally, + and the non-ANSI way under -traditional. */ + +#define __CONCAT(x,y) x ## y +#define __STRING(x) #x + +/* This is not a typedef so `const __ptr_t' does the right thing. */ +#define __ptr_t void * + + +/* C++ needs to know that types and declarations are C, not C++. */ +#ifdef __cplusplus +# define __BEGIN_DECLS extern "C" { +# define __END_DECLS } +#else +# define __BEGIN_DECLS +# define __END_DECLS +#endif + + +/* Fortify support. */ +#define __bos(ptr) __builtin_object_size (ptr, __USE_FORTIFY_LEVEL > 1) +#define __bos0(ptr) __builtin_object_size (ptr, 0) + +#if __GNUC_PREREQ (4,3) +# define __warndecl(name, msg) \ + extern void name (void) __attribute__((__warning__ (msg))) +# define __warnattr(msg) __attribute__((__warning__ (msg))) +# define __errordecl(name, msg) \ + extern void name (void) __attribute__((__error__ (msg))) +#else +# define __warndecl(name, msg) extern void name (void) +# define __warnattr(msg) +# define __errordecl(name, msg) extern void name (void) +#endif + +/* Support for flexible arrays. + Headers that should use flexible arrays only if they're "real" + (e.g. only if they won't affect sizeof()) should test + #if __glibc_c99_flexarr_available. */ +#if defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L && !defined __HP_cc +# define __flexarr [] +# define __glibc_c99_flexarr_available 1 +#elif __GNUC_PREREQ (2,97) +/* GCC 2.97 supports C99 flexible array members as an extension, + even when in C89 mode or compiling C++ (any version). */ +# define __flexarr [] +# define __glibc_c99_flexarr_available 1 +#elif defined __GNUC__ +/* Pre-2.97 GCC did not support C99 flexible arrays but did have + an equivalent extension with slightly different notation. */ +# define __flexarr [0] +# define __glibc_c99_flexarr_available 1 +#else +/* Some other non-C99 compiler. Approximate with [1]. */ +# define __flexarr [1] +# define __glibc_c99_flexarr_available 0 +#endif + + +/* __asm__ ("xyz") is used throughout the headers to rename functions + at the assembly language level. This is wrapped by the __REDIRECT + macro, in order to support compilers that can do this some other + way. When compilers don't support asm-names at all, we have to do + preprocessor tricks instead (which don't have exactly the right + semantics, but it's the best we can do). + + Example: + int __REDIRECT(setpgrp, (__pid_t pid, __pid_t pgrp), setpgid); */ + +#if defined __GNUC__ && __GNUC__ >= 2 + +# define __REDIRECT(name, proto, alias) name proto __asm__ (__ASMNAME (#alias)) +# ifdef __cplusplus +# define __REDIRECT_NTH(name, proto, alias) \ + name proto __THROW __asm__ (__ASMNAME (#alias)) +# define __REDIRECT_NTHNL(name, proto, alias) \ + name proto __THROWNL __asm__ (__ASMNAME (#alias)) +# else +# define __REDIRECT_NTH(name, proto, alias) \ + name proto __asm__ (__ASMNAME (#alias)) __THROW +# define __REDIRECT_NTHNL(name, proto, alias) \ + name proto __asm__ (__ASMNAME (#alias)) __THROWNL +# endif +# define __ASMNAME(cname) __ASMNAME2 (__USER_LABEL_PREFIX__, cname) +# define __ASMNAME2(prefix, cname) __STRING (prefix) cname + +/* +#elif __SOME_OTHER_COMPILER__ + +# define __REDIRECT(name, proto, alias) name proto; \ + _Pragma("let " #name " = " #alias) +*/ +#endif + +/* GCC has various useful declarations that can be made with the + `__attribute__' syntax. All of the ways we use this do fine if + they are omitted for compilers that don't understand it. */ +#if !defined __GNUC__ || __GNUC__ < 2 +# define __attribute__(xyz) /* Ignore */ +#endif + +/* At some point during the gcc 2.96 development the `malloc' attribute + for functions was introduced. We don't want to use it unconditionally + (although this would be possible) since it generates warnings. */ +#if __GNUC_PREREQ (2,96) +# define __attribute_malloc__ __attribute__ ((__malloc__)) +#else +# define __attribute_malloc__ /* Ignore */ +#endif + +/* Tell the compiler which arguments to an allocation function + indicate the size of the allocation. */ +#if __GNUC_PREREQ (4, 3) +# define __attribute_alloc_size__(params) \ + __attribute__ ((__alloc_size__ params)) +#else +# define __attribute_alloc_size__(params) /* Ignore. */ +#endif + +/* At some point during the gcc 2.96 development the `pure' attribute + for functions was introduced. We don't want to use it unconditionally + (although this would be possible) since it generates warnings. */ +#if __GNUC_PREREQ (2,96) +# define __attribute_pure__ __attribute__ ((__pure__)) +#else +# define __attribute_pure__ /* Ignore */ +#endif + +/* This declaration tells the compiler that the value is constant. */ +#if __GNUC_PREREQ (2,5) +# define __attribute_const__ __attribute__ ((__const__)) +#else +# define __attribute_const__ /* Ignore */ +#endif + +/* At some point during the gcc 3.1 development the `used' attribute + for functions was introduced. We don't want to use it unconditionally + (although this would be possible) since it generates warnings. */ +#if __GNUC_PREREQ (3,1) +# define __attribute_used__ __attribute__ ((__used__)) +# define __attribute_noinline__ __attribute__ ((__noinline__)) +#else +# define __attribute_used__ __attribute__ ((__unused__)) +# define __attribute_noinline__ /* Ignore */ +#endif + +/* Since version 3.2, gcc allows marking deprecated functions. */ +#if __GNUC_PREREQ (3,2) +# define __attribute_deprecated__ __attribute__ ((__deprecated__)) +#else +# define __attribute_deprecated__ /* Ignore */ +#endif + +/* Since version 4.5, gcc also allows one to specify the message printed + when a deprecated function is used. clang claims to be gcc 4.2, but + may also support this feature. */ +#if __GNUC_PREREQ (4,5) || \ + __glibc_clang_has_extension (__attribute_deprecated_with_message__) +# define __attribute_deprecated_msg__(msg) \ + __attribute__ ((__deprecated__ (msg))) +#else +# define __attribute_deprecated_msg__(msg) __attribute_deprecated__ +#endif + +/* At some point during the gcc 2.8 development the `format_arg' attribute + for functions was introduced. We don't want to use it unconditionally + (although this would be possible) since it generates warnings. + If several `format_arg' attributes are given for the same function, in + gcc-3.0 and older, all but the last one are ignored. In newer gccs, + all designated arguments are considered. */ +#if __GNUC_PREREQ (2,8) +# define __attribute_format_arg__(x) __attribute__ ((__format_arg__ (x))) +#else +# define __attribute_format_arg__(x) /* Ignore */ +#endif + +/* At some point during the gcc 2.97 development the `strfmon' format + attribute for functions was introduced. We don't want to use it + unconditionally (although this would be possible) since it + generates warnings. */ +#if __GNUC_PREREQ (2,97) +# define __attribute_format_strfmon__(a,b) \ + __attribute__ ((__format__ (__strfmon__, a, b))) +#else +# define __attribute_format_strfmon__(a,b) /* Ignore */ +#endif + +/* The nonnull function attribute marks pointer parameters that + must not be NULL. Do not define __nonnull if it is already defined, + for portability when this file is used in Gnulib. */ +#ifndef __nonnull +# if __GNUC_PREREQ (3,3) +# define __nonnull(params) __attribute__ ((__nonnull__ params)) +# else +# define __nonnull(params) +# endif +#endif + +/* If fortification mode, we warn about unused results of certain + function calls which can lead to problems. */ +#if __GNUC_PREREQ (3,4) +# define __attribute_warn_unused_result__ \ + __attribute__ ((__warn_unused_result__)) +# if defined __USE_FORTIFY_LEVEL && __USE_FORTIFY_LEVEL > 0 +# define __wur __attribute_warn_unused_result__ +# endif +#else +# define __attribute_warn_unused_result__ /* empty */ +#endif +#ifndef __wur +# define __wur /* Ignore */ +#endif + +/* Forces a function to be always inlined. */ +#if __GNUC_PREREQ (3,2) +/* The Linux kernel defines __always_inline in stddef.h (283d7573), and + it conflicts with this definition. Therefore undefine it first to + allow either header to be included first. */ +# undef __always_inline +# define __always_inline __inline __attribute__ ((__always_inline__)) +#else +# undef __always_inline +# define __always_inline __inline +#endif + +/* Associate error messages with the source location of the call site rather + than with the source location inside the function. */ +#if __GNUC_PREREQ (4,3) +# define __attribute_artificial__ __attribute__ ((__artificial__)) +#else +# define __attribute_artificial__ /* Ignore */ +#endif + +/* GCC 4.3 and above with -std=c99 or -std=gnu99 implements ISO C99 + inline semantics, unless -fgnu89-inline is used. Using __GNUC_STDC_INLINE__ + or __GNUC_GNU_INLINE is not a good enough check for gcc because gcc versions + older than 4.3 may define these macros and still not guarantee GNU inlining + semantics. + + clang++ identifies itself as gcc-4.2, but has support for GNU inlining + semantics, that can be checked for by using the __GNUC_STDC_INLINE_ and + __GNUC_GNU_INLINE__ macro definitions. */ +#if (!defined __cplusplus || __GNUC_PREREQ (4,3) \ + || (defined __clang__ && (defined __GNUC_STDC_INLINE__ \ + || defined __GNUC_GNU_INLINE__))) +# if defined __GNUC_STDC_INLINE__ || defined __cplusplus +# define __extern_inline extern __inline __attribute__ ((__gnu_inline__)) +# define __extern_always_inline \ + extern __always_inline __attribute__ ((__gnu_inline__)) +# else +# define __extern_inline extern __inline +# define __extern_always_inline extern __always_inline +# endif +#endif + +#ifdef __extern_always_inline +# define __fortify_function __extern_always_inline __attribute_artificial__ +#endif + +/* GCC 4.3 and above allow passing all anonymous arguments of an + __extern_always_inline function to some other vararg function. */ +#if __GNUC_PREREQ (4,3) +# define __va_arg_pack() __builtin_va_arg_pack () +# define __va_arg_pack_len() __builtin_va_arg_pack_len () +#endif + +/* It is possible to compile containing GCC extensions even if GCC is + run in pedantic mode if the uses are carefully marked using the + `__extension__' keyword. But this is not generally available before + version 2.8. */ +#if !__GNUC_PREREQ (2,8) +# define __extension__ /* Ignore */ +#endif + +/* __restrict is known in EGCS 1.2 and above. */ +#if !__GNUC_PREREQ (2,92) +# if defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L +# define __restrict restrict +# else +# define __restrict /* Ignore */ +# endif +#endif + +/* ISO C99 also allows to declare arrays as non-overlapping. The syntax is + array_name[restrict] + GCC 3.1 supports this. */ +#if __GNUC_PREREQ (3,1) && !defined __GNUG__ +# define __restrict_arr __restrict +#else +# ifdef __GNUC__ +# define __restrict_arr /* Not supported in old GCC. */ +# else +# if defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L +# define __restrict_arr restrict +# else +/* Some other non-C99 compiler. */ +# define __restrict_arr /* Not supported. */ +# endif +# endif +#endif + +#if __GNUC__ >= 3 +# define __glibc_unlikely(cond) __builtin_expect ((cond), 0) +# define __glibc_likely(cond) __builtin_expect ((cond), 1) +#else +# define __glibc_unlikely(cond) (cond) +# define __glibc_likely(cond) (cond) +#endif + +#ifdef __has_attribute +# define __glibc_has_attribute(attr) __has_attribute (attr) +#else +# define __glibc_has_attribute(attr) 0 +#endif + +#if (!defined _Noreturn \ + && (defined __STDC_VERSION__ ? __STDC_VERSION__ : 0) < 201112 \ + && !__GNUC_PREREQ (4,7)) +# if __GNUC_PREREQ (2,8) +# define _Noreturn __attribute__ ((__noreturn__)) +# else +# define _Noreturn +# endif +#endif + +#if __GNUC_PREREQ (8, 0) +/* Describes a char array whose address can safely be passed as the first + argument to strncpy and strncat, as the char array is not necessarily + a NUL-terminated string. */ +# define __attribute_nonstring__ __attribute__ ((__nonstring__)) +#else +# define __attribute_nonstring__ +#endif + +#if (!defined _Static_assert && !defined __cplusplus \ + && (defined __STDC_VERSION__ ? __STDC_VERSION__ : 0) < 201112 \ + && (!__GNUC_PREREQ (4, 6) || defined __STRICT_ANSI__)) +# define _Static_assert(expr, diagnostic) \ + extern int (*__Static_assert_function (void)) \ + [!!sizeof (struct { int __error_if_negative: (expr) ? 2 : -1; })] +#endif + +/* The #ifndef lets Gnulib avoid including these on non-glibc + platforms, where the includes typically do not exist. */ +#ifndef __WORDSIZE +# include <bits/wordsize.h> +# include <bits/long-double.h> +#endif + +#if defined __LONG_DOUBLE_MATH_OPTIONAL && defined __NO_LONG_DOUBLE_MATH +# define __LDBL_COMPAT 1 +# ifdef __REDIRECT +# define __LDBL_REDIR1(name, proto, alias) __REDIRECT (name, proto, alias) +# define __LDBL_REDIR(name, proto) \ + __LDBL_REDIR1 (name, proto, __nldbl_##name) +# define __LDBL_REDIR1_NTH(name, proto, alias) __REDIRECT_NTH (name, proto, alias) +# define __LDBL_REDIR_NTH(name, proto) \ + __LDBL_REDIR1_NTH (name, proto, __nldbl_##name) +# define __LDBL_REDIR1_DECL(name, alias) \ + extern __typeof (name) name __asm (__ASMNAME (#alias)); +# define __LDBL_REDIR_DECL(name) \ + extern __typeof (name) name __asm (__ASMNAME ("__nldbl_" #name)); +# define __REDIRECT_LDBL(name, proto, alias) \ + __LDBL_REDIR1 (name, proto, __nldbl_##alias) +# define __REDIRECT_NTH_LDBL(name, proto, alias) \ + __LDBL_REDIR1_NTH (name, proto, __nldbl_##alias) +# endif +#endif +#if !defined __LDBL_COMPAT || !defined __REDIRECT +# define __LDBL_REDIR1(name, proto, alias) name proto +# define __LDBL_REDIR(name, proto) name proto +# define __LDBL_REDIR1_NTH(name, proto, alias) name proto __THROW +# define __LDBL_REDIR_NTH(name, proto) name proto __THROW +# define __LDBL_REDIR_DECL(name) +# ifdef __REDIRECT +# define __REDIRECT_LDBL(name, proto, alias) __REDIRECT (name, proto, alias) +# define __REDIRECT_NTH_LDBL(name, proto, alias) \ + __REDIRECT_NTH (name, proto, alias) +# endif +#endif + +/* __glibc_macro_warning (MESSAGE) issues warning MESSAGE. This is + intended for use in preprocessor macros. + + Note: MESSAGE must be a _single_ string; concatenation of string + literals is not supported. */ +#if __GNUC_PREREQ (4,8) || __glibc_clang_prereq (3,5) +# define __glibc_macro_warning1(message) _Pragma (#message) +# define __glibc_macro_warning(message) \ + __glibc_macro_warning1 (GCC warning message) +#else +# define __glibc_macro_warning(msg) +#endif + +/* Generic selection (ISO C11) is a C-only feature, available in GCC + since version 4.9. Previous versions do not provide generic + selection, even though they might set __STDC_VERSION__ to 201112L, + when in -std=c11 mode. Thus, we must check for !defined __GNUC__ + when testing __STDC_VERSION__ for generic selection support. + On the other hand, Clang also defines __GNUC__, so a clang-specific + check is required to enable the use of generic selection. */ +#if !defined __cplusplus \ + && (__GNUC_PREREQ (4, 9) \ + || __glibc_clang_has_extension (c_generic_selections) \ + || (!defined __GNUC__ && defined __STDC_VERSION__ \ + && __STDC_VERSION__ >= 201112L)) +# define __HAVE_GENERIC_SELECTION 1 +#else +# define __HAVE_GENERIC_SELECTION 0 +#endif + +#endif /* sys/cdefs.h */ diff --git a/lib/cloexec.c b/lib/cloexec.c index 2b31d26efd6..db425766a0e 100644 --- a/lib/cloexec.c +++ b/lib/cloexec.c @@ -1,7 +1,6 @@ /* cloexec.c - set or clear the close-on-exec descriptor flag - Copyright (C) 1991, 2004-2006, 2009-2019 Free Software Foundation, - Inc. + Copyright (C) 1991, 2004-2006, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/close-stream.c b/lib/close-stream.c index 6de923af69f..5458c4f29f3 100644 --- a/lib/close-stream.c +++ b/lib/close-stream.c @@ -1,7 +1,6 @@ /* Close a stream, with nicer error checking than fclose's. - Copyright (C) 1998-2002, 2004, 2006-2019 Free Software Foundation, - Inc. + Copyright (C) 1998-2002, 2004, 2006-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/diffseq.h b/lib/diffseq.h index 88a6c4d7ada..c6aac3d8120 100644 --- a/lib/diffseq.h +++ b/lib/diffseq.h @@ -1,7 +1,7 @@ /* Analyze differences between two vectors. - Copyright (C) 1988-1989, 1992-1995, 2001-2004, 2006-2019 Free - Software Foundation, Inc. + Copyright (C) 1988-1989, 1992-1995, 2001-2004, 2006-2019 Free Software + Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/dosname.h b/lib/dosname.h index 3aaf104c6ca..c0ab6848a58 100644 --- a/lib/dosname.h +++ b/lib/dosname.h @@ -1,7 +1,6 @@ /* File names on MS-DOS/Windows systems. - Copyright (C) 2000-2001, 2004-2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2000-2001, 2004-2006, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -21,9 +20,8 @@ #ifndef _DOSNAME_H #define _DOSNAME_H -#if (defined _WIN32 || defined __WIN32__ || \ - defined __MSDOS__ || defined __CYGWIN__ || \ - defined __EMX__ || defined __DJGPP__) +#if (defined _WIN32 || defined __CYGWIN__ \ + || defined __EMX__ || defined __MSDOS__ || defined __DJGPP__) /* This internal macro assumes ASCII, but all hosts that support drive letters use ASCII. */ # define _IS_DRIVE_LETTER(C) (((unsigned int) (C) | ('a' - 'A')) - 'a' \ diff --git a/lib/dtotimespec.c b/lib/dtotimespec.c index 65aae10b56e..a8ff04a9859 100644 --- a/lib/dtotimespec.c +++ b/lib/dtotimespec.c @@ -32,20 +32,20 @@ dtotimespec (double sec) if (! (TYPE_MINIMUM (time_t) < sec)) return make_timespec (TYPE_MINIMUM (time_t), 0); else if (! (sec < 1.0 + TYPE_MAXIMUM (time_t))) - return make_timespec (TYPE_MAXIMUM (time_t), TIMESPEC_RESOLUTION - 1); + return make_timespec (TYPE_MAXIMUM (time_t), TIMESPEC_HZ - 1); else { time_t s = sec; - double frac = TIMESPEC_RESOLUTION * (sec - s); + double frac = TIMESPEC_HZ * (sec - s); long ns = frac; ns += ns < frac; - s += ns / TIMESPEC_RESOLUTION; - ns %= TIMESPEC_RESOLUTION; + s += ns / TIMESPEC_HZ; + ns %= TIMESPEC_HZ; if (ns < 0) { s--; - ns += TIMESPEC_RESOLUTION; + ns += TIMESPEC_HZ; } return make_timespec (s, ns); diff --git a/lib/dup2.c b/lib/dup2.c index 583417c0eec..d3aafa458b5 100644 --- a/lib/dup2.c +++ b/lib/dup2.c @@ -1,7 +1,6 @@ /* Duplicate an open file descriptor to a specified file descriptor. - Copyright (C) 1999, 2004-2007, 2009-2019 Free Software Foundation, - Inc. + Copyright (C) 1999, 2004-2007, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -30,7 +29,7 @@ # undef dup2 -# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# if defined _WIN32 && ! defined __CYGWIN__ /* Get declarations of the native Windows API functions. */ # define WIN32_LEAN_AND_MEAN diff --git a/lib/errno.in.h b/lib/errno.in.h index 3bc064602d5..3bd27f1cc9f 100644 --- a/lib/errno.in.h +++ b/lib/errno.in.h @@ -30,7 +30,7 @@ /* On native Windows platforms, many macros are not defined. */ -# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# if defined _WIN32 && ! defined __CYGWIN__ /* These are the same values as defined by MSVC 10, for interoperability. */ @@ -248,7 +248,7 @@ interoperability. */ # define EOWNERDEAD 58 # define ENOTRECOVERABLE 59 -# elif (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# elif defined _WIN32 && ! defined __CYGWIN__ /* We have a conflict here: pthreads-win32 defines these values differently than MSVC 10. It's hairy to decide which one to use. */ # if defined __MINGW32__ && !defined USE_WINDOWS_THREADS diff --git a/lib/euidaccess.c b/lib/euidaccess.c index c7b057b792e..fece4cfc5e0 100644 --- a/lib/euidaccess.c +++ b/lib/euidaccess.c @@ -29,8 +29,11 @@ #include <sys/types.h> #include <sys/stat.h> #include <unistd.h> - -#include "root-uid.h" +#if defined _WIN32 && ! defined __CYGWIN__ +# include <io.h> +#else +# include "root-uid.h" +#endif #if HAVE_LIBGEN_H # include <libgen.h> @@ -84,7 +87,9 @@ euidaccess (const char *file, int mode) return accessx (file, mode, ACC_SELF); #elif HAVE_EACCESS /* FreeBSD */ return eaccess (file, mode); -#else /* Mac OS X, NetBSD, OpenBSD, HP-UX, Solaris, Cygwin, mingw, BeOS */ +#elif defined _WIN32 && ! defined __CYGWIN__ /* mingw */ + return _access (file, mode); +#else /* Mac OS X, NetBSD, OpenBSD, HP-UX, Solaris, Cygwin, BeOS */ uid_t uid = getuid (); gid_t gid = getgid (); diff --git a/lib/explicit_bzero.c b/lib/explicit_bzero.c index a4abc337b59..79ef1860fb7 100644 --- a/lib/explicit_bzero.c +++ b/lib/explicit_bzero.c @@ -27,9 +27,11 @@ #include <string.h> +#if _LIBC /* glibc-internal users use __explicit_bzero_chk, and explicit_bzero redirects to that. */ -#undef explicit_bzero +# undef explicit_bzero +#endif /* Set LEN bytes of S to 0. The compiler will not delete a call to this function, even if S is dead after the call. */ diff --git a/lib/fcntl.c b/lib/fcntl.c index 06d8e51f4b9..51f62ef78a8 100644 --- a/lib/fcntl.c +++ b/lib/fcntl.c @@ -27,12 +27,12 @@ #include <stdarg.h> #include <unistd.h> -#if !HAVE_FCNTL -# define rpl_fcntl fcntl +#ifdef __KLIBC__ +# define INCL_DOS +# include <os2.h> #endif -#undef fcntl -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ /* Get declarations of the native Windows API functions. */ # define WIN32_LEAN_AND_MEAN # include <windows.h> @@ -166,93 +166,18 @@ dupfd (int oldfd, int newfd, int flags) } #endif /* W32 */ +/* Forward declarations, because we '#undef fcntl' in the middle of this + compilation unit. */ +/* Our implementation of fcntl (fd, F_DUPFD, target). */ +static int rpl_fcntl_DUPFD (int fd, int target); +/* Our implementation of fcntl (fd, F_DUPFD_CLOEXEC, target). */ +static int rpl_fcntl_DUPFD_CLOEXEC (int fd, int target); #ifdef __KLIBC__ - -# define INCL_DOS -# include <os2.h> - -static int -klibc_fcntl (int fd, int action, /* arg */...) -{ - va_list arg_ptr; - int arg; - struct stat sbuf; - int result = -1; - - va_start (arg_ptr, action); - arg = va_arg (arg_ptr, int); - result = fcntl (fd, action, arg); - /* EPERM for F_DUPFD, ENOTSUP for others */ - if (result == -1 && (errno == EPERM || errno == ENOTSUP) - && !fstat (fd, &sbuf) && S_ISDIR (sbuf.st_mode)) - { - ULONG ulMode; - - switch (action) - { - case F_DUPFD: - /* Find available fd */ - while (fcntl (arg, F_GETFL) != -1 || errno != EBADF) - arg++; - - result = dup2 (fd, arg); - break; - - /* Using underlying APIs is right ? */ - case F_GETFD: - if (DosQueryFHState (fd, &ulMode)) - break; - - result = (ulMode & OPEN_FLAGS_NOINHERIT) ? FD_CLOEXEC : 0; - break; - - case F_SETFD: - if (arg & ~FD_CLOEXEC) - break; - - if (DosQueryFHState (fd, &ulMode)) - break; - - if (arg & FD_CLOEXEC) - ulMode |= OPEN_FLAGS_NOINHERIT; - else - ulMode &= ~OPEN_FLAGS_NOINHERIT; - - /* Filter supported flags. */ - ulMode &= (OPEN_FLAGS_WRITE_THROUGH | OPEN_FLAGS_FAIL_ON_ERROR - | OPEN_FLAGS_NO_CACHE | OPEN_FLAGS_NOINHERIT); - - if (DosSetFHState (fd, ulMode)) - break; - - result = 0; - break; - - case F_GETFL: - result = 0; - break; - - case F_SETFL: - if (arg != 0) - break; - - result = 0; - break; - - default : - errno = EINVAL; - break; - } - } - - va_end (arg_ptr); - - return result; -} - -# define fcntl klibc_fcntl +/* Adds support for fcntl on directories. */ +static int klibc_fcntl (int fd, int action, /* arg */...); #endif + /* Perform the specified ACTION on the file descriptor FD, possibly using the argument ARG further described below. This replacement handles the following actions, and forwards all others on to the @@ -273,110 +198,35 @@ klibc_fcntl (int fd, int action, /* arg */...) return -1 and set errno. */ int -rpl_fcntl (int fd, int action, /* arg */...) +fcntl (int fd, int action, /* arg */...) +#undef fcntl +#ifdef __KLIBC__ +# define fcntl klibc_fcntl +#endif { va_list arg; int result = -1; va_start (arg, action); switch (action) { - -#if !HAVE_FCNTL case F_DUPFD: { int target = va_arg (arg, int); - result = dupfd (fd, target, 0); + result = rpl_fcntl_DUPFD (fd, target); break; } -#elif FCNTL_DUPFD_BUGGY || REPLACE_FCHDIR - case F_DUPFD: - { - int target = va_arg (arg, int); - /* Detect invalid target; needed for cygwin 1.5.x. */ - if (target < 0 || getdtablesize () <= target) - errno = EINVAL; - else - { - /* Haiku alpha 2 loses fd flags on original. */ - int flags = fcntl (fd, F_GETFD); - if (flags < 0) - { - result = -1; - break; - } - result = fcntl (fd, action, target); - if (0 <= result && fcntl (fd, F_SETFD, flags) == -1) - { - int saved_errno = errno; - close (result); - result = -1; - errno = saved_errno; - } -# if REPLACE_FCHDIR - if (0 <= result) - result = _gl_register_dup (fd, result); -# endif - } - break; - } /* F_DUPFD */ -#endif /* FCNTL_DUPFD_BUGGY || REPLACE_FCHDIR */ case F_DUPFD_CLOEXEC: { int target = va_arg (arg, int); - -#if !HAVE_FCNTL - result = dupfd (fd, target, O_CLOEXEC); - break; -#else /* HAVE_FCNTL */ - /* Try the system call first, if the headers claim it exists - (that is, if GNULIB_defined_F_DUPFD_CLOEXEC is 0), since we - may be running with a glibc that has the macro but with an - older kernel that does not support it. Cache the - information on whether the system call really works, but - avoid caching failure if the corresponding F_DUPFD fails - for any reason. 0 = unknown, 1 = yes, -1 = no. */ - static int have_dupfd_cloexec = GNULIB_defined_F_DUPFD_CLOEXEC ? -1 : 0; - if (0 <= have_dupfd_cloexec) - { - result = fcntl (fd, action, target); - if (0 <= result || errno != EINVAL) - { - have_dupfd_cloexec = 1; -# if REPLACE_FCHDIR - if (0 <= result) - result = _gl_register_dup (fd, result); -# endif - } - else - { - result = rpl_fcntl (fd, F_DUPFD, target); - if (result < 0) - break; - have_dupfd_cloexec = -1; - } - } - else - result = rpl_fcntl (fd, F_DUPFD, target); - if (0 <= result && have_dupfd_cloexec == -1) - { - int flags = fcntl (result, F_GETFD); - if (flags < 0 || fcntl (result, F_SETFD, flags | FD_CLOEXEC) == -1) - { - int saved_errno = errno; - close (result); - errno = saved_errno; - result = -1; - } - } + result = rpl_fcntl_DUPFD_CLOEXEC (fd, target); break; -#endif /* HAVE_FCNTL */ - } /* F_DUPFD_CLOEXEC */ + } #if !HAVE_FCNTL case F_GETFD: { -# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# if defined _WIN32 && ! defined __CYGWIN__ HANDLE handle = (HANDLE) _get_osfhandle (fd); DWORD flags; if (handle == INVALID_HANDLE_VALUE @@ -405,8 +255,183 @@ rpl_fcntl (int fd, int action, /* arg */...) default: { #if HAVE_FCNTL - void *p = va_arg (arg, void *); - result = fcntl (fd, action, p); + switch (action) + { + #ifdef F_BARRIERFSYNC /* macOS */ + case F_BARRIERFSYNC: + #endif + #ifdef F_CHKCLEAN /* macOS */ + case F_CHKCLEAN: + #endif + #ifdef F_CLOSEM /* NetBSD, HP-UX */ + case F_CLOSEM: + #endif + #ifdef F_FLUSH_DATA /* macOS */ + case F_FLUSH_DATA: + #endif + #ifdef F_FREEZE_FS /* macOS */ + case F_FREEZE_FS: + #endif + #ifdef F_FULLFSYNC /* macOS */ + case F_FULLFSYNC: + #endif + #ifdef F_GETCONFINED /* macOS */ + case F_GETCONFINED: + #endif + #ifdef F_GETDEFAULTPROTLEVEL /* macOS */ + case F_GETDEFAULTPROTLEVEL: + #endif + #ifdef F_GETFD /* POSIX */ + case F_GETFD: + #endif + #ifdef F_GETFL /* POSIX */ + case F_GETFL: + #endif + #ifdef F_GETLEASE /* Linux */ + case F_GETLEASE: + #endif + #ifdef F_GETNOSIGPIPE /* macOS */ + case F_GETNOSIGPIPE: + #endif + #ifdef F_GETOWN /* POSIX */ + case F_GETOWN: + #endif + #ifdef F_GETPIPE_SZ /* Linux */ + case F_GETPIPE_SZ: + #endif + #ifdef F_GETPROTECTIONCLASS /* macOS */ + case F_GETPROTECTIONCLASS: + #endif + #ifdef F_GETPROTECTIONLEVEL /* macOS */ + case F_GETPROTECTIONLEVEL: + #endif + #ifdef F_GET_SEALS /* Linux */ + case F_GET_SEALS: + #endif + #ifdef F_GETSIG /* Linux */ + case F_GETSIG: + #endif + #ifdef F_MAXFD /* NetBSD */ + case F_MAXFD: + #endif + #ifdef F_RECYCLE /* macOS */ + case F_RECYCLE: + #endif + #ifdef F_SETFIFOENH /* HP-UX */ + case F_SETFIFOENH: + #endif + #ifdef F_THAW_FS /* macOS */ + case F_THAW_FS: + #endif + /* These actions take no argument. */ + result = fcntl (fd, action); + break; + + #ifdef F_ADD_SEALS /* Linux */ + case F_ADD_SEALS: + #endif + #ifdef F_BADFD /* Solaris */ + case F_BADFD: + #endif + #ifdef F_CHECK_OPENEVT /* macOS */ + case F_CHECK_OPENEVT: + #endif + #ifdef F_DUP2FD /* FreeBSD, AIX, Solaris */ + case F_DUP2FD: + #endif + #ifdef F_DUP2FD_CLOEXEC /* FreeBSD, Solaris */ + case F_DUP2FD_CLOEXEC: + #endif + #ifdef F_DUP2FD_CLOFORK /* Solaris */ + case F_DUP2FD_CLOFORK: + #endif + #ifdef F_DUPFD /* POSIX */ + case F_DUPFD: + #endif + #ifdef F_DUPFD_CLOEXEC /* POSIX */ + case F_DUPFD_CLOEXEC: + #endif + #ifdef F_DUPFD_CLOFORK /* Solaris */ + case F_DUPFD_CLOFORK: + #endif + #ifdef F_GETXFL /* Solaris */ + case F_GETXFL: + #endif + #ifdef F_GLOBAL_NOCACHE /* macOS */ + case F_GLOBAL_NOCACHE: + #endif + #ifdef F_MAKECOMPRESSED /* macOS */ + case F_MAKECOMPRESSED: + #endif + #ifdef F_MOVEDATAEXTENTS /* macOS */ + case F_MOVEDATAEXTENTS: + #endif + #ifdef F_NOCACHE /* macOS */ + case F_NOCACHE: + #endif + #ifdef F_NODIRECT /* macOS */ + case F_NODIRECT: + #endif + #ifdef F_NOTIFY /* Linux */ + case F_NOTIFY: + #endif + #ifdef F_OPLKACK /* IRIX */ + case F_OPLKACK: + #endif + #ifdef F_OPLKREG /* IRIX */ + case F_OPLKREG: + #endif + #ifdef F_RDAHEAD /* macOS */ + case F_RDAHEAD: + #endif + #ifdef F_SETBACKINGSTORE /* macOS */ + case F_SETBACKINGSTORE: + #endif + #ifdef F_SETCONFINED /* macOS */ + case F_SETCONFINED: + #endif + #ifdef F_SETFD /* POSIX */ + case F_SETFD: + #endif + #ifdef F_SETFL /* POSIX */ + case F_SETFL: + #endif + #ifdef F_SETLEASE /* Linux */ + case F_SETLEASE: + #endif + #ifdef F_SETNOSIGPIPE /* macOS */ + case F_SETNOSIGPIPE: + #endif + #ifdef F_SETOWN /* POSIX */ + case F_SETOWN: + #endif + #ifdef F_SETPIPE_SZ /* Linux */ + case F_SETPIPE_SZ: + #endif + #ifdef F_SETPROTECTIONCLASS /* macOS */ + case F_SETPROTECTIONCLASS: + #endif + #ifdef F_SETSIG /* Linux */ + case F_SETSIG: + #endif + #ifdef F_SINGLE_WRITER /* macOS */ + case F_SINGLE_WRITER: + #endif + /* These actions take an 'int' argument. */ + { + int x = va_arg (arg, int); + result = fcntl (fd, action, x); + } + break; + + default: + /* Other actions take a pointer argument. */ + { + void *p = va_arg (arg, void *); + result = fcntl (fd, action, p); + } + break; + } #else errno = EINVAL; #endif @@ -416,3 +441,186 @@ rpl_fcntl (int fd, int action, /* arg */...) va_end (arg); return result; } + +static int +rpl_fcntl_DUPFD (int fd, int target) +{ + int result; +#if !HAVE_FCNTL + result = dupfd (fd, target, 0); +#elif FCNTL_DUPFD_BUGGY || REPLACE_FCHDIR + /* Detect invalid target; needed for cygwin 1.5.x. */ + if (target < 0 || getdtablesize () <= target) + { + result = -1; + errno = EINVAL; + } + else + { + /* Haiku alpha 2 loses fd flags on original. */ + int flags = fcntl (fd, F_GETFD); + if (flags < 0) + result = -1; + else + { + result = fcntl (fd, F_DUPFD, target); + if (0 <= result && fcntl (fd, F_SETFD, flags) == -1) + { + int saved_errno = errno; + close (result); + result = -1; + errno = saved_errno; + } +# if REPLACE_FCHDIR + if (0 <= result) + result = _gl_register_dup (fd, result); +# endif + } + } +#else + result = fcntl (fd, F_DUPFD, target); +#endif + return result; +} + +static int +rpl_fcntl_DUPFD_CLOEXEC (int fd, int target) +{ + int result; +#if !HAVE_FCNTL + result = dupfd (fd, target, O_CLOEXEC); +#else /* HAVE_FCNTL */ +# if defined __HAIKU__ + /* On Haiku, the system fcntl (fd, F_DUPFD_CLOEXEC, target) sets + the FD_CLOEXEC flag on fd, not on target. Therefore avoid the + system fcntl in this case. */ +# define have_dupfd_cloexec -1 +# else + /* Try the system call first, if the headers claim it exists + (that is, if GNULIB_defined_F_DUPFD_CLOEXEC is 0), since we + may be running with a glibc that has the macro but with an + older kernel that does not support it. Cache the + information on whether the system call really works, but + avoid caching failure if the corresponding F_DUPFD fails + for any reason. 0 = unknown, 1 = yes, -1 = no. */ + static int have_dupfd_cloexec = GNULIB_defined_F_DUPFD_CLOEXEC ? -1 : 0; + if (0 <= have_dupfd_cloexec) + { + result = fcntl (fd, F_DUPFD_CLOEXEC, target); + if (0 <= result || errno != EINVAL) + { + have_dupfd_cloexec = 1; +# if REPLACE_FCHDIR + if (0 <= result) + result = _gl_register_dup (fd, result); +# endif + } + else + { + result = rpl_fcntl_DUPFD (fd, target); + if (result >= 0) + have_dupfd_cloexec = -1; + } + } + else +# endif + result = rpl_fcntl_DUPFD (fd, target); + if (0 <= result && have_dupfd_cloexec == -1) + { + int flags = fcntl (result, F_GETFD); + if (flags < 0 || fcntl (result, F_SETFD, flags | FD_CLOEXEC) == -1) + { + int saved_errno = errno; + close (result); + errno = saved_errno; + result = -1; + } + } +#endif /* HAVE_FCNTL */ + return result; +} + +#undef fcntl + +#ifdef __KLIBC__ + +static int +klibc_fcntl (int fd, int action, /* arg */...) +{ + va_list arg_ptr; + int arg; + struct stat sbuf; + int result; + + va_start (arg_ptr, action); + arg = va_arg (arg_ptr, int); + result = fcntl (fd, action, arg); + /* EPERM for F_DUPFD, ENOTSUP for others */ + if (result == -1 && (errno == EPERM || errno == ENOTSUP) + && !fstat (fd, &sbuf) && S_ISDIR (sbuf.st_mode)) + { + ULONG ulMode; + + switch (action) + { + case F_DUPFD: + /* Find available fd */ + while (fcntl (arg, F_GETFL) != -1 || errno != EBADF) + arg++; + + result = dup2 (fd, arg); + break; + + /* Using underlying APIs is right ? */ + case F_GETFD: + if (DosQueryFHState (fd, &ulMode)) + break; + + result = (ulMode & OPEN_FLAGS_NOINHERIT) ? FD_CLOEXEC : 0; + break; + + case F_SETFD: + if (arg & ~FD_CLOEXEC) + break; + + if (DosQueryFHState (fd, &ulMode)) + break; + + if (arg & FD_CLOEXEC) + ulMode |= OPEN_FLAGS_NOINHERIT; + else + ulMode &= ~OPEN_FLAGS_NOINHERIT; + + /* Filter supported flags. */ + ulMode &= (OPEN_FLAGS_WRITE_THROUGH | OPEN_FLAGS_FAIL_ON_ERROR + | OPEN_FLAGS_NO_CACHE | OPEN_FLAGS_NOINHERIT); + + if (DosSetFHState (fd, ulMode)) + break; + + result = 0; + break; + + case F_GETFL: + result = 0; + break; + + case F_SETFL: + if (arg != 0) + break; + + result = 0; + break; + + default: + errno = EINVAL; + break; + } + } + + va_end (arg_ptr); + + return result; +} + +#endif diff --git a/lib/fcntl.in.h b/lib/fcntl.in.h index 5f17d8f65a1..eb70dc61eca 100644 --- a/lib/fcntl.in.h +++ b/lib/fcntl.in.h @@ -68,7 +68,7 @@ /* Native Windows platforms declare open(), creat() in <io.h>. */ #if (@GNULIB_OPEN@ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) + && (defined _WIN32 && ! defined __CYGWIN__) # include <io.h> #endif diff --git a/lib/fdatasync.c b/lib/fdatasync.c deleted file mode 100644 index 9fa93297499..00000000000 --- a/lib/fdatasync.c +++ /dev/null @@ -1,27 +0,0 @@ -/* Emulate fdatasync on platforms that lack it. - - Copyright (C) 2011-2019 Free Software Foundation, Inc. - - This library is free software; you can redistribute it and/or - modify it under the terms of the GNU General Public - License as published by the Free Software Foundation; either - version 3 of the License, or (at your option) any later version. - - This library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program. If not, see <https://www.gnu.org/licenses/>. */ - -#include <config.h> -#include <unistd.h> - -int -fdatasync (int fd) -{ - /* This does more work than strictly necessary, but is the best we - can do portably. */ - return fsync (fd); -} diff --git a/lib/filemode.h b/lib/filemode.h index 3a578cb9052..5ae9da0ebe3 100644 --- a/lib/filemode.h +++ b/lib/filemode.h @@ -1,7 +1,7 @@ /* Make a string describing file modes. - Copyright (C) 1998-1999, 2003, 2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1998-1999, 2003, 2006, 2009-2019 Free Software Foundation, + Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/fingerprint.c b/lib/fingerprint.c new file mode 100644 index 00000000000..e55de9c6da3 --- /dev/null +++ b/lib/fingerprint.c @@ -0,0 +1,66 @@ +/* Placeholder fingerprint for Emacs + +Copyright 2019 Free Software Foundation, Inc. + +This file is part of GNU Emacs. + +GNU Emacs is free software: you can redistribute it and/or modify +it under the terms of the GNU General Public License as published by +the Free Software Foundation, either version 3 of the License, or (at +your option) any later version. + +GNU Emacs is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +#include "fingerprint.h" + +/* This random fingerprint was generated by the shell command: + + shuf -i 0-255 -n 32 -r | awk '{printf " 0x%.02X,\n", $0}' + + In the final Emacs executable, this random fingerprint is replaced + by a fingerprint of the temporary Emacs executable that was built + along the way. */ + +unsigned char const fingerprint[] = + { + 0xDE, + 0x86, + 0xBB, + 0x99, + 0xFF, + 0xF5, + 0x46, + 0x9A, + 0x9E, + 0x3F, + 0x9F, + 0x5D, + 0x9A, + 0xDF, + 0xF0, + 0x91, + 0xBD, + 0xCD, + 0xC1, + 0xE8, + 0x0C, + 0x16, + 0x1E, + 0xAF, + 0xB8, + 0x6C, + 0xE2, + 0x2B, + 0xB1, + 0x24, + 0xCE, + 0xB0, + }; diff --git a/lib/fingerprint.h b/lib/fingerprint.h new file mode 100644 index 00000000000..0b195fd0ca7 --- /dev/null +++ b/lib/fingerprint.h @@ -0,0 +1,29 @@ +/* Header file for the Emacs build fingerprint. + +Copyright (C) 2016, 2018-2019 Free Software Foundation, Inc. + +This file is part of GNU Emacs. + +GNU Emacs is free software: you can redistribute it and/or modify +it under the terms of the GNU General Public License as published by +the Free Software Foundation, either version 3 of the License, or (at +your option) any later version. + +GNU Emacs is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ + +#ifndef EMACS_FINGERPRINT_H +#define EMACS_FINGERPRINT_H + +/* We generate fingerprint.c and fingerprint.o from all the sources in + Emacs. This way, we have a unique value that we can use to pair + data files (like a portable dump image) with a specific build of + Emacs. */ +extern unsigned char const fingerprint[32]; + +#endif diff --git a/lib/flexmember.h b/lib/flexmember.h index 0d65f6da60f..af17b41d29f 100644 --- a/lib/flexmember.h +++ b/lib/flexmember.h @@ -33,11 +33,26 @@ # define FLEXALIGNOF(type) _Alignof (type) #endif -/* Upper bound on the size of a struct of type TYPE with a flexible - array member named MEMBER that is followed by N bytes of other data. - This is not simply sizeof (TYPE) + N, since it may require - alignment on unusually picky C11 platforms, and - FLEXIBLE_ARRAY_MEMBER may be 1 on pre-C11 platforms. +/* Yield a properly aligned upper bound on the size of a struct of + type TYPE with a flexible array member named MEMBER that is + followed by N bytes of other data. The result is suitable as an + argument to malloc. For example: + + struct s { int n; char d[FLEXIBLE_ARRAY_MEMBER]; }; + struct s *p = malloc (FLEXSIZEOF (struct s, d, n * sizeof (char))); + + FLEXSIZEOF (TYPE, MEMBER, N) is not simply (sizeof (TYPE) + N), + since FLEXIBLE_ARRAY_MEMBER may be 1 on pre-C11 platforms. Nor is + it simply (offsetof (TYPE, MEMBER) + N), as that might yield a size + that causes malloc to yield a pointer that is not properly aligned + for TYPE; for example, if sizeof (int) == alignof (int) == 4, + malloc (offsetof (struct s, d) + 3 * sizeof (char)) is equivalent + to malloc (7) and might yield a pointer that is not a multiple of 4 + (which means the pointer is not properly aligned for struct s), + whereas malloc (FLEXSIZEOF (struct s, d, 3 * sizeof (char))) is + equivalent to malloc (8) and must yield a pointer that is a + multiple of 4. + Yield a value less than N if and only if arithmetic overflow occurs. */ #define FLEXSIZEOF(type, member, n) \ diff --git a/lib/fpending.c b/lib/fpending.c index 8d995e9de85..3c01285bb83 100644 --- a/lib/fpending.c +++ b/lib/fpending.c @@ -1,6 +1,6 @@ /* fpending.c -- return the number of pending output bytes on a stream - Copyright (C) 2000, 2004, 2006-2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2000, 2004, 2006-2007, 2009-2019 Free Software Foundation, + Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -24,6 +24,9 @@ #include "stdio-impl.h" +/* This file is not used on systems that already have the __fpending function, + namely glibc >= 2.2, Solaris >= 7, Android API >= 23. */ + /* Return the number of pending (aka buffered, unflushed) bytes on the stream, FP, that is open for writing. */ size_t @@ -32,7 +35,8 @@ __fpending (FILE *fp) /* Most systems provide FILE as a struct and the necessary bitmask in <stdio.h>, because they need it for implementing getc() and putc() as fast macros. */ -#if defined _IO_ftrylockfile || __GNU_LIBRARY__ == 1 /* GNU libc, BeOS, Haiku, Linux libc5 */ +#if defined _IO_EOF_SEEN || defined _IO_ftrylockfile || __GNU_LIBRARY__ == 1 + /* GNU libc, BeOS, Haiku, Linux libc5 */ return fp->_IO_write_ptr - fp->_IO_write_base; #elif defined __sferror || defined __DragonFly__ || defined __ANDROID__ /* FreeBSD, NetBSD, OpenBSD, DragonFly, Mac OS X, Cygwin, Minix 3, Android */ diff --git a/lib/fpending.h b/lib/fpending.h index 0d77266da4c..097a3ef0717 100644 --- a/lib/fpending.h +++ b/lib/fpending.h @@ -1,7 +1,7 @@ /* Declare __fpending. - Copyright (C) 2000, 2003, 2005-2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2000, 2003, 2005-2006, 2009-2019 Free Software Foundation, + Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/fstatat.c b/lib/fstatat.c index 515b5693991..019d3c61638 100644 --- a/lib/fstatat.c +++ b/lib/fstatat.c @@ -36,10 +36,14 @@ orig_fstatat (int fd, char const *filename, struct stat *buf, int flags) } #endif +#ifdef __osf__ /* Write "sys/stat.h" here, not <sys/stat.h>, otherwise OSF/1 5.1 DTK cc eliminates this include because of the preliminary #include <sys/stat.h> above. */ -#include "sys/stat.h" +# include "sys/stat.h" +#else +# include <sys/stat.h> +#endif #include "stat-time.h" diff --git a/lib/fsusage.c b/lib/fsusage.c new file mode 100644 index 00000000000..7ddeb52fc0f --- /dev/null +++ b/lib/fsusage.c @@ -0,0 +1,237 @@ +/* fsusage.c -- return space usage of mounted file systems + + Copyright (C) 1991-1992, 1996, 1998-1999, 2002-2006, 2009-2019 Free Software + Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <https://www.gnu.org/licenses/>. */ + +#include <config.h> + +#include "fsusage.h" + +#include <limits.h> +#include <sys/types.h> + +#if STAT_STATVFS || STAT_STATVFS64 /* POSIX 1003.1-2001 (and later) with XSI */ +# include <sys/statvfs.h> +#else +/* Don't include backward-compatibility files unless they're needed. + Eventually we'd like to remove all this cruft. */ +# include <fcntl.h> +# include <unistd.h> +# include <sys/stat.h> +#if HAVE_SYS_PARAM_H +# include <sys/param.h> +#endif +#if HAVE_SYS_MOUNT_H +# include <sys/mount.h> +#endif +#if HAVE_SYS_VFS_H +# include <sys/vfs.h> +#endif +# if HAVE_SYS_FS_S5PARAM_H /* Fujitsu UXP/V */ +# include <sys/fs/s5param.h> +# endif +# if HAVE_SYS_STATFS_H +# include <sys/statfs.h> +# endif +#endif + +/* Many space usage primitives use all 1 bits to denote a value that is + not applicable or unknown. Propagate this information by returning + a uintmax_t value that is all 1 bits if X is all 1 bits, even if X + is unsigned and narrower than uintmax_t. */ +#define PROPAGATE_ALL_ONES(x) \ + ((sizeof (x) < sizeof (uintmax_t) \ + && (~ (x) == (sizeof (x) < sizeof (int) \ + ? - (1 << (sizeof (x) * CHAR_BIT)) \ + : 0))) \ + ? UINTMAX_MAX : (uintmax_t) (x)) + +/* Extract the top bit of X as an uintmax_t value. */ +#define EXTRACT_TOP_BIT(x) ((x) \ + & ((uintmax_t) 1 << (sizeof (x) * CHAR_BIT - 1))) + +/* If a value is negative, many space usage primitives store it into an + integer variable by assignment, even if the variable's type is unsigned. + So, if a space usage variable X's top bit is set, convert X to the + uintmax_t value V such that (- (uintmax_t) V) is the negative of + the original value. If X's top bit is clear, just yield X. + Use PROPAGATE_TOP_BIT if the original value might be negative; + otherwise, use PROPAGATE_ALL_ONES. */ +#define PROPAGATE_TOP_BIT(x) ((x) | ~ (EXTRACT_TOP_BIT (x) - 1)) + +#ifdef STAT_STATVFS +/* Return true if statvfs works. This is false for statvfs on systems + with GNU libc on Linux kernels before 2.6.36, which stats all + preceding entries in /proc/mounts; that makes df hang if even one + of the corresponding file systems is hard-mounted but not available. */ +# if ! (__linux__ && (__GLIBC__ || __UCLIBC__)) +/* The FRSIZE fallback is not required in this case. */ +# undef STAT_STATFS2_FRSIZE +static int statvfs_works (void) { return 1; } +# else +# include <string.h> /* for strverscmp */ +# include <sys/utsname.h> +# include <sys/statfs.h> +# define STAT_STATFS2_BSIZE 1 + +static int +statvfs_works (void) +{ + static int statvfs_works_cache = -1; + struct utsname name; + if (statvfs_works_cache < 0) + statvfs_works_cache = (uname (&name) == 0 + && 0 <= strverscmp (name.release, "2.6.36")); + return statvfs_works_cache; +} +# endif +#endif + + +/* Fill in the fields of FSP with information about space usage for + the file system on which FILE resides. + DISK is the device on which FILE is mounted, for space-getting + methods that need to know it. + Return 0 if successful, -1 if not. When returning -1, ensure that + ERRNO is either a system error value, or zero if DISK is NULL + on a system that requires a non-NULL value. */ +int +get_fs_usage (char const *file, char const *disk, struct fs_usage *fsp) +{ +#ifdef STAT_STATVFS /* POSIX, except pre-2.6.36 glibc/Linux */ + + if (statvfs_works ()) + { + struct statvfs vfsd; + + if (statvfs (file, &vfsd) < 0) + return -1; + + /* f_frsize isn't guaranteed to be supported. */ + fsp->fsu_blocksize = (vfsd.f_frsize + ? PROPAGATE_ALL_ONES (vfsd.f_frsize) + : PROPAGATE_ALL_ONES (vfsd.f_bsize)); + + fsp->fsu_blocks = PROPAGATE_ALL_ONES (vfsd.f_blocks); + fsp->fsu_bfree = PROPAGATE_ALL_ONES (vfsd.f_bfree); + fsp->fsu_bavail = PROPAGATE_TOP_BIT (vfsd.f_bavail); + fsp->fsu_bavail_top_bit_set = EXTRACT_TOP_BIT (vfsd.f_bavail) != 0; + fsp->fsu_files = PROPAGATE_ALL_ONES (vfsd.f_files); + fsp->fsu_ffree = PROPAGATE_ALL_ONES (vfsd.f_ffree); + return 0; + } + +#endif + +#if defined STAT_STATVFS64 /* AIX */ + + struct statvfs64 fsd; + + if (statvfs64 (file, &fsd) < 0) + return -1; + + /* f_frsize isn't guaranteed to be supported. */ + fsp->fsu_blocksize = (fsd.f_frsize + ? PROPAGATE_ALL_ONES (fsd.f_frsize) + : PROPAGATE_ALL_ONES (fsd.f_bsize)); + +#elif defined STAT_STATFS3_OSF1 /* OSF/1 */ + + struct statfs fsd; + + if (statfs (file, &fsd, sizeof (struct statfs)) != 0) + return -1; + + fsp->fsu_blocksize = PROPAGATE_ALL_ONES (fsd.f_fsize); + +#elif defined STAT_STATFS2_FRSIZE /* 2.6 < glibc/Linux < 2.6.36 */ + + struct statfs fsd; + + if (statfs (file, &fsd) < 0) + return -1; + + fsp->fsu_blocksize = PROPAGATE_ALL_ONES (fsd.f_frsize); + +#elif defined STAT_STATFS2_BSIZE /* glibc/Linux < 2.6, 4.3BSD, SunOS 4, \ + Mac OS X < 10.4, FreeBSD < 5.0, \ + NetBSD < 3.0, OpenBSD < 4.4 */ + + struct statfs fsd; + + if (statfs (file, &fsd) < 0) + return -1; + + fsp->fsu_blocksize = PROPAGATE_ALL_ONES (fsd.f_bsize); + +# ifdef STATFS_TRUNCATES_BLOCK_COUNTS + + /* In SunOS 4.1.2, 4.1.3, and 4.1.3_U1, the block counts in the + struct statfs are truncated to 2GB. These conditions detect that + truncation, presumably without botching the 4.1.1 case, in which + the values are not truncated. The correct counts are stored in + undocumented spare fields. */ + if (fsd.f_blocks == 0x7fffffff / fsd.f_bsize && fsd.f_spare[0] > 0) + { + fsd.f_blocks = fsd.f_spare[0]; + fsd.f_bfree = fsd.f_spare[1]; + fsd.f_bavail = fsd.f_spare[2]; + } +# endif /* STATFS_TRUNCATES_BLOCK_COUNTS */ + +#elif defined STAT_STATFS2_FSIZE /* 4.4BSD and older NetBSD */ + + struct statfs fsd; + + if (statfs (file, &fsd) < 0) + return -1; + + fsp->fsu_blocksize = PROPAGATE_ALL_ONES (fsd.f_fsize); + +#elif defined STAT_STATFS4 /* SVR3, old Irix */ + + struct statfs fsd; + + if (statfs (file, &fsd, sizeof fsd, 0) < 0) + return -1; + + /* Empirically, the block counts on most SVR3 and SVR3-derived + systems seem to always be in terms of 512-byte blocks, + no matter what value f_bsize has. */ +# if defined _CRAY + fsp->fsu_blocksize = PROPAGATE_ALL_ONES (fsd.f_bsize); +# else + fsp->fsu_blocksize = 512; +# endif + +#endif + +#if (defined STAT_STATVFS64 || defined STAT_STATFS3_OSF1 \ + || defined STAT_STATFS2_FRSIZE || defined STAT_STATFS2_BSIZE \ + || defined STAT_STATFS2_FSIZE || defined STAT_STATFS4) + + fsp->fsu_blocks = PROPAGATE_ALL_ONES (fsd.f_blocks); + fsp->fsu_bfree = PROPAGATE_ALL_ONES (fsd.f_bfree); + fsp->fsu_bavail = PROPAGATE_TOP_BIT (fsd.f_bavail); + fsp->fsu_bavail_top_bit_set = EXTRACT_TOP_BIT (fsd.f_bavail) != 0; + fsp->fsu_files = PROPAGATE_ALL_ONES (fsd.f_files); + fsp->fsu_ffree = PROPAGATE_ALL_ONES (fsd.f_ffree); + +#endif + + (void) disk; /* avoid argument-unused warning */ + return 0; +} diff --git a/lib/fsusage.h b/lib/fsusage.h new file mode 100644 index 00000000000..1d550bc451a --- /dev/null +++ b/lib/fsusage.h @@ -0,0 +1,40 @@ +/* fsusage.h -- declarations for file system space usage info + + Copyright (C) 1991-1992, 1997, 2003-2006, 2009-2019 Free Software + Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <https://www.gnu.org/licenses/>. */ + +/* Space usage statistics for a file system. Blocks are 512-byte. */ + +#if !defined FSUSAGE_H_ +# define FSUSAGE_H_ + +# include <stdint.h> +# include <stdbool.h> + +struct fs_usage +{ + uintmax_t fsu_blocksize; /* Size of a block. */ + uintmax_t fsu_blocks; /* Total blocks. */ + uintmax_t fsu_bfree; /* Free blocks available to superuser. */ + uintmax_t fsu_bavail; /* Free blocks available to non-superuser. */ + bool fsu_bavail_top_bit_set; /* 1 if fsu_bavail represents a value < 0. */ + uintmax_t fsu_files; /* Total file nodes. */ + uintmax_t fsu_ffree; /* Free file nodes. */ +}; + +int get_fs_usage (char const *file, char const *disk, struct fs_usage *fsp); + +#endif diff --git a/lib/fsync.c b/lib/fsync.c index 182fd95b88f..bfb6d28a5cf 100644 --- a/lib/fsync.c +++ b/lib/fsync.c @@ -25,7 +25,7 @@ #include <config.h> #include <unistd.h> -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ /* FlushFileBuffers */ # define WIN32_LEAN_AND_MEAN diff --git a/lib/ftoastr.c b/lib/ftoastr.c index 6c0ce261dca..55f1e02e80f 100644 --- a/lib/ftoastr.c +++ b/lib/ftoastr.c @@ -40,9 +40,7 @@ # define FLOAT_PREC_BOUND _GL_LDBL_PREC_BOUND # define FTOASTR ldtoastr # define PROMOTED_FLOAT long double -# if HAVE_C99_STRTOLD -# define STRTOF strtold -# endif +# define STRTOF strtold #elif LENGTH == 2 # define FLOAT double # define FLOAT_DIG DBL_DIG @@ -63,7 +61,7 @@ # endif #endif -/* On pre-C99 hosts, approximate strtof and strtold with strtod. This +/* On pre-C99 hosts, approximate strtof with strtod. This may generate one or two extra digits, but that's better than not working at all. */ #ifndef STRTOF diff --git a/lib/get-permissions.c b/lib/get-permissions.c index 354693e599d..99b4664aae7 100644 --- a/lib/get-permissions.c +++ b/lib/get-permissions.c @@ -31,16 +31,16 @@ int get_permissions (const char *name, int desc, mode_t mode, - struct permission_context *ctx) + struct permission_context *ctx) { memset (ctx, 0, sizeof *ctx); ctx->mode = mode; #if USE_ACL && HAVE_ACL_GET_FILE /* POSIX 1003.1e (draft 17 -- abandoned) specific version. */ - /* Linux, FreeBSD, Mac OS X, IRIX, Tru64 */ + /* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ # if !HAVE_ACL_TYPE_EXTENDED - /* Linux, FreeBSD, IRIX, Tru64 */ + /* Linux, FreeBSD, IRIX, Tru64, Cygwin >= 2.5 */ if (HAVE_ACL_GET_FD && desc != -1) ctx->acl = acl_get_fd (desc); @@ -57,16 +57,16 @@ get_permissions (const char *name, int desc, mode_t mode, { ctx->default_acl = acl_get_file (name, ACL_TYPE_DEFAULT); if (ctx->default_acl == NULL) - return -1; + return -1; } -# if HAVE_ACL_TYPE_NFS4 /* FreeBSD */ +# if HAVE_ACL_TYPE_NFS4 /* FreeBSD */ /* TODO (see set_permissions). */ -# endif +# endif -# else /* HAVE_ACL_TYPE_EXTENDED */ +# else /* HAVE_ACL_TYPE_EXTENDED */ /* Mac OS X */ /* On Mac OS X, acl_get_file (name, ACL_TYPE_ACCESS) @@ -115,16 +115,16 @@ get_permissions (const char *name, int desc, mode_t mode, int ret; if (desc != -1) - ret = facl (desc, ACE_GETACLCNT, 0, NULL); + ret = facl (desc, ACE_GETACLCNT, 0, NULL); else - ret = acl (name, ACE_GETACLCNT, 0, NULL); + ret = acl (name, ACE_GETACLCNT, 0, NULL); if (ret < 0) - { - if (errno == ENOSYS || errno == EINVAL) - ret = 0; - else - return -1; - } + { + if (errno == ENOSYS || errno == EINVAL) + ret = 0; + else + return -1; + } ctx->ace_count = ret; if (ctx->ace_count == 0) @@ -138,15 +138,15 @@ get_permissions (const char *name, int desc, mode_t mode, } if (desc != -1) - ret = facl (desc, ACE_GETACL, ctx->ace_count, ctx->ace_entries); + ret = facl (desc, ACE_GETACL, ctx->ace_count, ctx->ace_entries); else - ret = acl (name, ACE_GETACL, ctx->ace_count, ctx->ace_entries); + ret = acl (name, ACE_GETACL, ctx->ace_count, ctx->ace_entries); if (ret < 0) { if (errno == ENOSYS || errno == EINVAL) { - free (ctx->ace_entries); - ctx->ace_entries = NULL; + free (ctx->ace_entries); + ctx->ace_entries = NULL; ctx->ace_count = 0; break; } @@ -154,10 +154,10 @@ get_permissions (const char *name, int desc, mode_t mode, return -1; } if (ret <= ctx->ace_count) - { - ctx->ace_count = ret; - break; - } + { + ctx->ace_count = ret; + break; + } /* Huh? The number of ACL entries has increased since the last call. Repeat. */ free (ctx->ace_entries); @@ -170,20 +170,20 @@ get_permissions (const char *name, int desc, mode_t mode, int ret; if (desc != -1) - ret = facl (desc, GETACLCNT, 0, NULL); + ret = facl (desc, GETACLCNT, 0, NULL); else - ret = acl (name, GETACLCNT, 0, NULL); + ret = acl (name, GETACLCNT, 0, NULL); if (ret < 0) - { - if (errno == ENOSYS || errno == ENOTSUP || errno == EOPNOTSUPP) - ret = 0; - else - return -1; - } + { + if (errno == ENOSYS || errno == ENOTSUP || errno == EOPNOTSUPP) + ret = 0; + else + return -1; + } ctx->count = ret; if (ctx->count == 0) - break; + break; ctx->entries = (aclent_t *) malloc (ctx->count * sizeof (aclent_t)); if (ctx->entries == NULL) @@ -193,26 +193,26 @@ get_permissions (const char *name, int desc, mode_t mode, } if (desc != -1) - ret = facl (desc, GETACL, ctx->count, ctx->entries); + ret = facl (desc, GETACL, ctx->count, ctx->entries); else - ret = acl (name, GETACL, ctx->count, ctx->entries); + ret = acl (name, GETACL, ctx->count, ctx->entries); if (ret < 0) - { - if (errno == ENOSYS || errno == ENOTSUP || errno == EOPNOTSUPP) - { - free (ctx->entries); - ctx->entries = NULL; - ctx->count = 0; - break; - } - else - return -1; - } + { + if (errno == ENOSYS || errno == ENOTSUP || errno == EOPNOTSUPP) + { + free (ctx->entries); + ctx->entries = NULL; + ctx->count = 0; + break; + } + else + return -1; + } if (ret <= ctx->count) - { - ctx->count = ret; - break; - } + { + ctx->count = ret; + break; + } /* Huh? The number of ACL entries has increased since the last call. Repeat. */ free (ctx->entries); diff --git a/lib/getdtablesize.c b/lib/getdtablesize.c index 1bee093bc15..03a92435f0b 100644 --- a/lib/getdtablesize.c +++ b/lib/getdtablesize.c @@ -20,7 +20,7 @@ /* Specification. */ #include <unistd.h> -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ # include <stdio.h> diff --git a/lib/getgroups.c b/lib/getgroups.c index 7dc672d1f4d..d8c77e9a650 100644 --- a/lib/getgroups.c +++ b/lib/getgroups.c @@ -1,7 +1,6 @@ /* provide consistent interface to getgroups for systems that don't allow N==0 - Copyright (C) 1996, 1999, 2003, 2006-2019 Free Software Foundation, - Inc. + Copyright (C) 1996, 1999, 2003, 2006-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -59,8 +58,8 @@ int posix_getgroups (int, gid_t []) __asm ("_getgroups"); # define getgroups posix_getgroups # endif -/* On at least Ultrix 4.3 and NextStep 3.2, getgroups (0, NULL) always - fails. On other systems, it returns the number of supplemental +/* On at least NeXTstep 3.2, getgroups (0, NULL) always fails. + On other systems, it returns the number of supplemental groups for the process. This function handles that special case and lets the system-provided function handle all others. However, it can fail with ENOMEM if memory is tight. It is unspecified diff --git a/lib/getloadavg.c b/lib/getloadavg.c index c07f7db175d..08c14efcfce 100644 --- a/lib/getloadavg.c +++ b/lib/getloadavg.c @@ -1,7 +1,7 @@ /* Get the system load averages. - Copyright (C) 1985-1989, 1991-1995, 1997, 1999-2000, 2003-2019 Free - Software Foundation, Inc. + Copyright (C) 1985-1989, 1991-1995, 1997, 1999-2000, 2003-2019 Free Software + Foundation, Inc. NOTE: The canonical source of this file is maintained with gnulib. Bugs can be reported to bug-gnulib@gnu.org. @@ -47,29 +47,25 @@ N_NAME_POINTER The nlist n_name element is a pointer, not an array. HAVE_STRUCT_NLIST_N_UN_N_NAME 'n_un.n_name' is member of 'struct nlist'. - LINUX_LDAV_FILE [__linux__, __CYGWIN__]: File containing - load averages. + LINUX_LDAV_FILE [__linux__, __ANDROID__, __CYGWIN__]: File + containing load averages. Specific system predefines this file uses, aside from setting default values if not emacs: apollo BSD Real BSD, not just BSD-like. - convex DGUX eunice UNIX emulator under VMS. hpux __MSDOS__ No-op for MSDOS. NeXT sgi - sequent Sequent Dynix 3.x.x (BSD) - _SEQUENT_ Sequent DYNIX/ptx 1.x.x (SYSV) - sony_news NEWS-OS (works at least for 4.1C) UMAX UMAX4_3 VMS - WINDOWS32 No-op for Windows95/NT. - __linux__ Linux: assumes /proc file system mounted. + _WIN32 Native Windows (possibly also defined on Cygwin) + __linux__, __ANDROID__ Linux: assumes /proc file system mounted. Support from Michael K. Johnson. __CYGWIN__ Cygwin emulates linux /proc/loadavg. __NetBSD__ NetBSD: assumes /kern file system mounted. @@ -97,9 +93,8 @@ # include "intprops.h" -# if !defined (BSD) && defined (ultrix) -/* Ultrix behaves like BSD on Vaxen. */ -# define BSD +# if defined _WIN32 && ! defined __CYGWIN__ && ! defined WINDOWS32 +# define WINDOWS32 # endif # ifdef NeXT @@ -141,10 +136,6 @@ # define MORE_BSD # endif -# if defined (ultrix) && defined (mips) -# define decstation -# endif - # if defined (__SVR4) && !defined (SVR4) # define SVR4 # endif @@ -168,13 +159,6 @@ # include <sys/table.h> # endif -/* UTek's /bin/cc on the 4300 has no architecture specific cpp define by - default, but _MACH_IND_SYS_TYPES is defined in <sys/types.h>. Combine - that with a couple of other things and we'll have a unique match. */ -# if !defined (tek4300) && defined (unix) && defined (m68k) && defined (mc68000) && defined (mc68020) && defined (_MACH_IND_SYS_TYPES) -# define tek4300 /* Define by emacs, but not by other users. */ -# endif - /* VAX C can't handle multi-line #ifs, or lines longer than 256 chars. */ # ifndef LOAD_AVE_TYPE @@ -187,14 +171,6 @@ # define LOAD_AVE_TYPE long # endif -# ifdef decstation -# define LOAD_AVE_TYPE long -# endif - -# ifdef _SEQUENT_ -# define LOAD_AVE_TYPE long -# endif - # ifdef sgi # define LOAD_AVE_TYPE long # endif @@ -203,41 +179,14 @@ # define LOAD_AVE_TYPE long # endif -# ifdef sony_news -# define LOAD_AVE_TYPE long -# endif - -# ifdef sequent -# define LOAD_AVE_TYPE long -# endif - # ifdef OSF_ALPHA # define LOAD_AVE_TYPE long # endif -# if defined (ardent) && defined (titan) -# define LOAD_AVE_TYPE long -# endif - -# ifdef tek4300 -# define LOAD_AVE_TYPE long -# endif - -# if defined (alliant) && defined (i860) /* Alliant FX/2800 */ -# define LOAD_AVE_TYPE long -# endif - # if defined _AIX && ! defined HAVE_LIBPERFSTAT # define LOAD_AVE_TYPE long # endif -# ifdef convex -# define LOAD_AVE_TYPE double -# ifndef LDAV_CVT -# define LDAV_CVT(n) (n) -# endif -# endif - # endif /* No LOAD_AVE_TYPE. */ # ifdef OSF_ALPHA @@ -247,13 +196,6 @@ # define FSCALE 1024.0 # endif -# if defined (alliant) && defined (i860) /* Alliant FX/2800 */ -/* <sys/param.h> defines an incorrect value for FSCALE on an - Alliant FX/2800 Concentrix 2.2, according to ghazi@noc.rutgers.edu. */ -# undef FSCALE -# define FSCALE 100.0 -# endif - # ifndef FSCALE @@ -263,25 +205,17 @@ # define FSCALE 2048.0 # endif -# if defined (MIPS) || defined (SVR4) || defined (decstation) +# if defined (MIPS) || defined (SVR4) # define FSCALE 256 # endif -# if defined (sgi) || defined (sequent) +# if defined (sgi) /* Sometimes both MIPS and sgi are defined, so FSCALE was just defined above under #ifdef MIPS. But we want the sgi value. */ # undef FSCALE # define FSCALE 1000.0 # endif -# if defined (ardent) && defined (titan) -# define FSCALE 65536.0 -# endif - -# ifdef tek4300 -# define FSCALE 100.0 -# endif - # if defined _AIX && !defined HAVE_LIBPERFSTAT # define FSCALE 65536.0 # endif @@ -303,28 +237,22 @@ # endif -# if !defined (KERNEL_FILE) && defined (sequent) -# define KERNEL_FILE "/dynix" -# endif - # if !defined (KERNEL_FILE) && defined (hpux) # define KERNEL_FILE "/hp-ux" # endif -# if !defined (KERNEL_FILE) && (defined (_SEQUENT_) || defined (MIPS) || defined (SVR4) || defined (ISC) || defined (sgi) || (defined (ardent) && defined (titan))) +# if !defined (KERNEL_FILE) && (defined (MIPS) || defined (SVR4) || defined (ISC) || defined (sgi)) # define KERNEL_FILE "/unix" # endif -# if !defined (LDAV_SYMBOL) && defined (alliant) -# define LDAV_SYMBOL "_Loadavg" -# endif - -# if !defined (LDAV_SYMBOL) && ((defined (hpux) && !defined (hp9000s300)) || defined (_SEQUENT_) || defined (SVR4) || defined (ISC) || defined (sgi) || (defined (ardent) && defined (titan)) || (defined (_AIX) && !defined(HAVE_LIBPERFSTAT))) +# if !defined (LDAV_SYMBOL) && (defined (hpux) || defined (SVR4) || defined (ISC) || defined (sgi) || (defined (_AIX) && !defined(HAVE_LIBPERFSTAT))) # define LDAV_SYMBOL "avenrun" # endif -# include <unistd.h> +# ifdef HAVE_UNISTD_H +# include <unistd.h> +# endif /* LOAD_AVE_TYPE should only get defined if we're going to use the nlist method. */ @@ -335,7 +263,7 @@ # ifdef LOAD_AVE_TYPE # ifndef __VMS -# ifndef __linux__ +# if !(defined __linux__ || defined __ANDROID__) # ifndef NLIST_STRUCT # include <a.out.h> # else /* NLIST_STRUCT */ @@ -358,7 +286,7 @@ # ifndef LDAV_SYMBOL # define LDAV_SYMBOL "_avenrun" # endif /* LDAV_SYMBOL */ -# endif /* __linux__ */ +# endif /* __linux__ || __ANDROID__ */ # else /* __VMS */ @@ -431,7 +359,8 @@ # include <sys/dg_sys_info.h> # endif -# if (defined __linux__ || defined __CYGWIN__ || defined SUNOS_5 \ +# if (defined __linux__ || defined __ANDROID__ \ + || defined __CYGWIN__ || defined SUNOS_5 \ || (defined LOAD_AVE_TYPE && ! defined __VMS)) # include <fcntl.h> # endif @@ -460,7 +389,7 @@ static bool getloadavg_initialized; /* Offset in kmem to seek to read load average, or 0 means invalid. */ static long offset; -# if ! defined __VMS && ! defined sgi && ! defined __linux__ +# if ! defined __VMS && ! defined sgi && ! (defined __linux__ || defined __ANDROID__) static struct nlist name_list[2]; # endif @@ -495,17 +424,17 @@ getloadavg (double loadavg[], int nelem) int saved_errno; kc = kstat_open (); - if (kc == 0) + if (kc == NULL) return -1; ksp = kstat_lookup (kc, "unix", 0, "system_misc"); - if (ksp == 0) + if (ksp == NULL) return -1; if (kstat_read (kc, ksp, 0) == -1) return -1; kn = kstat_data_lookup (ksp, "avenrun_1min"); - if (kn == 0) + if (kn == NULL) { /* Return -1 if no load average information is available. */ nelem = 0; @@ -518,14 +447,14 @@ getloadavg (double loadavg[], int nelem) if (nelem >= 2) { kn = kstat_data_lookup (ksp, "avenrun_5min"); - if (kn != 0) + if (kn != NULL) { loadavg[elem++] = (double) kn->value.ul / FSCALE; if (nelem >= 3) { kn = kstat_data_lookup (ksp, "avenrun_15min"); - if (kn != 0) + if (kn != NULL) loadavg[elem++] = (double) kn->value.ul / FSCALE; } } @@ -570,8 +499,8 @@ getloadavg (double loadavg[], int nelem) } # endif -# if !defined (LDAV_DONE) && (defined (__linux__) || defined (__CYGWIN__)) - /* Linux without glibc, Cygwin */ +# if !defined (LDAV_DONE) && (defined __linux__ || defined __ANDROID__ || defined __CYGWIN__) + /* Linux without glibc, Android, Cygwin */ # define LDAV_DONE # undef LOAD_AVE_TYPE @@ -626,7 +555,7 @@ getloadavg (double loadavg[], int nelem) return elem; -# endif /* __linux__ || __CYGWIN__ */ +# endif /* __linux__ || __ANDROID__ || __CYGWIN__ */ # if !defined (LDAV_DONE) && defined (__NetBSD__) /* NetBSD < 0.9 */ # define LDAV_DONE @@ -915,7 +844,7 @@ getloadavg (double loadavg[], int nelem) # ifndef SUNOS_5 if ( -# if !(defined (_AIX) && !defined (ps2)) +# if !defined (_AIX) nlist (KERNEL_FILE, name_list) # else /* _AIX */ knlist (name_list, 1, sizeof (name_list[0])) @@ -966,7 +895,7 @@ getloadavg (double loadavg[], int nelem) /* We pass 0 for the kernel, corefile, and swapfile names to use the currently running kernel. */ kd = kvm_open (0, 0, 0, O_RDONLY, 0); - if (kd != 0) + if (kd != NULL) { /* nlist the currently running kernel. */ kvm_nlist (kd, name_list); diff --git a/lib/getopt.c b/lib/getopt.c index cb0f338b5d3..8ee075a8091 100644 --- a/lib/getopt.c +++ b/lib/getopt.c @@ -46,7 +46,7 @@ /* When used standalone, flockfile and funlockfile might not be available. */ # if (!defined _POSIX_THREAD_SAFE_FUNCTIONS \ - || ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__)) + || (defined _WIN32 && ! defined __CYGWIN__)) # define flockfile(fp) /* nop */ # define funlockfile(fp) /* nop */ # endif diff --git a/lib/gettext.h b/lib/gettext.h index 9492b46c49b..c7c0fdb5311 100644 --- a/lib/gettext.h +++ b/lib/gettext.h @@ -1,6 +1,6 @@ /* Convenience header for conditional use of GNU <libintl.h>. - Copyright (C) 1995-1998, 2000-2002, 2004-2006, 2009-2019 Free - Software Foundation, Inc. + Copyright (C) 1995-1998, 2000-2002, 2004-2006, 2009-2019 Free Software + Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -184,9 +184,16 @@ npgettext_aux (const char *domain, #include <string.h> -#if (((__GNUC__ >= 3 || __GNUG__ >= 2) && !defined __STRICT_ANSI__) \ - /* || __STDC_VERSION__ == 199901L - || (__STDC_VERSION__ >= 201112L && !defined __STDC_NO_VLA__) */ ) +/* GNULIB_NO_VLA can be defined to disable use of VLAs even if supported. + This relates to the -Wvla and -Wvla-larger-than warnings, enabled in + the default GCC many warnings set. This allows programs to disable use + of VLAs, which may be unintended, or may be awkward to support portably, + or may have security implications due to non-deterministic stack usage. */ + +#if (!defined GNULIB_NO_VLA \ + && (((__GNUC__ >= 3 || __GNUG__ >= 2) && !defined __STRICT_ANSI__) \ + /* || (__STDC_VERSION__ == 199901L && !defined __HP_cc) + || (__STDC_VERSION__ >= 201112L && !defined __STDC_NO_VLA__) */ )) # define _LIBGETTEXT_HAVE_VARIABLE_SIZE_ARRAYS 1 #else # define _LIBGETTEXT_HAVE_VARIABLE_SIZE_ARRAYS 0 diff --git a/lib/gettime.c b/lib/gettime.c index 6e61a09a418..1fd153f6dce 100644 --- a/lib/gettime.c +++ b/lib/gettime.c @@ -1,7 +1,6 @@ /* gettime -- get the system clock - Copyright (C) 2002, 2004-2007, 2009-2019 Free Software Foundation, - Inc. + Copyright (C) 2002, 2004-2007, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -29,21 +28,22 @@ void gettime (struct timespec *ts) { -#if HAVE_NANOTIME - nanotime (ts); +#if defined CLOCK_REALTIME && HAVE_CLOCK_GETTIME + clock_gettime (CLOCK_REALTIME, ts); #else + struct timeval tv; + gettimeofday (&tv, NULL); + ts->tv_sec = tv.tv_sec; + ts->tv_nsec = tv.tv_usec * 1000; +#endif +} -# if defined CLOCK_REALTIME && HAVE_CLOCK_GETTIME - if (clock_gettime (CLOCK_REALTIME, ts) == 0) - return; -# endif - - { - struct timeval tv; - gettimeofday (&tv, NULL); - ts->tv_sec = tv.tv_sec; - ts->tv_nsec = tv.tv_usec * 1000; - } +/* Return the current system time as a struct timespec. */ -#endif +struct timespec +current_timespec (void) +{ + struct timespec ts; + gettime (&ts); + return ts; } diff --git a/lib/gettimeofday.c b/lib/gettimeofday.c index db4e8f48891..e728bf47355 100644 --- a/lib/gettimeofday.c +++ b/lib/gettimeofday.c @@ -1,7 +1,6 @@ /* Provide gettimeofday for systems that don't have it or for which it's broken. - Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -25,7 +24,7 @@ #include <time.h> -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ # define WINDOWS_NATIVE # include <windows.h> #endif @@ -34,6 +33,10 @@ #ifdef WINDOWS_NATIVE +/* Avoid warnings from gcc -Wcast-function-type. */ +# define GetProcAddress \ + (void *) GetProcAddress + /* GetSystemTimePreciseAsFileTime was introduced only in Windows 8. */ typedef void (WINAPI * GetSystemTimePreciseAsFileTimeFuncType) (FILETIME *lpTime); static GetSystemTimePreciseAsFileTimeFuncType GetSystemTimePreciseAsFileTimeFunc = NULL; @@ -46,7 +49,7 @@ initialize (void) if (kernel32 != NULL) { GetSystemTimePreciseAsFileTimeFunc = - (GetSystemTimePreciseAsFileTimeFuncType) GetProcAddress (kernel32, "GetSystemTimePreciseAsFileTime"); + (GetSystemTimePreciseAsFileTimeFuncType) GetProcAddress (kernel32, "GetSystemTimePreciseAsFileTime"); } initialized = TRUE; } @@ -69,10 +72,10 @@ gettimeofday (struct timeval *restrict tv, void *restrict tz) /* On native Windows, there are two ways to get the current time: GetSystemTimeAsFileTime - <https://msdn.microsoft.com/en-us/library/ms724397.aspx> + <https://docs.microsoft.com/en-us/windows/desktop/api/sysinfoapi/nf-sysinfoapi-getsystemtimeasfiletime> or GetSystemTimePreciseAsFileTime - <https://msdn.microsoft.com/en-us/library/hh706895.aspx>. + <https://docs.microsoft.com/en-us/windows/desktop/api/sysinfoapi/nf-sysinfoapi-getsystemtimepreciseasfiletime>. GetSystemTimeAsFileTime produces values that jump by increments of 15.627 milliseconds (!) on average. Whereas GetSystemTimePreciseAsFileTime values usually jump by 1 or 2 @@ -89,7 +92,7 @@ gettimeofday (struct timeval *restrict tv, void *restrict tz) GetSystemTimeAsFileTime (¤t_time); /* Convert from FILETIME to 'struct timeval'. */ - /* FILETIME: <https://msdn.microsoft.com/en-us/library/ms724284.aspx> */ + /* FILETIME: <https://docs.microsoft.com/en-us/windows/desktop/api/minwinbase/ns-minwinbase-filetime> */ ULONGLONG since_1601 = ((ULONGLONG) current_time.dwHighDateTime << 32) | (ULONGLONG) current_time.dwLowDateTime; diff --git a/lib/gnulib.mk.in b/lib/gnulib.mk.in index efa87bc45dd..1cbbff212bb 100644 --- a/lib/gnulib.mk.in +++ b/lib/gnulib.mk.in @@ -1,5 +1,4 @@ ## DO NOT EDIT! GENERATED AUTOMATICALLY! -## Process this file with automake to produce Makefile.in. # Copyright (C) 2002-2019 Free Software Foundation, Inc. # # This file is free software; you can redistribute it and/or modify @@ -21,7 +20,137 @@ # the same distribution terms as the rest of that program. # # Generated by gnulib-tool. -# Reproduce by: gnulib-tool --import --lib=libgnu --source-base=lib --m4-base=m4 --doc-base=doc --tests-base=tests --aux-dir=build-aux --avoid=close --avoid=dup --avoid=fchdir --avoid=fstat --avoid=malloc-posix --avoid=msvc-inval --avoid=msvc-nothrow --avoid=openat-die --avoid=opendir --avoid=raise --avoid=save-cwd --avoid=select --avoid=setenv --avoid=sigprocmask --avoid=stat --avoid=stdarg --avoid=stdbool --avoid=threadlib --avoid=tzset --avoid=unsetenv --avoid=utime --avoid=utime-h --gnu-make --makefile-name=gnulib.mk.in --conditional-dependencies --no-libtool --macro-prefix=gl --no-vc-files alloca-opt binary-io byteswap c-ctype c-strcase careadlinkat close-stream count-leading-zeros count-one-bits count-trailing-zeros crypto/md5 crypto/sha1 crypto/sha256 crypto/sha512 d-type diffseq dtoastr dtotimespec dup2 environ execinfo explicit_bzero faccessat fcntl fcntl-h fdatasync fdopendir filemode filevercmp flexmember fpieee fstatat fsync getloadavg getopt-gnu gettime gettimeofday gitlog-to-changelog ignore-value intprops largefile lstat manywarnings memrchr minmax mkostemp mktime nstrftime pipe2 pselect pthread_sigmask putenv qcopy-acl readlink readlinkat sig2str socklen stat-time std-gnu11 stdalign stddef stdio stpcpy strtoimax symlink sys_stat sys_time tempname time time_r time_rz timegm timer-time timespec-add timespec-sub unlocked-io update-copyright utimens vla warnings +# Reproduce by: +# gnulib-tool --import \ +# --lib=libgnu \ +# --source-base=lib \ +# --m4-base=m4 \ +# --doc-base=doc \ +# --tests-base=tests \ +# --aux-dir=build-aux \ +# --gnu-make \ +# --makefile-name=gnulib.mk.in \ +# --conditional-dependencies \ +# --no-libtool \ +# --macro-prefix=gl \ +# --no-vc-files \ +# --avoid=btowc \ +# --avoid=close \ +# --avoid=dup \ +# --avoid=fchdir \ +# --avoid=fstat \ +# --avoid=langinfo \ +# --avoid=lock \ +# --avoid=malloc-posix \ +# --avoid=mbrtowc \ +# --avoid=mbsinit \ +# --avoid=memchr \ +# --avoid=mkdir \ +# --avoid=msvc-inval \ +# --avoid=msvc-nothrow \ +# --avoid=nl_langinfo \ +# --avoid=openat-die \ +# --avoid=opendir \ +# --avoid=raise \ +# --avoid=save-cwd \ +# --avoid=select \ +# --avoid=setenv \ +# --avoid=sigprocmask \ +# --avoid=stat \ +# --avoid=stdarg \ +# --avoid=stdbool \ +# --avoid=threadlib \ +# --avoid=tzset \ +# --avoid=unsetenv \ +# --avoid=utime \ +# --avoid=utime-h \ +# --avoid=wchar \ +# --avoid=wcrtomb \ +# --avoid=wctype-h \ +# alloca-opt \ +# binary-io \ +# byteswap \ +# c-ctype \ +# c-strcase \ +# careadlinkat \ +# close-stream \ +# count-leading-zeros \ +# count-one-bits \ +# count-trailing-zeros \ +# crypto/md5-buffer \ +# crypto/sha1-buffer \ +# crypto/sha256-buffer \ +# crypto/sha512-buffer \ +# d-type \ +# diffseq \ +# dosname \ +# dtoastr \ +# dtotimespec \ +# dup2 \ +# environ \ +# execinfo \ +# explicit_bzero \ +# faccessat \ +# fcntl \ +# fcntl-h \ +# fdopendir \ +# filemode \ +# filevercmp \ +# flexmember \ +# fpieee \ +# fstatat \ +# fsusage \ +# fsync \ +# getloadavg \ +# getopt-gnu \ +# gettime \ +# gettimeofday \ +# gitlog-to-changelog \ +# ieee754-h \ +# ignore-value \ +# intprops \ +# largefile \ +# lstat \ +# manywarnings \ +# memmem-simple \ +# memrchr \ +# minmax \ +# mkostemp \ +# mktime \ +# nstrftime \ +# pipe2 \ +# pselect \ +# pthread_sigmask \ +# putenv \ +# qcopy-acl \ +# readlink \ +# readlinkat \ +# regex \ +# sig2str \ +# socklen \ +# stat-time \ +# std-gnu11 \ +# stdalign \ +# stddef \ +# stdio \ +# stpcpy \ +# strtoimax \ +# symlink \ +# sys_stat \ +# sys_time \ +# tempname \ +# time \ +# time_r \ +# time_rz \ +# timegm \ +# timer-time \ +# timespec-add \ +# timespec-sub \ +# unlocked-io \ +# update-copyright \ +# utimens \ +# vla \ +# warnings MOSTLYCLEANFILES += core *.stackdump @@ -44,14 +173,15 @@ BITSIZEOF_SIZE_T = @BITSIZEOF_SIZE_T@ BITSIZEOF_WCHAR_T = @BITSIZEOF_WCHAR_T@ BITSIZEOF_WINT_T = @BITSIZEOF_WINT_T@ BLESSMAIL_TARGET = @BLESSMAIL_TARGET@ +BREW = @BREW@ BUILD_DETAILS = @BUILD_DETAILS@ BYTESWAP_H = @BYTESWAP_H@ CAIRO_CFLAGS = @CAIRO_CFLAGS@ CAIRO_LIBS = @CAIRO_LIBS@ -CANNOT_DUMP = @CANNOT_DUMP@ CC = @CC@ CFLAGS = @CFLAGS@ CFLAGS_SOUND = @CFLAGS_SOUND@ +CHECK_STRUCTS = @CHECK_STRUCTS@ CLIENTRES = @CLIENTRES@ CLIENTW = @CLIENTW@ CM_OBJ = @CM_OBJ@ @@ -69,6 +199,7 @@ DBUS_OBJ = @DBUS_OBJ@ DEFS = @DEFS@ DESLIB = @DESLIB@ DOCMISC_W32 = @DOCMISC_W32@ +DUMPING = @DUMPING@ ECHO_C = @ECHO_C@ ECHO_N = @ECHO_N@ ECHO_T = @ECHO_T@ @@ -99,16 +230,20 @@ GETOPT_CDEFS_H = @GETOPT_CDEFS_H@ GETOPT_H = @GETOPT_H@ GFILENOTIFY_CFLAGS = @GFILENOTIFY_CFLAGS@ GFILENOTIFY_LIBS = @GFILENOTIFY_LIBS@ +GLIBC21 = @GLIBC21@ GL_COND_LIBTOOL = @GL_COND_LIBTOOL@ GL_GENERATE_ALLOCA_H = @GL_GENERATE_ALLOCA_H@ GL_GENERATE_BYTESWAP_H = @GL_GENERATE_BYTESWAP_H@ GL_GENERATE_ERRNO_H = @GL_GENERATE_ERRNO_H@ GL_GENERATE_EXECINFO_H = @GL_GENERATE_EXECINFO_H@ +GL_GENERATE_IEEE754_H = @GL_GENERATE_IEEE754_H@ GL_GENERATE_LIMITS_H = @GL_GENERATE_LIMITS_H@ GL_GENERATE_STDALIGN_H = @GL_GENERATE_STDALIGN_H@ GL_GENERATE_STDDEF_H = @GL_GENERATE_STDDEF_H@ GL_GENERATE_STDINT_H = @GL_GENERATE_STDINT_H@ GMALLOC_OBJ = @GMALLOC_OBJ@ +GMP_LIB = @GMP_LIB@ +GMP_OBJ = @GMP_OBJ@ GNULIB_ALPHASORT = @GNULIB_ALPHASORT@ GNULIB_ATOLL = @GNULIB_ATOLL@ GNULIB_CALLOC_POSIX = @GNULIB_CALLOC_POSIX@ @@ -172,6 +307,7 @@ GNULIB_GETLOADAVG = @GNULIB_GETLOADAVG@ GNULIB_GETLOGIN = @GNULIB_GETLOGIN@ GNULIB_GETLOGIN_R = @GNULIB_GETLOGIN_R@ GNULIB_GETPAGESIZE = @GNULIB_GETPAGESIZE@ +GNULIB_GETPASS = @GNULIB_GETPASS@ GNULIB_GETSUBOPT = @GNULIB_GETSUBOPT@ GNULIB_GETTIMEOFDAY = @GNULIB_GETTIMEOFDAY@ GNULIB_GETUSERSHELL = @GNULIB_GETUSERSHELL@ @@ -298,6 +434,7 @@ GNULIB_STRSTR = @GNULIB_STRSTR@ GNULIB_STRTOD = @GNULIB_STRTOD@ GNULIB_STRTOIMAX = @GNULIB_STRTOIMAX@ GNULIB_STRTOK_R = @GNULIB_STRTOK_R@ +GNULIB_STRTOLD = @GNULIB_STRTOLD@ GNULIB_STRTOLL = @GNULIB_STRTOLL@ GNULIB_STRTOULL = @GNULIB_STRTOULL@ GNULIB_STRTOUMAX = @GNULIB_STRTOUMAX@ @@ -345,6 +482,7 @@ GTK_CFLAGS = @GTK_CFLAGS@ GTK_LIBS = @GTK_LIBS@ GTK_OBJ = @GTK_OBJ@ GZIP_PROG = @GZIP_PROG@ +HAVE_ALLOCA_H = @HAVE_ALLOCA_H@ HAVE_ALPHASORT = @HAVE_ALPHASORT@ HAVE_ATOLL = @HAVE_ATOLL@ HAVE_C99_STDINT_H = @HAVE_C99_STDINT_H@ @@ -386,6 +524,7 @@ HAVE_DECL_STRSIGNAL = @HAVE_DECL_STRSIGNAL@ HAVE_DECL_STRTOIMAX = @HAVE_DECL_STRTOIMAX@ HAVE_DECL_STRTOK_R = @HAVE_DECL_STRTOK_R@ HAVE_DECL_STRTOUMAX = @HAVE_DECL_STRTOUMAX@ +HAVE_DECL_TRUNCATE = @HAVE_DECL_TRUNCATE@ HAVE_DECL_TTYNAME_R = @HAVE_DECL_TTYNAME_R@ HAVE_DECL_UNSETENV = @HAVE_DECL_UNSETENV@ HAVE_DECL_VSNPRINTF = @HAVE_DECL_VSNPRINTF@ @@ -416,10 +555,13 @@ HAVE_GETHOSTNAME = @HAVE_GETHOSTNAME@ HAVE_GETLOGIN = @HAVE_GETLOGIN@ HAVE_GETOPT_H = @HAVE_GETOPT_H@ HAVE_GETPAGESIZE = @HAVE_GETPAGESIZE@ +HAVE_GETPASS = @HAVE_GETPASS@ HAVE_GETSUBOPT = @HAVE_GETSUBOPT@ HAVE_GETTIMEOFDAY = @HAVE_GETTIMEOFDAY@ HAVE_GRANTPT = @HAVE_GRANTPT@ HAVE_GROUP_MEMBER = @HAVE_GROUP_MEMBER@ +HAVE_IMAXDIV_T = @HAVE_IMAXDIV_T@ +HAVE_INITSTATE = @HAVE_INITSTATE@ HAVE_INTTYPES_H = @HAVE_INTTYPES_H@ HAVE_LCHMOD = @HAVE_LCHMOD@ HAVE_LCHOWN = @HAVE_LCHOWN@ @@ -430,6 +572,7 @@ HAVE_LSTAT = @HAVE_LSTAT@ HAVE_MAKEINFO = @HAVE_MAKEINFO@ HAVE_MAX_ALIGN_T = @HAVE_MAX_ALIGN_T@ HAVE_MBSLEN = @HAVE_MBSLEN@ +HAVE_MBTOWC = @HAVE_MBTOWC@ HAVE_MEMCHR = @HAVE_MEMCHR@ HAVE_MEMPCPY = @HAVE_MEMPCPY@ HAVE_MKDIRAT = @HAVE_MKDIRAT@ @@ -448,6 +591,7 @@ HAVE_OPENAT = @HAVE_OPENAT@ HAVE_OPENDIR = @HAVE_OPENDIR@ HAVE_OS_H = @HAVE_OS_H@ HAVE_PCLOSE = @HAVE_PCLOSE@ +HAVE_PDUMPER = @HAVE_PDUMPER@ HAVE_PIPE = @HAVE_PIPE@ HAVE_PIPE2 = @HAVE_PIPE2@ HAVE_POPEN = @HAVE_POPEN@ @@ -477,6 +621,7 @@ HAVE_SCANDIR = @HAVE_SCANDIR@ HAVE_SECURE_GETENV = @HAVE_SECURE_GETENV@ HAVE_SETENV = @HAVE_SETENV@ HAVE_SETHOSTNAME = @HAVE_SETHOSTNAME@ +HAVE_SETSTATE = @HAVE_SETSTATE@ HAVE_SIGACTION = @HAVE_SIGACTION@ HAVE_SIGHANDLER_T = @HAVE_SIGHANDLER_T@ HAVE_SIGINFO_T = @HAVE_SIGINFO_T@ @@ -494,6 +639,7 @@ HAVE_STRPBRK = @HAVE_STRPBRK@ HAVE_STRPTIME = @HAVE_STRPTIME@ HAVE_STRSEP = @HAVE_STRSEP@ HAVE_STRTOD = @HAVE_STRTOD@ +HAVE_STRTOLD = @HAVE_STRTOLD@ HAVE_STRTOLL = @HAVE_STRTOLL@ HAVE_STRTOULL = @HAVE_STRTOULL@ HAVE_STRUCT_RANDOM_DATA = @HAVE_STRUCT_RANDOM_DATA@ @@ -512,7 +658,6 @@ HAVE_SYS_TIME_H = @HAVE_SYS_TIME_H@ HAVE_SYS_TYPES_H = @HAVE_SYS_TYPES_H@ HAVE_TIMEGM = @HAVE_TIMEGM@ HAVE_TIMEZONE_T = @HAVE_TIMEZONE_T@ -HAVE_TRUNCATE = @HAVE_TRUNCATE@ HAVE_TYPE_VOLATILE_SIG_ATOMIC_T = @HAVE_TYPE_VOLATILE_SIG_ATOMIC_T@ HAVE_TZSET = @HAVE_TZSET@ HAVE_UNISTD_H = @HAVE_UNISTD_H@ @@ -529,6 +674,7 @@ HAVE_WINSOCK2_H = @HAVE_WINSOCK2_H@ HAVE_XSERVER = @HAVE_XSERVER@ HAVE__EXIT = @HAVE__EXIT@ HYBRID_MALLOC = @HYBRID_MALLOC@ +IEEE754_H = @IEEE754_H@ IMAGEMAGICK_CFLAGS = @IMAGEMAGICK_CFLAGS@ IMAGEMAGICK_LIBS = @IMAGEMAGICK_LIBS@ INCLUDE_NEXT = @INCLUDE_NEXT@ @@ -540,10 +686,15 @@ INSTALL_PROGRAM = @INSTALL_PROGRAM@ INSTALL_SCRIPT = @INSTALL_SCRIPT@ INT32_MAX_LT_INTMAX_MAX = @INT32_MAX_LT_INTMAX_MAX@ INT64_MAX_EQ_LONG_MAX = @INT64_MAX_EQ_LONG_MAX@ +JSON_CFLAGS = @JSON_CFLAGS@ +JSON_LIBS = @JSON_LIBS@ +JSON_OBJ = @JSON_OBJ@ KQUEUE_CFLAGS = @KQUEUE_CFLAGS@ KQUEUE_LIBS = @KQUEUE_LIBS@ KRB4LIB = @KRB4LIB@ KRB5LIB = @KRB5LIB@ +LCMS2_CFLAGS = @LCMS2_CFLAGS@ +LCMS2_LIBS = @LCMS2_LIBS@ LDFLAGS = @LDFLAGS@ LD_SWITCH_SYSTEM = @LD_SWITCH_SYSTEM@ LD_SWITCH_SYSTEM_TEMACS = @LD_SWITCH_SYSTEM_TEMACS@ @@ -558,7 +709,6 @@ LIBGPM = @LIBGPM@ LIBHESIOD = @LIBHESIOD@ LIBINTL = @LIBINTL@ LIBJPEG = @LIBJPEG@ -LIBLCMS2 = @LIBLCMS2@ LIBMODULES = @LIBMODULES@ LIBOBJS = @LIBOBJS@ LIBOTF_CFLAGS = @LIBOTF_CFLAGS@ @@ -590,7 +740,6 @@ LIB_ACL = @LIB_ACL@ LIB_CLOCK_GETTIME = @LIB_CLOCK_GETTIME@ LIB_EACCESS = @LIB_EACCESS@ LIB_EXECINFO = @LIB_EXECINFO@ -LIB_FDATASYNC = @LIB_FDATASYNC@ LIB_MATH = @LIB_MATH@ LIB_PTHREAD = @LIB_PTHREAD@ LIB_PTHREAD_SIGMASK = @LIB_PTHREAD_SIGMASK@ @@ -668,6 +817,7 @@ PKG_CONFIG = @PKG_CONFIG@ PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@ PKG_CONFIG_PATH = @PKG_CONFIG_PATH@ PNG_CFLAGS = @PNG_CFLAGS@ +PNG_LIBS = @PNG_LIBS@ POST_ALLOC_OBJ = @POST_ALLOC_OBJ@ PRAGMA_COLUMNS = @PRAGMA_COLUMNS@ PRAGMA_SYSTEM_HEADER = @PRAGMA_SYSTEM_HEADER@ @@ -716,8 +866,10 @@ REPLACE_GETGROUPS = @REPLACE_GETGROUPS@ REPLACE_GETLINE = @REPLACE_GETLINE@ REPLACE_GETLOGIN_R = @REPLACE_GETLOGIN_R@ REPLACE_GETPAGESIZE = @REPLACE_GETPAGESIZE@ +REPLACE_GETPASS = @REPLACE_GETPASS@ REPLACE_GETTIMEOFDAY = @REPLACE_GETTIMEOFDAY@ REPLACE_GMTIME = @REPLACE_GMTIME@ +REPLACE_INITSTATE = @REPLACE_INITSTATE@ REPLACE_ISATTY = @REPLACE_ISATTY@ REPLACE_LCHOWN = @REPLACE_LCHOWN@ REPLACE_LINK = @REPLACE_LINK@ @@ -753,6 +905,7 @@ REPLACE_PUTENV = @REPLACE_PUTENV@ REPLACE_PWRITE = @REPLACE_PWRITE@ REPLACE_QSORT_R = @REPLACE_QSORT_R@ REPLACE_RAISE = @REPLACE_RAISE@ +REPLACE_RANDOM = @REPLACE_RANDOM@ REPLACE_RANDOM_R = @REPLACE_RANDOM_R@ REPLACE_READ = @REPLACE_READ@ REPLACE_READLINK = @REPLACE_READLINK@ @@ -765,6 +918,7 @@ REPLACE_RENAMEAT = @REPLACE_RENAMEAT@ REPLACE_RMDIR = @REPLACE_RMDIR@ REPLACE_SELECT = @REPLACE_SELECT@ REPLACE_SETENV = @REPLACE_SETENV@ +REPLACE_SETSTATE = @REPLACE_SETSTATE@ REPLACE_SLEEP = @REPLACE_SLEEP@ REPLACE_SNPRINTF = @REPLACE_SNPRINTF@ REPLACE_SPRINTF = @REPLACE_SPRINTF@ @@ -786,6 +940,7 @@ REPLACE_STRSTR = @REPLACE_STRSTR@ REPLACE_STRTOD = @REPLACE_STRTOD@ REPLACE_STRTOIMAX = @REPLACE_STRTOIMAX@ REPLACE_STRTOK_R = @REPLACE_STRTOK_R@ +REPLACE_STRTOLD = @REPLACE_STRTOLD@ REPLACE_STRTOUMAX = @REPLACE_STRTOUMAX@ REPLACE_STRUCT_TIMEVAL = @REPLACE_STRUCT_TIMEVAL@ REPLACE_SYMLINK = @REPLACE_SYMLINK@ @@ -868,6 +1023,7 @@ XMKMF = @XMKMF@ XOBJ = @XOBJ@ XRANDR_CFLAGS = @XRANDR_CFLAGS@ XRANDR_LIBS = @XRANDR_LIBS@ +XRENDER_LIBS = @XRENDER_LIBS@ XWIDGETS_OBJ = @XWIDGETS_OBJ@ X_TOOLKIT_TYPE = @X_TOOLKIT_TYPE@ ac_ct_CC = @ac_ct_CC@ @@ -890,6 +1046,7 @@ datadir = @datadir@ datarootdir = @datarootdir@ docdir = @docdir@ dvidir = @dvidir@ +emacs_major_version = @emacs_major_version@ etcdir = @etcdir@ etcdocdir = @etcdocdir@ exec_prefix = @exec_prefix@ @@ -898,6 +1055,7 @@ gamegroup = @gamegroup@ gameuser = @gameuser@ gl_GNULIB_ENABLED_03e0aaad4cb89ca757653bd367a6ccb7 = @gl_GNULIB_ENABLED_03e0aaad4cb89ca757653bd367a6ccb7@ gl_GNULIB_ENABLED_2049e887c7e5308faad27b3f894bb8c9 = @gl_GNULIB_ENABLED_2049e887c7e5308faad27b3f894bb8c9@ +gl_GNULIB_ENABLED_21ee726a3540c09237a8e70c0baf7467 = @gl_GNULIB_ENABLED_21ee726a3540c09237a8e70c0baf7467@ gl_GNULIB_ENABLED_260941c0e5dc67ec9e87d1fb321c300b = @gl_GNULIB_ENABLED_260941c0e5dc67ec9e87d1fb321c300b@ gl_GNULIB_ENABLED_5264294aa0a5557541b53c8c741f7f31 = @gl_GNULIB_ENABLED_5264294aa0a5557541b53c8c741f7f31@ gl_GNULIB_ENABLED_6099e9737f757db36c47fa9d9f02e88c = @gl_GNULIB_ENABLED_6099e9737f757db36c47fa9d9f02e88c@ @@ -906,7 +1064,6 @@ gl_GNULIB_ENABLED_a9786850e999ae65a836a6041e8e5ed1 = @gl_GNULIB_ENABLED_a9786850 gl_GNULIB_ENABLED_be453cec5eecf5731a274f2de7f2db36 = @gl_GNULIB_ENABLED_be453cec5eecf5731a274f2de7f2db36@ gl_GNULIB_ENABLED_cloexec = @gl_GNULIB_ENABLED_cloexec@ gl_GNULIB_ENABLED_dirfd = @gl_GNULIB_ENABLED_dirfd@ -gl_GNULIB_ENABLED_dosname = @gl_GNULIB_ENABLED_dosname@ gl_GNULIB_ENABLED_euidaccess = @gl_GNULIB_ENABLED_euidaccess@ gl_GNULIB_ENABLED_getdtablesize = @gl_GNULIB_ENABLED_getdtablesize@ gl_GNULIB_ENABLED_getgroups = @gl_GNULIB_ENABLED_getgroups@ @@ -997,7 +1154,7 @@ ifneq (,$(GL_GENERATE_ALLOCA_H)) alloca.h: alloca.in.h $(top_builddir)/config.status $(AM_V_GEN)rm -f $@-t $@ && \ { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */'; \ - cat $(srcdir)/alloca.in.h; \ + sed -e 's|@''HAVE_ALLOCA_H''@|$(HAVE_ALLOCA_H)|g' < $(srcdir)/alloca.in.h; \ } > $@-t && \ mv -f $@-t $@ else @@ -1145,45 +1302,45 @@ EXTRA_DIST += count-trailing-zeros.h endif ## end gnulib module count-trailing-zeros -## begin gnulib module crypto/md5 -ifeq (,$(OMIT_GNULIB_MODULE_crypto/md5)) +## begin gnulib module crypto/md5-buffer +ifeq (,$(OMIT_GNULIB_MODULE_crypto/md5-buffer)) libgnu_a_SOURCES += md5.c EXTRA_DIST += gl_openssl.h md5.h endif -## end gnulib module crypto/md5 +## end gnulib module crypto/md5-buffer -## begin gnulib module crypto/sha1 -ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha1)) +## begin gnulib module crypto/sha1-buffer +ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha1-buffer)) libgnu_a_SOURCES += sha1.c EXTRA_DIST += gl_openssl.h sha1.h endif -## end gnulib module crypto/sha1 +## end gnulib module crypto/sha1-buffer -## begin gnulib module crypto/sha256 -ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha256)) +## begin gnulib module crypto/sha256-buffer +ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha256-buffer)) libgnu_a_SOURCES += sha256.c EXTRA_DIST += gl_openssl.h sha256.h endif -## end gnulib module crypto/sha256 +## end gnulib module crypto/sha256-buffer -## begin gnulib module crypto/sha512 -ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha512)) +## begin gnulib module crypto/sha512-buffer +ifeq (,$(OMIT_GNULIB_MODULE_crypto/sha512-buffer)) libgnu_a_SOURCES += sha512.c EXTRA_DIST += gl_openssl.h sha512.h endif -## end gnulib module crypto/sha512 +## end gnulib module crypto/sha512-buffer ## begin gnulib module diffseq ifeq (,$(OMIT_GNULIB_MODULE_diffseq)) @@ -1259,9 +1416,7 @@ endif ## begin gnulib module dosname ifeq (,$(OMIT_GNULIB_MODULE_dosname)) -ifneq (,$(gl_GNULIB_ENABLED_dosname)) -endif EXTRA_DIST += dosname.h endif @@ -1444,17 +1599,6 @@ EXTRA_DIST += fcntl.in.h endif ## end gnulib module fcntl-h -## begin gnulib module fdatasync -ifeq (,$(OMIT_GNULIB_MODULE_fdatasync)) - - -EXTRA_DIST += fdatasync.c - -EXTRA_libgnu_a_SOURCES += fdatasync.c - -endif -## end gnulib module fdatasync - ## begin gnulib module fdopendir ifeq (,$(OMIT_GNULIB_MODULE_fdopendir)) @@ -1517,6 +1661,17 @@ EXTRA_libgnu_a_SOURCES += at-func.c fstatat.c endif ## end gnulib module fstatat +## begin gnulib module fsusage +ifeq (,$(OMIT_GNULIB_MODULE_fsusage)) + + +EXTRA_DIST += fsusage.c fsusage.h + +EXTRA_libgnu_a_SOURCES += fsusage.c + +endif +## end gnulib module fsusage + ## begin gnulib module fsync ifeq (,$(OMIT_GNULIB_MODULE_fsync)) @@ -1572,7 +1727,7 @@ BUILT_SOURCES += $(GETOPT_H) $(GETOPT_CDEFS_H) # We need the following in order to create <getopt.h> when the system # doesn't have one that works with the given compiler. -getopt.h: getopt.in.h $(top_builddir)/config.status +getopt.h: getopt.in.h $(top_builddir)/config.status $(ARG_NONNULL_H) $(AM_V_GEN)rm -f $@-t $@ && \ { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */'; \ sed -e 's|@''GUARD_PREFIX''@|GL|g' \ @@ -1654,6 +1809,32 @@ EXTRA_libgnu_a_SOURCES += group-member.c endif ## end gnulib module group-member +## begin gnulib module ieee754-h +ifeq (,$(OMIT_GNULIB_MODULE_ieee754-h)) + +BUILT_SOURCES += $(IEEE754_H) + +# We need the following in order to create <ieee754.h> when the system +# doesn't have one that works with the given compiler. +ifneq (,$(GL_GENERATE_IEEE754_H)) +ieee754.h: ieee754.in.h $(top_builddir)/config.status + $(AM_V_GEN)rm -f $@-t && \ + { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */'; \ + sed -e 's/ifndef _GL_GNULIB_HEADER/if 0/g' \ + $(srcdir)/ieee754.in.h; \ + } > $@-t && \ + mv -f $@-t $@ +else +ieee754.h: $(top_builddir)/config.status + rm -f $@ +endif +MOSTLYCLEANFILES += ieee754.h ieee754.h-t + +EXTRA_DIST += ieee754.in.h + +endif +## end gnulib module ieee754-h + ## begin gnulib module ignore-value ifeq (,$(OMIT_GNULIB_MODULE_ignore-value)) @@ -1700,6 +1881,7 @@ inttypes.h: inttypes.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(WARN_ON_U -e 's/@''HAVE_DECL_IMAXDIV''@/$(HAVE_DECL_IMAXDIV)/g' \ -e 's/@''HAVE_DECL_STRTOIMAX''@/$(HAVE_DECL_STRTOIMAX)/g' \ -e 's/@''HAVE_DECL_STRTOUMAX''@/$(HAVE_DECL_STRTOUMAX)/g' \ + -e 's/@''HAVE_IMAXDIV_T''@/$(HAVE_IMAXDIV_T)/g' \ -e 's/@''REPLACE_STRTOIMAX''@/$(REPLACE_STRTOIMAX)/g' \ -e 's/@''REPLACE_STRTOUMAX''@/$(REPLACE_STRTOUMAX)/g' \ -e 's/@''INT32_MAX_LT_INTMAX_MAX''@/$(INT32_MAX_LT_INTMAX_MAX)/g' \ @@ -1719,6 +1901,17 @@ EXTRA_DIST += inttypes.in.h endif ## end gnulib module inttypes-incomplete +## begin gnulib module libc-config +ifeq (,$(OMIT_GNULIB_MODULE_libc-config)) + +ifneq (,$(gl_GNULIB_ENABLED_21ee726a3540c09237a8e70c0baf7467)) + +endif +EXTRA_DIST += cdefs.h libc-config.h + +endif +## end gnulib module libc-config + ## begin gnulib module limits-h ifeq (,$(OMIT_GNULIB_MODULE_limits-h)) @@ -1773,6 +1966,17 @@ EXTRA_libgnu_a_SOURCES += lstat.c endif ## end gnulib module lstat +## begin gnulib module memmem-simple +ifeq (,$(OMIT_GNULIB_MODULE_memmem-simple)) + + +EXTRA_DIST += memmem.c str-two-way.h + +EXTRA_libgnu_a_SOURCES += memmem.c + +endif +## end gnulib module memmem-simple + ## begin gnulib module memrchr ifeq (,$(OMIT_GNULIB_MODULE_memrchr)) @@ -1932,6 +2136,17 @@ EXTRA_libgnu_a_SOURCES += at-func.c readlinkat.c endif ## end gnulib module readlinkat +## begin gnulib module regex +ifeq (,$(OMIT_GNULIB_MODULE_regex)) + + +EXTRA_DIST += regcomp.c regex.c regex.h regex_internal.c regex_internal.h regexec.c + +EXTRA_libgnu_a_SOURCES += regcomp.c regex.c regex_internal.c regexec.c + +endif +## end gnulib module regex + ## begin gnulib module root-uid ifeq (,$(OMIT_GNULIB_MODULE_root-uid)) @@ -1969,8 +2184,8 @@ signal.h: signal.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's|@''PRAGMA_SYSTEM_HEADER''@|@PRAGMA_SYSTEM_HEADER@|g' \ -e 's|@''PRAGMA_COLUMNS''@|@PRAGMA_COLUMNS@|g' \ -e 's|@''NEXT_SIGNAL_H''@|$(NEXT_SIGNAL_H)|g' \ - -e 's|@''GNULIB_PTHREAD_SIGMASK''@|$(GNULIB_PTHREAD_SIGMASK)|g' \ - -e 's|@''GNULIB_RAISE''@|$(GNULIB_RAISE)|g' \ + -e 's/@''GNULIB_PTHREAD_SIGMASK''@/$(GNULIB_PTHREAD_SIGMASK)/g' \ + -e 's/@''GNULIB_RAISE''@/$(GNULIB_RAISE)/g' \ -e 's/@''GNULIB_SIGNAL_H_SIGPIPE''@/$(GNULIB_SIGNAL_H_SIGPIPE)/g' \ -e 's/@''GNULIB_SIGPROCMASK''@/$(GNULIB_SIGPROCMASK)/g' \ -e 's/@''GNULIB_SIGACTION''@/$(GNULIB_SIGACTION)/g' \ @@ -2350,6 +2565,7 @@ stdlib.h: stdlib.in.h $(top_builddir)/config.status $(CXXDEFS_H) \ -e 's/@''GNULIB_SECURE_GETENV''@/$(GNULIB_SECURE_GETENV)/g' \ -e 's/@''GNULIB_SETENV''@/$(GNULIB_SETENV)/g' \ -e 's/@''GNULIB_STRTOD''@/$(GNULIB_STRTOD)/g' \ + -e 's/@''GNULIB_STRTOLD''@/$(GNULIB_STRTOLD)/g' \ -e 's/@''GNULIB_STRTOLL''@/$(GNULIB_STRTOLL)/g' \ -e 's/@''GNULIB_STRTOULL''@/$(GNULIB_STRTOULL)/g' \ -e 's/@''GNULIB_SYSTEM_POSIX''@/$(GNULIB_SYSTEM_POSIX)/g' \ @@ -2363,7 +2579,9 @@ stdlib.h: stdlib.in.h $(top_builddir)/config.status $(CXXDEFS_H) \ -e 's|@''HAVE_DECL_GETLOADAVG''@|$(HAVE_DECL_GETLOADAVG)|g' \ -e 's|@''HAVE_GETSUBOPT''@|$(HAVE_GETSUBOPT)|g' \ -e 's|@''HAVE_GRANTPT''@|$(HAVE_GRANTPT)|g' \ + -e 's|@''HAVE_INITSTATE''@|$(HAVE_INITSTATE)|g' \ -e 's|@''HAVE_DECL_INITSTATE''@|$(HAVE_DECL_INITSTATE)|g' \ + -e 's|@''HAVE_MBTOWC''@|$(HAVE_MBTOWC)|g' \ -e 's|@''HAVE_MKDTEMP''@|$(HAVE_MKDTEMP)|g' \ -e 's|@''HAVE_MKOSTEMP''@|$(HAVE_MKOSTEMP)|g' \ -e 's|@''HAVE_MKOSTEMPS''@|$(HAVE_MKOSTEMPS)|g' \ @@ -2381,8 +2599,10 @@ stdlib.h: stdlib.in.h $(top_builddir)/config.status $(CXXDEFS_H) \ -e 's|@''HAVE_RPMATCH''@|$(HAVE_RPMATCH)|g' \ -e 's|@''HAVE_SECURE_GETENV''@|$(HAVE_SECURE_GETENV)|g' \ -e 's|@''HAVE_DECL_SETENV''@|$(HAVE_DECL_SETENV)|g' \ + -e 's|@''HAVE_SETSTATE''@|$(HAVE_SETSTATE)|g' \ -e 's|@''HAVE_DECL_SETSTATE''@|$(HAVE_DECL_SETSTATE)|g' \ -e 's|@''HAVE_STRTOD''@|$(HAVE_STRTOD)|g' \ + -e 's|@''HAVE_STRTOLD''@|$(HAVE_STRTOLD)|g' \ -e 's|@''HAVE_STRTOLL''@|$(HAVE_STRTOLL)|g' \ -e 's|@''HAVE_STRTOULL''@|$(HAVE_STRTOULL)|g' \ -e 's|@''HAVE_STRUCT_RANDOM_DATA''@|$(HAVE_STRUCT_RANDOM_DATA)|g' \ @@ -2391,6 +2611,7 @@ stdlib.h: stdlib.in.h $(top_builddir)/config.status $(CXXDEFS_H) \ -e 's|@''HAVE_DECL_UNSETENV''@|$(HAVE_DECL_UNSETENV)|g' \ -e 's|@''REPLACE_CALLOC''@|$(REPLACE_CALLOC)|g' \ -e 's|@''REPLACE_CANONICALIZE_FILE_NAME''@|$(REPLACE_CANONICALIZE_FILE_NAME)|g' \ + -e 's|@''REPLACE_INITSTATE''@|$(REPLACE_INITSTATE)|g' \ -e 's|@''REPLACE_MALLOC''@|$(REPLACE_MALLOC)|g' \ -e 's|@''REPLACE_MBTOWC''@|$(REPLACE_MBTOWC)|g' \ -e 's|@''REPLACE_MKSTEMP''@|$(REPLACE_MKSTEMP)|g' \ @@ -2398,11 +2619,14 @@ stdlib.h: stdlib.in.h $(top_builddir)/config.status $(CXXDEFS_H) \ -e 's|@''REPLACE_PTSNAME_R''@|$(REPLACE_PTSNAME_R)|g' \ -e 's|@''REPLACE_PUTENV''@|$(REPLACE_PUTENV)|g' \ -e 's|@''REPLACE_QSORT_R''@|$(REPLACE_QSORT_R)|g' \ + -e 's|@''REPLACE_RANDOM''@|$(REPLACE_RANDOM)|g' \ -e 's|@''REPLACE_RANDOM_R''@|$(REPLACE_RANDOM_R)|g' \ -e 's|@''REPLACE_REALLOC''@|$(REPLACE_REALLOC)|g' \ -e 's|@''REPLACE_REALPATH''@|$(REPLACE_REALPATH)|g' \ -e 's|@''REPLACE_SETENV''@|$(REPLACE_SETENV)|g' \ + -e 's|@''REPLACE_SETSTATE''@|$(REPLACE_SETSTATE)|g' \ -e 's|@''REPLACE_STRTOD''@|$(REPLACE_STRTOD)|g' \ + -e 's|@''REPLACE_STRTOLD''@|$(REPLACE_STRTOLD)|g' \ -e 's|@''REPLACE_UNSETENV''@|$(REPLACE_UNSETENV)|g' \ -e 's|@''REPLACE_WCTOMB''@|$(REPLACE_WCTOMB)|g' \ -e '/definitions of _GL_FUNCDECL_RPL/r $(CXXDEFS_H)' \ @@ -2910,6 +3134,7 @@ unistd.h: unistd.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's/@''GNULIB_GETLOGIN''@/$(GNULIB_GETLOGIN)/g' \ -e 's/@''GNULIB_GETLOGIN_R''@/$(GNULIB_GETLOGIN_R)/g' \ -e 's/@''GNULIB_GETPAGESIZE''@/$(GNULIB_GETPAGESIZE)/g' \ + -e 's/@''GNULIB_GETPASS''@/$(GNULIB_GETPASS)/g' \ -e 's/@''GNULIB_GETUSERSHELL''@/$(GNULIB_GETUSERSHELL)/g' \ -e 's/@''GNULIB_GROUP_MEMBER''@/$(GNULIB_GROUP_MEMBER)/g' \ -e 's/@''GNULIB_ISATTY''@/$(GNULIB_ISATTY)/g' \ @@ -2953,6 +3178,7 @@ unistd.h: unistd.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's|@''HAVE_GETGROUPS''@|$(HAVE_GETGROUPS)|g' \ -e 's|@''HAVE_GETHOSTNAME''@|$(HAVE_GETHOSTNAME)|g' \ -e 's|@''HAVE_GETPAGESIZE''@|$(HAVE_GETPAGESIZE)|g' \ + -e 's|@''HAVE_GETPASS''@|$(HAVE_GETPASS)|g' \ -e 's|@''HAVE_GROUP_MEMBER''@|$(HAVE_GROUP_MEMBER)|g' \ -e 's|@''HAVE_LCHOWN''@|$(HAVE_LCHOWN)|g' \ -e 's|@''HAVE_LINK''@|$(HAVE_LINK)|g' \ @@ -2967,7 +3193,6 @@ unistd.h: unistd.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's|@''HAVE_SLEEP''@|$(HAVE_SLEEP)|g' \ -e 's|@''HAVE_SYMLINK''@|$(HAVE_SYMLINK)|g' \ -e 's|@''HAVE_SYMLINKAT''@|$(HAVE_SYMLINKAT)|g' \ - -e 's|@''HAVE_TRUNCATE''@|$(HAVE_TRUNCATE)|g' \ -e 's|@''HAVE_UNLINKAT''@|$(HAVE_UNLINKAT)|g' \ -e 's|@''HAVE_USLEEP''@|$(HAVE_USLEEP)|g' \ -e 's|@''HAVE_DECL_ENVIRON''@|$(HAVE_DECL_ENVIRON)|g' \ @@ -2979,6 +3204,7 @@ unistd.h: unistd.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's|@''HAVE_DECL_GETPAGESIZE''@|$(HAVE_DECL_GETPAGESIZE)|g' \ -e 's|@''HAVE_DECL_GETUSERSHELL''@|$(HAVE_DECL_GETUSERSHELL)|g' \ -e 's|@''HAVE_DECL_SETHOSTNAME''@|$(HAVE_DECL_SETHOSTNAME)|g' \ + -e 's|@''HAVE_DECL_TRUNCATE''@|$(HAVE_DECL_TRUNCATE)|g' \ -e 's|@''HAVE_DECL_TTYNAME_R''@|$(HAVE_DECL_TTYNAME_R)|g' \ -e 's|@''HAVE_OS_H''@|$(HAVE_OS_H)|g' \ -e 's|@''HAVE_SYS_PARAM_H''@|$(HAVE_SYS_PARAM_H)|g' \ @@ -2996,6 +3222,7 @@ unistd.h: unistd.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNULL_H -e 's|@''REPLACE_GETLOGIN_R''@|$(REPLACE_GETLOGIN_R)|g' \ -e 's|@''REPLACE_GETGROUPS''@|$(REPLACE_GETGROUPS)|g' \ -e 's|@''REPLACE_GETPAGESIZE''@|$(REPLACE_GETPAGESIZE)|g' \ + -e 's|@''REPLACE_GETPASS''@|$(REPLACE_GETPASS)|g' \ -e 's|@''REPLACE_ISATTY''@|$(REPLACE_ISATTY)|g' \ -e 's|@''REPLACE_LCHOWN''@|$(REPLACE_LCHOWN)|g' \ -e 's|@''REPLACE_LINK''@|$(REPLACE_LINK)|g' \ diff --git a/lib/group-member.c b/lib/group-member.c index a14056f9a82..ad61cf0b630 100644 --- a/lib/group-member.c +++ b/lib/group-member.c @@ -1,7 +1,7 @@ /* group-member.c -- determine whether group id is in calling user's group list - Copyright (C) 1994, 1997-1998, 2003, 2005-2006, 2009-2019 Free - Software Foundation, Inc. + Copyright (C) 1994, 1997-1998, 2003, 2005-2006, 2009-2019 Free Software + Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/ieee754.in.h b/lib/ieee754.in.h new file mode 100644 index 00000000000..a079e59d791 --- /dev/null +++ b/lib/ieee754.in.h @@ -0,0 +1,222 @@ +/* Copyright (C) 1992-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ + +#ifndef _IEEE754_H + +#define _IEEE754_H 1 + +#ifndef _GL_GNULIB_HEADER +/* Ordinary glibc usage. */ +# include <features.h> +# include <endian.h> +#else +/* Gnulib usage. */ +# ifndef __BEGIN_DECLS +# ifdef __cplusplus +# define __BEGIN_DECLS extern "C" { +# define __END_DECLS } +# else +# define __BEGIN_DECLS +# define __END_DECLS +# endif +# endif +# ifndef __FLOAT_WORD_ORDER +# define __LITTLE_ENDIAN 1234 +# define __BIG_ENDIAN 4321 +# ifdef WORDS_BIGENDIAN +# define __BYTE_ORDER __BIG_ENDIAN +# else +# define __BYTE_ORDER __LITTLE_ENDIAN +# endif +# define __FLOAT_WORD_ORDER __BYTE_ORDER +# endif +#endif + +__BEGIN_DECLS + +union ieee754_float + { + float f; + + /* This is the IEEE 754 single-precision format. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:8; + unsigned int mantissa:23; +#endif /* Big endian. */ +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int mantissa:23; + unsigned int exponent:8; + unsigned int negative:1; +#endif /* Little endian. */ + } ieee; + + /* This format makes it easier to see if a NaN is a signalling NaN. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:8; + unsigned int quiet_nan:1; + unsigned int mantissa:22; +#endif /* Big endian. */ +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int mantissa:22; + unsigned int quiet_nan:1; + unsigned int exponent:8; + unsigned int negative:1; +#endif /* Little endian. */ + } ieee_nan; + }; + +#define IEEE754_FLOAT_BIAS 0x7f /* Added to exponent. */ + + +union ieee754_double + { + double d; + + /* This is the IEEE 754 double-precision format. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:11; + /* Together these comprise the mantissa. */ + unsigned int mantissa0:20; + unsigned int mantissa1:32; +#endif /* Big endian. */ +#if __BYTE_ORDER == __LITTLE_ENDIAN +# if __FLOAT_WORD_ORDER == __BIG_ENDIAN + unsigned int mantissa0:20; + unsigned int exponent:11; + unsigned int negative:1; + unsigned int mantissa1:32; +# else + /* Together these comprise the mantissa. */ + unsigned int mantissa1:32; + unsigned int mantissa0:20; + unsigned int exponent:11; + unsigned int negative:1; +# endif +#endif /* Little endian. */ + } ieee; + + /* This format makes it easier to see if a NaN is a signalling NaN. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:11; + unsigned int quiet_nan:1; + /* Together these comprise the mantissa. */ + unsigned int mantissa0:19; + unsigned int mantissa1:32; +#else +# if __FLOAT_WORD_ORDER == __BIG_ENDIAN + unsigned int mantissa0:19; + unsigned int quiet_nan:1; + unsigned int exponent:11; + unsigned int negative:1; + unsigned int mantissa1:32; +# else + /* Together these comprise the mantissa. */ + unsigned int mantissa1:32; + unsigned int mantissa0:19; + unsigned int quiet_nan:1; + unsigned int exponent:11; + unsigned int negative:1; +# endif +#endif + } ieee_nan; + }; + +#define IEEE754_DOUBLE_BIAS 0x3ff /* Added to exponent. */ + + +union ieee854_long_double + { + long double d; + + /* This is the IEEE 854 double-extended-precision format. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:15; + unsigned int empty:16; + unsigned int mantissa0:32; + unsigned int mantissa1:32; +#endif +#if __BYTE_ORDER == __LITTLE_ENDIAN +# if __FLOAT_WORD_ORDER == __BIG_ENDIAN + unsigned int exponent:15; + unsigned int negative:1; + unsigned int empty:16; + unsigned int mantissa0:32; + unsigned int mantissa1:32; +# else + unsigned int mantissa1:32; + unsigned int mantissa0:32; + unsigned int exponent:15; + unsigned int negative:1; + unsigned int empty:16; +# endif +#endif + } ieee; + + /* This is for NaNs in the IEEE 854 double-extended-precision format. */ + struct + { +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int negative:1; + unsigned int exponent:15; + unsigned int empty:16; + unsigned int one:1; + unsigned int quiet_nan:1; + unsigned int mantissa0:30; + unsigned int mantissa1:32; +#endif +#if __BYTE_ORDER == __LITTLE_ENDIAN +# if __FLOAT_WORD_ORDER == __BIG_ENDIAN + unsigned int exponent:15; + unsigned int negative:1; + unsigned int empty:16; + unsigned int mantissa0:30; + unsigned int quiet_nan:1; + unsigned int one:1; + unsigned int mantissa1:32; +# else + unsigned int mantissa1:32; + unsigned int mantissa0:30; + unsigned int quiet_nan:1; + unsigned int one:1; + unsigned int exponent:15; + unsigned int negative:1; + unsigned int empty:16; +# endif +#endif + } ieee_nan; + }; + +#define IEEE854_LONG_DOUBLE_BIAS 0x3fff + +__END_DECLS + +#endif /* ieee754.h */ diff --git a/lib/intprops.h b/lib/intprops.h index 592371469d4..1a44ae55653 100644 --- a/lib/intprops.h +++ b/lib/intprops.h @@ -22,12 +22,13 @@ #include <limits.h> -/* Return a value with the common real type of E and V and the value of V. */ -#define _GL_INT_CONVERT(e, v) (0 * (e) + (v)) +/* Return a value with the common real type of E and V and the value of V. + Do not evaluate E. */ +#define _GL_INT_CONVERT(e, v) ((1 ? 0 : (e)) + (v)) /* Act like _GL_INT_CONVERT (E, -V) but work around a bug in IRIX 6.5 cc; see <https://lists.gnu.org/r/bug-gnulib/2011-05/msg00406.html>. */ -#define _GL_INT_NEGATE_CONVERT(e, v) (0 * (e) - (v)) +#define _GL_INT_NEGATE_CONVERT(e, v) ((1 ? 0 : (e)) - (v)) /* The extra casts in the following macros work around compiler bugs, e.g., in Cray C 5.0.3.0. */ @@ -40,13 +41,14 @@ #define TYPE_SIGNED(t) (! ((t) 0 < (t) -1)) /* Return 1 if the real expression E, after promotion, has a - signed or floating type. */ + signed or floating type. Do not evaluate E. */ #define EXPR_SIGNED(e) (_GL_INT_NEGATE_CONVERT (e, 1) < 0) /* Minimum and maximum values for integer types and expressions. */ /* The width in bits of the integer type or expression T. + Do not evaluate T. Padding bits are not supported; this is checked at compile-time below. */ #define TYPE_WIDTH(t) (sizeof (t) * CHAR_BIT) @@ -58,7 +60,7 @@ : ((((t) 1 << (TYPE_WIDTH (t) - 2)) - 1) * 2 + 1))) /* The maximum and minimum values for the type of the expression E, - after integer promotion. E should not have side effects. */ + after integer promotion. E is not evaluated. */ #define _GL_INT_MINIMUM(e) \ (EXPR_SIGNED (e) \ ? ~ _GL_SIGNED_INT_MAXIMUM (e) \ @@ -340,8 +342,8 @@ Arguments should be free of side effects. */ #define _GL_BINARY_OP_OVERFLOW(a, b, op_result_overflow) \ op_result_overflow (a, b, \ - _GL_INT_MINIMUM (0 * (b) + (a)), \ - _GL_INT_MAXIMUM (0 * (b) + (a))) + _GL_INT_MINIMUM (_GL_INT_CONVERT (a, b)), \ + _GL_INT_MAXIMUM (_GL_INT_CONVERT (a, b))) /* Store the low-order bits of A + B, A - B, A * B, respectively, into *R. Return 1 if the result overflows. See above for restrictions. */ diff --git a/lib/inttypes.in.h b/lib/inttypes.in.h index 8fe89db4659..d3c735c12ce 100644 --- a/lib/inttypes.in.h +++ b/lib/inttypes.in.h @@ -52,7 +52,7 @@ /* Get CHAR_BIT. */ #include <limits.h> /* On mingw, __USE_MINGW_ANSI_STDIO only works if <stdio.h> is also included */ -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ # include <stdio.h> #endif @@ -1067,11 +1067,13 @@ _GL_WARN_ON_USE (imaxabs, "imaxabs is unportable - " #endif #if @GNULIB_IMAXDIV@ -# if !@HAVE_DECL_IMAXDIV@ +# if !@HAVE_IMAXDIV_T@ # if !GNULIB_defined_imaxdiv_t typedef struct { intmax_t quot; intmax_t rem; } imaxdiv_t; # define GNULIB_defined_imaxdiv_t 1 # endif +# endif +# if !@HAVE_DECL_IMAXDIV@ extern imaxdiv_t imaxdiv (intmax_t, intmax_t); # endif #elif defined GNULIB_POSIXCHECK diff --git a/lib/libc-config.h b/lib/libc-config.h new file mode 100644 index 00000000000..57c69661d2f --- /dev/null +++ b/lib/libc-config.h @@ -0,0 +1,174 @@ +/* System definitions for code taken from the GNU C Library + + Copyright 2017-2019 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with this program; if not, see + <https://www.gnu.org/licenses/>. */ + +/* Written by Paul Eggert. */ + +/* This is intended to be a good-enough substitute for glibc system + macros like those defined in <sys/cdefs.h>, so that Gnulib code + shared with glibc can do this as the first #include: + + #ifndef _LIBC + # include <libc-config.h> + #endif + + When compiled as part of glibc this is a no-op; when compiled as + part of Gnulib this includes Gnulib's <config.h> and defines macros + that glibc library code would normally assume. */ + +#include <config.h> + +/* On glibc this includes <features.h> and <sys/cdefs.h> and #defines + _FEATURES_H, __WORDSIZE, and __set_errno. On FreeBSD 11 it + includes <sys/cdefs.h> which defines __nonnull. Elsewhere it + is harmless. */ +#include <errno.h> + +/* From glibc <errno.h>. */ +#ifndef __set_errno +# define __set_errno(val) (errno = (val)) +#endif + +/* From glibc <features.h>. */ + +#ifndef __GNUC_PREREQ +# if defined __GNUC__ && defined __GNUC_MINOR__ +# define __GNUC_PREREQ(maj, min) ((maj) < __GNUC__ + ((min) <= __GNUC_MINOR__)) +# else +# define __GNUC_PREREQ(maj, min) 0 +# endif +#endif + +#ifndef __glibc_clang_prereq +# if defined __clang_major__ && defined __clang_minor__ +# define __glibc_clang_prereq(maj, min) \ + ((maj) < __clang_major__ + ((min) <= __clang_minor__)) +# else +# define __glibc_clang_prereq(maj, min) 0 +# endif +#endif + + +/* Prepare to include <cdefs.h>, which is our copy of glibc + <sys/cdefs.h>. */ + +/* Define _FEATURES_H so that <cdefs.h> does not include <features.h>. */ +#ifndef _FEATURES_H +# define _FEATURES_H 1 +#endif +/* Define __WORDSIZE so that <cdefs.h> does not attempt to include + nonexistent files. Make it a syntax error, since Gnulib does not + use __WORDSIZE now, and if Gnulib uses it later the syntax error + will let us know that __WORDSIZE needs configuring. */ +#ifndef __WORDSIZE +# define __WORDSIZE %%% +#endif +/* Undef the macros unconditionally defined by our copy of glibc + <sys/cdefs.h>, so that they do not clash with any system-defined + versions. */ +#undef _SYS_CDEFS_H +#undef __ASMNAME +#undef __ASMNAME2 +#undef __BEGIN_DECLS +#undef __CONCAT +#undef __END_DECLS +#undef __HAVE_GENERIC_SELECTION +#undef __LDBL_COMPAT +#undef __LDBL_REDIR +#undef __LDBL_REDIR1 +#undef __LDBL_REDIR1_DECL +#undef __LDBL_REDIR1_NTH +#undef __LDBL_REDIR_DECL +#undef __LDBL_REDIR_NTH +#undef __LEAF +#undef __LEAF_ATTR +#undef __NTH +#undef __NTHNL +#undef __P +#undef __PMT +#undef __REDIRECT +#undef __REDIRECT_LDBL +#undef __REDIRECT_NTH +#undef __REDIRECT_NTHNL +#undef __REDIRECT_NTH_LDBL +#undef __STRING +#undef __THROW +#undef __THROWNL +#undef __always_inline +#undef __attribute__ +#undef __attribute_alloc_size__ +#undef __attribute_artificial__ +#undef __attribute_const__ +#undef __attribute_deprecated__ +#undef __attribute_deprecated_msg__ +#undef __attribute_format_arg__ +#undef __attribute_format_strfmon__ +#undef __attribute_malloc__ +#undef __attribute_noinline__ +#undef __attribute_nonstring__ +#undef __attribute_pure__ +#undef __attribute_used__ +#undef __attribute_warn_unused_result__ +#undef __bos +#undef __bos0 +#undef __errordecl +#undef __extension__ +#undef __extern_always_inline +#undef __extern_inline +#undef __flexarr +#undef __fortify_function +#undef __glibc_c99_flexarr_available +#undef __glibc_clang_has_extension +#undef __glibc_likely +#undef __glibc_macro_warning +#undef __glibc_macro_warning1 +#undef __glibc_unlikely +#undef __inline +#undef __ptr_t +#undef __restrict +#undef __restrict_arr +#undef __va_arg_pack +#undef __va_arg_pack_len +#undef __warnattr +#undef __warndecl + +/* Include our copy of glibc <sys/cdefs.h>. */ +#include <cdefs.h> + +/* <cdefs.h> __inline is too pessimistic for non-GCC. */ +#undef __inline +#ifndef HAVE___INLINE +# if 199901 <= __STDC_VERSION__ || defined inline +# define __inline inline +# else +# define __inline +# endif +#endif + + +/* A substitute for glibc <libc-symbols.h>, good enough for Gnulib. */ +#define attribute_hidden +#define libc_hidden_proto(name, ...) +#define libc_hidden_def(name) +#define libc_hidden_weak(name) +#define libc_hidden_ver(local, name) +#define strong_alias(name, aliasname) +#define weak_alias(name, aliasname) + +/* A substitute for glibc <shlib-compat.h>, good enough for Gnulib. */ +#define SHLIB_COMPAT(lib, introduced, obsoleted) 0 +#define versioned_symbol(lib, local, symbol, version) diff --git a/lib/limits.in.h b/lib/limits.in.h index 0788a4e67e6..39750b38d1a 100644 --- a/lib/limits.in.h +++ b/lib/limits.in.h @@ -28,15 +28,32 @@ #ifndef _@GUARD_PREFIX@_LIMITS_H #define _@GUARD_PREFIX@_LIMITS_H -/* For HP-UX 11.31. */ -#if defined LONG_LONG_MIN && !defined LLONG_MIN -# define LLONG_MIN LONG_LONG_MIN +#ifndef LLONG_MIN +# if defined LONG_LONG_MIN /* HP-UX 11.31 */ +# define LLONG_MIN LONG_LONG_MIN +# elif defined LONGLONG_MIN /* IRIX 6.5 */ +# define LLONG_MIN LONGLONG_MIN +# elif defined __GNUC__ +# define LLONG_MIN (- __LONG_LONG_MAX__ - 1LL) +# endif #endif -#if defined LONG_LONG_MAX && !defined LLONG_MAX -# define LLONG_MAX LONG_LONG_MAX +#ifndef LLONG_MAX +# if defined LONG_LONG_MAX /* HP-UX 11.31 */ +# define LLONG_MAX LONG_LONG_MAX +# elif defined LONGLONG_MAX /* IRIX 6.5 */ +# define LLONG_MAX LONGLONG_MAX +# elif defined __GNUC__ +# define LLONG_MAX __LONG_LONG_MAX__ +# endif #endif -#if defined ULONG_LONG_MAX && !defined ULLONG_MAX -# define ULLONG_MAX ULONG_LONG_MAX +#ifndef ULLONG_MAX +# if defined ULONG_LONG_MAX /* HP-UX 11.31 */ +# define ULLONG_MAX ULONG_LONG_MAX +# elif defined ULONGLONG_MAX /* IRIX 6.5 */ +# define ULLONG_MAX ULONGLONG_MAX +# elif defined __GNUC__ +# define ULLONG_MAX (__LONG_LONG_MAX__ * 2ULL + 1ULL) +# endif #endif /* The number of usable bits in an unsigned or signed integer type @@ -53,6 +70,19 @@ #define _GL_COB8(n) (_GL_COB4 ((n) >> 4) + _GL_COB4 (n)) #define _GL_COB4(n) (!!((n) & 8) + !!((n) & 4) + !!((n) & 2) + !!((n) & 1)) +#ifndef WORD_BIT +/* Assume 'int' is 32 bits wide. */ +# define WORD_BIT 32 +#endif +#ifndef LONG_BIT +/* Assume 'long' is 32 or 64 bits wide. */ +# if LONG_MAX == INT_MAX +# define LONG_BIT 32 +# else +# define LONG_BIT 64 +# endif +#endif + /* Macros specified by ISO/IEC TS 18661-1:2014. */ #if (! defined ULLONG_WIDTH \ diff --git a/lib/localtime-buffer.c b/lib/localtime-buffer.c index f98e0b8f37a..b65ea45af20 100644 --- a/lib/localtime-buffer.c +++ b/lib/localtime-buffer.c @@ -1,7 +1,6 @@ /* Provide access to the last buffer returned by localtime() or gmtime(). - Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -35,6 +34,7 @@ struct tm *localtime_buffer_addr = &tm_zero_buffer; struct tm * rpl_localtime (time_t const *timep) +#undef localtime { struct tm *tm = localtime (timep); @@ -47,6 +47,7 @@ rpl_localtime (time_t const *timep) /* Same as above, since gmtime and localtime use the same buffer. */ struct tm * rpl_gmtime (time_t const *timep) +#undef gmtime { struct tm *tm = gmtime (timep); diff --git a/lib/localtime-buffer.h b/lib/localtime-buffer.h index cf15cbd3433..031111a752f 100644 --- a/lib/localtime-buffer.h +++ b/lib/localtime-buffer.h @@ -1,7 +1,6 @@ /* Provide access to the last buffer returned by localtime() or gmtime(). - Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 2001-2003, 2005-2007, 2009-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/lstat.c b/lib/lstat.c index d57ca105fd3..a3e40d826f0 100644 --- a/lib/lstat.c +++ b/lib/lstat.c @@ -42,10 +42,14 @@ orig_lstat (const char *filename, struct stat *buf) } /* Specification. */ +# ifdef __osf__ /* Write "sys/stat.h" here, not <sys/stat.h>, otherwise OSF/1 5.1 DTK cc eliminates this include because of the preliminary #include <sys/stat.h> above. */ -# include "sys/stat.h" +# include "sys/stat.h" +# else +# include <sys/stat.h> +# endif # include "stat-time.h" diff --git a/lib/md5.c b/lib/md5.c index 3d797ef8fea..5c8bcf91e48 100644 --- a/lib/md5.c +++ b/lib/md5.c @@ -1,7 +1,7 @@ /* Functions to compute MD5 message digest of files or memory blocks. according to the definition of MD5 in RFC 1321 from April 1992. - Copyright (C) 1995-1997, 1999-2001, 2005-2006, 2008-2019 Free - Software Foundation, Inc. + Copyright (C) 1995-1997, 1999-2001, 2005-2006, 2008-2019 Free Software + Foundation, Inc. This file is part of the GNU C Library. This program is free software; you can redistribute it and/or modify it @@ -52,9 +52,9 @@ # define md5_buffer __md5_buffer #endif +#include <byteswap.h> #ifdef WORDS_BIGENDIAN -# define SWAP(n) \ - (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24)) +# define SWAP(n) bswap_32 (n) #else # define SWAP(n) (n) #endif @@ -134,21 +134,29 @@ md5_finish_ctx (struct md5_ctx *ctx, void *resbuf) } #endif +#if defined _LIBC || defined GL_COMPILE_CRYPTO_STREAM + +#include "af_alg.h" + /* Compute MD5 message digest for bytes read from STREAM. The resulting message digest number will be written into the 16 bytes beginning at RESBLOCK. */ int md5_stream (FILE *stream, void *resblock) { - struct md5_ctx ctx; - size_t sum; + switch (afalg_stream (stream, "md5", resblock, MD5_DIGEST_SIZE)) + { + case 0: return 0; + case -EIO: return 1; + } char *buffer = malloc (BLOCKSIZE + 72); if (!buffer) return 1; - /* Initialize the computation context. */ + struct md5_ctx ctx; md5_init_ctx (&ctx); + size_t sum; /* Iterate over full file contents. */ while (1) @@ -162,6 +170,14 @@ md5_stream (FILE *stream, void *resblock) /* Read block. Take care for partial reads. */ while (1) { + /* Either process a partial fread() from this loop, + or the fread() in afalg_stream may have gotten EOF. + We need to avoid a subsequent fread() as EOF may + not be sticky. For details of such systems, see: + https://sourceware.org/bugzilla/show_bug.cgi?id=1190 */ + if (feof (stream)) + goto process_partial_block; + n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); sum += n; @@ -181,12 +197,6 @@ md5_stream (FILE *stream, void *resblock) } goto process_partial_block; } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; } /* Process buffer with BLOCKSIZE bytes. Note that @@ -206,6 +216,7 @@ process_partial_block: free (buffer); return 0; } +#endif #if ! HAVE_OPENSSL_MD5 /* Compute MD5 message digest for LEN bytes beginning at BUFFER. The diff --git a/lib/md5.h b/lib/md5.h index b41eaf42c13..478a27bc518 100644 --- a/lib/md5.h +++ b/lib/md5.h @@ -1,7 +1,7 @@ /* Declaration of functions and data types used for MD5 sum computing library functions. - Copyright (C) 1995-1997, 1999-2001, 2004-2006, 2008-2019 Free - Software Foundation, Inc. + Copyright (C) 1995-1997, 1999-2001, 2004-2006, 2008-2019 Free Software + Foundation, Inc. This file is part of the GNU C Library. This program is free software; you can redistribute it and/or modify it @@ -122,8 +122,11 @@ extern void *__md5_buffer (const char *buffer, size_t len, void *resblock) __THROW; # endif -/* Compute MD5 message digest for bytes read from STREAM. The - resulting message digest number will be written into the 16 bytes +/* Compute MD5 message digest for bytes read from STREAM. + STREAM is an open file stream. Regular files are handled more efficiently. + The contents of STREAM from its current position to its end will be read. + The case that the last operation on STREAM was an 'ungetc' is not supported. + The resulting message digest number will be written into the 16 bytes beginning at RESBLOCK. */ extern int __md5_stream (FILE *stream, void *resblock) __THROW; diff --git a/lib/memmem.c b/lib/memmem.c new file mode 100644 index 00000000000..12ae24f41b4 --- /dev/null +++ b/lib/memmem.c @@ -0,0 +1,71 @@ +/* Copyright (C) 1991-1994, 1996-1998, 2000, 2004, 2007-2019 Free Software + Foundation, Inc. + This file is part of the GNU C Library. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, see <https://www.gnu.org/licenses/>. */ + +/* This particular implementation was written by Eric Blake, 2008. */ + +#ifndef _LIBC +# include <config.h> +#endif + +/* Specification of memmem. */ +#include <string.h> + +#define RETURN_TYPE void * +#define AVAILABLE(h, h_l, j, n_l) ((j) <= (h_l) - (n_l)) +#include "str-two-way.h" + +/* Return the first occurrence of NEEDLE in HAYSTACK. Return HAYSTACK + if NEEDLE_LEN is 0, otherwise NULL if NEEDLE is not found in + HAYSTACK. */ +void * +memmem (const void *haystack_start, size_t haystack_len, + const void *needle_start, size_t needle_len) +{ + /* Abstract memory is considered to be an array of 'unsigned char' values, + not an array of 'char' values. See ISO C 99 section 6.2.6.1. */ + const unsigned char *haystack = (const unsigned char *) haystack_start; + const unsigned char *needle = (const unsigned char *) needle_start; + + if (needle_len == 0) + /* The first occurrence of the empty string is deemed to occur at + the beginning of the string. */ + return (void *) haystack; + + /* Sanity check, otherwise the loop might search through the whole + memory. */ + if (__builtin_expect (haystack_len < needle_len, 0)) + return NULL; + + /* Use optimizations in memchr when possible, to reduce the search + size of haystack using a linear algorithm with a smaller + coefficient. However, avoid memchr for long needles, since we + can often achieve sublinear performance. */ + if (needle_len < LONG_NEEDLE_THRESHOLD) + { + haystack = memchr (haystack, *needle, haystack_len); + if (!haystack || __builtin_expect (needle_len == 1, 0)) + return (void *) haystack; + haystack_len -= haystack - (const unsigned char *) haystack_start; + if (haystack_len < needle_len) + return NULL; + return two_way_short_needle (haystack, haystack_len, needle, needle_len); + } + else + return two_way_long_needle (haystack, haystack_len, needle, needle_len); +} + +#undef LONG_NEEDLE_THRESHOLD diff --git a/lib/memrchr.c b/lib/memrchr.c index da7c988efb6..96022835cef 100644 --- a/lib/memrchr.c +++ b/lib/memrchr.c @@ -1,7 +1,7 @@ /* memrchr -- find the last occurrence of a byte in a memory block - Copyright (C) 1991, 1993, 1996-1997, 1999-2000, 2003-2019 Free - Software Foundation, Inc. + Copyright (C) 1991, 1993, 1996-1997, 1999-2000, 2003-2019 Free Software + Foundation, Inc. Based on strlen implementation by Torbjorn Granlund (tege@sics.se), with help from Dan Sahlin (dan@sics.se) and @@ -68,7 +68,7 @@ __memrchr (void const *s, int c_in, size_t n) if (*--char_ptr == c) return (void *) char_ptr; - longword_ptr = (const longword *) char_ptr; + longword_ptr = (const void *) char_ptr; /* All these elucidatory comments refer to 4-byte longwords, but the theory applies equally well to any size longwords. */ diff --git a/lib/mktime-internal.h b/lib/mktime-internal.h index 8d1c97faa5e..52ab7814429 100644 --- a/lib/mktime-internal.h +++ b/lib/mktime-internal.h @@ -1,37 +1,79 @@ -/* mktime variant that also uses an offset guess - +/* Internals of mktime and related functions Copyright 2016-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Paul Eggert <eggert@cs.ucla.edu>. - This program is free software; you can redistribute it and/or + The GNU C Library is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version. - This program is distributed in the hope that it will be useful, + The GNU C Library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public - License along with this program; if not, see + License along with the GNU C Library; if not, see <https://www.gnu.org/licenses/>. */ -#include <time.h> +#ifndef _LIBC +# include <time.h> +#endif /* mktime_offset_t is a signed type wide enough to hold a UTC offset in seconds, and used as part of the type of the offset-guess - argument to mktime_internal. Use time_t on platforms where time_t + argument to mktime_internal. In Glibc, it is always long int. + When in Gnulib, use time_t on platforms where time_t is signed, to be compatible with platforms like BeOS that export this implementation detail of mktime. On platforms where time_t is unsigned, GNU and POSIX code can assume 'int' is at least 32 bits which is wide enough for a UTC offset. */ - -#if TIME_T_IS_SIGNED +#ifdef _LIBC +typedef long int mktime_offset_t; +#elif defined TIME_T_IS_SIGNED typedef time_t mktime_offset_t; #else typedef int mktime_offset_t; #endif -time_t mktime_internal (struct tm *, - struct tm * (*) (time_t const *, struct tm *), - mktime_offset_t *); +/* The source code uses identifiers like __time64_t for glibc + timestamps that can contain 64-bit values even when time_t is only + 32 bits. These are just macros for the ordinary identifiers unless + compiling within glibc when time_t is 32 bits. */ +#if ! (defined _LIBC && __TIMESIZE != 64) +# undef __time64_t +# define __time64_t time_t +# define __gmtime64_r __gmtime_r +# define __localtime64_r __localtime_r +# define __mktime64 mktime +# define __timegm64 timegm +#endif + +#ifndef _LIBC + +/* Although glibc source code uses leading underscores, Gnulib wants + ordinary names. + + Portable standalone applications should supply a <time.h> that + declares a POSIX-compliant localtime_r, for the benefit of older + implementations that lack localtime_r or have a nonstandard one. + Similarly for gmtime_r. See the gnulib time_r module for one way + to implement this. */ + +# undef __gmtime_r +# undef __localtime_r +# define __gmtime_r gmtime_r +# define __localtime_r localtime_r + +# define __mktime_internal mktime_internal + +#endif + +/* Subroutine of mktime. Return the time_t representation of TP and + normalize TP, given that a struct tm * maps to a time_t as performed + by FUNC. Record next guess for localtime-gmtime offset in *OFFSET. */ +extern __time64_t __mktime_internal (struct tm *tp, + struct tm *(*func) (__time64_t const *, + struct tm *), + mktime_offset_t *offset) attribute_hidden; diff --git a/lib/mktime.c b/lib/mktime.c index 43efa0ced5c..b00af96c8c4 100644 --- a/lib/mktime.c +++ b/lib/mktime.c @@ -17,27 +17,21 @@ License along with the GNU C Library; if not, see <https://www.gnu.org/licenses/>. */ -/* Define this to 1 to have a standalone program to test this implementation of - mktime. */ -#ifndef DEBUG_MKTIME -# define DEBUG_MKTIME 0 -#endif - /* The following macros influence what gets defined when this file is compiled: Macro/expression Which gnulib module This compilation unit should define + _LIBC (glibc proper) mktime + NEED_MKTIME_WORKING mktime rpl_mktime || NEED_MKTIME_WINDOWS NEED_MKTIME_INTERNAL mktime-internal mktime_internal - - DEBUG_MKTIME (defined manually) my_mktime, main */ -#if !defined _LIBC && !DEBUG_MKTIME -# include <config.h> +#ifndef _LIBC +# include <libc-config.h> #endif /* Assume that leap seconds are possible, unless told otherwise. @@ -49,47 +43,87 @@ #include <time.h> +#include <errno.h> #include <limits.h> #include <stdbool.h> +#include <stdlib.h> +#include <string.h> #include <intprops.h> #include <verify.h> -#if DEBUG_MKTIME -# include <stdio.h> -# include <stdlib.h> -# include <string.h> -/* Make it work even if the system's libc has its own mktime routine. */ -# undef mktime -# define mktime my_mktime +#ifndef NEED_MKTIME_INTERNAL +# define NEED_MKTIME_INTERNAL 0 #endif +#ifndef NEED_MKTIME_WINDOWS +# define NEED_MKTIME_WINDOWS 0 +#endif +#ifndef NEED_MKTIME_WORKING +# define NEED_MKTIME_WORKING 0 +#endif + +#include "mktime-internal.h" -#if NEED_MKTIME_WINDOWS /* on native Windows */ -# include <stdlib.h> -# include <string.h> +#if !defined _LIBC && (NEED_MKTIME_WORKING || NEED_MKTIME_WINDOWS) +static void +my_tzset (void) +{ +# if NEED_MKTIME_WINDOWS + /* Rectify the value of the environment variable TZ. + There are four possible kinds of such values: + - Traditional US time zone names, e.g. "PST8PDT". Syntax: see + <https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/tzset> + - Time zone names based on geography, that contain one or more + slashes, e.g. "Europe/Moscow". + - Time zone names based on geography, without slashes, e.g. + "Singapore". + - Time zone names that contain explicit DST rules. Syntax: see + <http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap08.html#tag_08_03> + The Microsoft CRT understands only the first kind. It produces incorrect + results if the value of TZ is of the other kinds. + But in a Cygwin environment, /etc/profile.d/tzset.sh sets TZ to a value + of the second kind for most geographies, or of the first kind in a few + other geographies. If it is of the second kind, neutralize it. For the + Microsoft CRT, an absent or empty TZ means the time zone that the user + has set in the Windows Control Panel. + If the value of TZ is of the third or fourth kind -- Cygwin programs + understand these syntaxes as well --, it does not matter whether we + neutralize it or not, since these values occur only when a Cygwin user + has set TZ explicitly; this case is 1. rare and 2. under the user's + responsibility. */ + const char *tz = getenv ("TZ"); + if (tz != NULL && strchr (tz, '/') != NULL) + _putenv ("TZ="); +# elif HAVE_TZSET + tzset (); +# endif +} +# undef __tzset +# define __tzset() my_tzset () #endif -#if NEED_MKTIME_WORKING || NEED_MKTIME_INTERNAL || DEBUG_MKTIME +#if defined _LIBC || NEED_MKTIME_WORKING || NEED_MKTIME_INTERNAL /* A signed type that can represent an integer number of years - multiplied by three times the number of seconds in a year. It is + multiplied by four times the number of seconds in a year. It is needed when converting a tm_year value times the number of seconds - in a year. The factor of three comes because these products need + in a year. The factor of four comes because these products need to be subtracted from each other, and sometimes with an offset - added to them, without worrying about overflow. + added to them, and then with another timestamp added, without + worrying about overflow. - Much of the code uses long_int to represent time_t values, to - lessen the hassle of dealing with platforms where time_t is + Much of the code uses long_int to represent __time64_t values, to + lessen the hassle of dealing with platforms where __time64_t is unsigned, and because long_int should suffice to represent all - time_t values that mktime can generate even on platforms where - time_t is excessively wide. */ + __time64_t values that mktime can generate even on platforms where + __time64_t is wider than the int components of struct tm. */ -#if INT_MAX <= LONG_MAX / 3 / 366 / 24 / 60 / 60 +#if INT_MAX <= LONG_MAX / 4 / 366 / 24 / 60 / 60 typedef long int long_int; #else typedef long long int long_int; #endif -verify (INT_MAX <= TYPE_MAXIMUM (long_int) / 3 / 366 / 24 / 60 / 60); +verify (INT_MAX <= TYPE_MAXIMUM (long_int) / 4 / 366 / 24 / 60 / 60); /* Shift A right by B bits portably, by dividing A by 2**B and truncating towards minus infinity. B should be in the range 0 <= B @@ -110,16 +144,15 @@ shr (long_int a, int b) : a / (one << b) - (a % (one << b) < 0)); } -/* Bounds for the intersection of time_t and long_int. */ +/* Bounds for the intersection of __time64_t and long_int. */ static long_int const mktime_min - = ((TYPE_SIGNED (time_t) && TYPE_MINIMUM (time_t) < TYPE_MINIMUM (long_int)) - ? TYPE_MINIMUM (long_int) : TYPE_MINIMUM (time_t)); + = ((TYPE_SIGNED (__time64_t) + && TYPE_MINIMUM (__time64_t) < TYPE_MINIMUM (long_int)) + ? TYPE_MINIMUM (long_int) : TYPE_MINIMUM (__time64_t)); static long_int const mktime_max - = (TYPE_MAXIMUM (long_int) < TYPE_MAXIMUM (time_t) - ? TYPE_MAXIMUM (long_int) : TYPE_MAXIMUM (time_t)); - -verify (TYPE_IS_INTEGER (time_t)); + = (TYPE_MAXIMUM (long_int) < TYPE_MAXIMUM (__time64_t) + ? TYPE_MAXIMUM (long_int) : TYPE_MAXIMUM (__time64_t)); #define EPOCH_YEAR 1970 #define TM_YEAR_BASE 1900 @@ -150,19 +183,6 @@ const unsigned short int __mon_yday[2][13] = }; -#ifdef _LIBC -typedef time_t mktime_offset_t; -#else -/* Portable standalone applications should supply a <time.h> that - declares a POSIX-compliant localtime_r, for the benefit of older - implementations that lack localtime_r or have a nonstandard one. - See the gnulib time_r module for one way to implement this. */ -# undef __localtime_r -# define __localtime_r localtime_r -# define __mktime_internal mktime_internal -# include "mktime-internal.h" -#endif - /* Do the values A and B differ according to the rules for tm_isdst? A and B differ if one is zero and the other positive. */ static bool @@ -176,9 +196,10 @@ isdst_differ (int a, int b) were not adjusted between the timestamps. The YEAR values uses the same numbering as TP->tm_year. Values - need not be in the usual range. However, YEAR1 must not overflow - when multiplied by three times the number of seconds in a year, and - likewise for YDAY1 and three times the number of seconds in a day. */ + need not be in the usual range. However, YEAR1 - YEAR0 must not + overflow even when multiplied by three times the number of seconds + in a year, and likewise for YDAY1 - YDAY0 and three times the + number of seconds in a day. */ static long_int ydhms_diff (long_int year1, long_int yday1, int hour1, int min1, int sec1, @@ -213,109 +234,94 @@ long_int_avg (long_int a, long_int b) return shr (a, 1) + shr (b, 1) + ((a | b) & 1); } -/* Return a time_t value corresponding to (YEAR-YDAY HOUR:MIN:SEC), - assuming that T corresponds to *TP and that no clock adjustments - occurred between *TP and the desired time. - Although T and the returned value are of type long_int, - they represent time_t values and must be in time_t range. - If TP is null, return a value not equal to T; this avoids false matches. +/* Return a long_int value corresponding to (YEAR-YDAY HOUR:MIN:SEC) + minus *TP seconds, assuming no clock adjustments occurred between + the two timestamps. + YEAR and YDAY must not be so large that multiplying them by three times the number of seconds in a year (or day, respectively) would overflow long_int. - If the returned value would be out of range, yield the minimal or - maximal in-range value, except do not yield a value equal to T. */ + *TP should be in the usual range. */ static long_int -guess_time_tm (long_int year, long_int yday, int hour, int min, int sec, - long_int t, const struct tm *tp) +tm_diff (long_int year, long_int yday, int hour, int min, int sec, + struct tm const *tp) { - if (tp) - { - long_int result; - long_int d = ydhms_diff (year, yday, hour, min, sec, - tp->tm_year, tp->tm_yday, - tp->tm_hour, tp->tm_min, tp->tm_sec); - if (! INT_ADD_WRAPV (t, d, &result)) - return result; - } - - /* Overflow occurred one way or another. Return the nearest result - that is actually in range, except don't report a zero difference - if the actual difference is nonzero, as that would cause a false - match; and don't oscillate between two values, as that would - confuse the spring-forward gap detector. */ - return (t < long_int_avg (mktime_min, mktime_max) - ? (t <= mktime_min + 1 ? t + 1 : mktime_min) - : (mktime_max - 1 <= t ? t - 1 : mktime_max)); + return ydhms_diff (year, yday, hour, min, sec, + tp->tm_year, tp->tm_yday, + tp->tm_hour, tp->tm_min, tp->tm_sec); } /* Use CONVERT to convert T to a struct tm value in *TM. T must be in - range for time_t. Return TM if successful, NULL if T is out of - range for CONVERT. */ + range for __time64_t. Return TM if successful, NULL (setting errno) on + failure. */ static struct tm * -convert_time (struct tm *(*convert) (const time_t *, struct tm *), +convert_time (struct tm *(*convert) (const __time64_t *, struct tm *), long_int t, struct tm *tm) { - time_t x = t; + __time64_t x = t; return convert (&x, tm); } /* Use CONVERT to convert *T to a broken down time in *TP. If *T is out of range for conversion, adjust it so that it is the nearest in-range value and then convert that. - A value is in range if it fits in both time_t and long_int. */ + A value is in range if it fits in both __time64_t and long_int. + Return TP on success, NULL (setting errno) on failure. */ static struct tm * -ranged_convert (struct tm *(*convert) (const time_t *, struct tm *), +ranged_convert (struct tm *(*convert) (const __time64_t *, struct tm *), long_int *t, struct tm *tp) { - struct tm *r; - if (*t < mktime_min) - *t = mktime_min; - else if (mktime_max < *t) - *t = mktime_max; - r = convert_time (convert, *t, tp); - - if (!r && *t) + long_int t1 = (*t < mktime_min ? mktime_min + : *t <= mktime_max ? *t : mktime_max); + struct tm *r = convert_time (convert, t1, tp); + if (r) { - long_int bad = *t; - long_int ok = 0; + *t = t1; + return r; + } + if (errno != EOVERFLOW) + return NULL; - /* BAD is a known unconvertible value, and OK is a known good one. - Use binary search to narrow the range between BAD and OK until - they differ by 1. */ - while (true) - { - long_int mid = long_int_avg (ok, bad); - if (mid != ok && mid != bad) - break; - r = convert_time (convert, mid, tp); - if (r) - ok = mid; - else - bad = mid; - } + long_int bad = t1; + long_int ok = 0; + struct tm oktm; oktm.tm_sec = -1; - if (!r && ok) - { - /* The last conversion attempt failed; - revert to the most recent successful attempt. */ - r = convert_time (convert, ok, tp); - } + /* BAD is a known out-of-range value, and OK is a known in-range one. + Use binary search to narrow the range between BAD and OK until + they differ by 1. */ + while (true) + { + long_int mid = long_int_avg (ok, bad); + if (mid == ok || mid == bad) + break; + if (convert_time (convert, mid, tp)) + ok = mid, oktm = *tp; + else if (errno != EOVERFLOW) + return NULL; + else + bad = mid; } - return r; + if (oktm.tm_sec < 0) + return NULL; + *t = ok; + *tp = oktm; + return tp; } -/* Convert *TP to a time_t value, inverting + +/* Convert *TP to a __time64_t value, inverting the monotonic and mostly-unit-linear conversion function CONVERT. Use *OFFSET to keep track of a guess at the offset of the result, compared to what the result would be for UTC without leap seconds. If *OFFSET's guess is correct, only one CONVERT call is needed. + If successful, set *TP to the canonicalized struct tm; + otherwise leave *TP alone, return ((time_t) -1) and set errno. This function is external because it is used also by timegm.c. */ -time_t +__time64_t __mktime_internal (struct tm *tp, - struct tm *(*convert) (const time_t *, struct tm *), + struct tm *(*convert) (const __time64_t *, struct tm *), mktime_offset_t *offset) { - long_int t, gt, t0, t1, t2, dt; struct tm tm; /* The maximum number of probes (calls to CONVERT) should be enough @@ -335,7 +341,7 @@ __mktime_internal (struct tm *tp, int isdst = tp->tm_isdst; /* 1 if the previous probe was DST. */ - int dst2; + int dst2 = 0; /* Ensure that mon is in range, and set year accordingly. */ int mon_remainder = mon % 12; @@ -355,6 +361,7 @@ __mktime_internal (struct tm *tp, long_int lmday = mday; long_int yday = mon_yday + lmday; + mktime_offset_t off = *offset; int negative_offset_guess; int sec_requested = sec; @@ -362,7 +369,7 @@ __mktime_internal (struct tm *tp, if (LEAP_SECONDS_POSSIBLE) { /* Handle out-of-range seconds specially, - since ydhms_tm_diff assumes every minute has 60 seconds. */ + since ydhms_diff assumes every minute has 60 seconds. */ if (sec < 0) sec = 0; if (59 < sec) @@ -372,34 +379,47 @@ __mktime_internal (struct tm *tp, /* Invert CONVERT by probing. First assume the same offset as last time. */ - INT_SUBTRACT_WRAPV (0, *offset, &negative_offset_guess); - t0 = ydhms_diff (year, yday, hour, min, sec, - EPOCH_YEAR - TM_YEAR_BASE, 0, 0, 0, negative_offset_guess); + INT_SUBTRACT_WRAPV (0, off, &negative_offset_guess); + long_int t0 = ydhms_diff (year, yday, hour, min, sec, + EPOCH_YEAR - TM_YEAR_BASE, 0, 0, 0, + negative_offset_guess); + long_int t = t0, t1 = t0, t2 = t0; /* Repeatedly use the error to improve the guess. */ - for (t = t1 = t2 = t0, dst2 = 0; - (gt = guess_time_tm (year, yday, hour, min, sec, t, - ranged_convert (convert, &t, &tm)), - t != gt); - t1 = t2, t2 = t, t = gt, dst2 = tm.tm_isdst != 0) - if (t == t1 && t != t2 - && (tm.tm_isdst < 0 - || (isdst < 0 - ? dst2 <= (tm.tm_isdst != 0) - : (isdst != 0) != (tm.tm_isdst != 0)))) - /* We can't possibly find a match, as we are oscillating - between two values. The requested time probably falls - within a spring-forward gap of size GT - T. Follow the common - practice in this case, which is to return a time that is GT - T - away from the requested time, preferring a time whose - tm_isdst differs from the requested value. (If no tm_isdst - was requested and only one of the two values has a nonzero - tm_isdst, prefer that value.) In practice, this is more - useful than returning -1. */ - goto offset_found; - else if (--remaining_probes == 0) - return -1; + while (true) + { + if (! ranged_convert (convert, &t, &tm)) + return -1; + long_int dt = tm_diff (year, yday, hour, min, sec, &tm); + if (dt == 0) + break; + + if (t == t1 && t != t2 + && (tm.tm_isdst < 0 + || (isdst < 0 + ? dst2 <= (tm.tm_isdst != 0) + : (isdst != 0) != (tm.tm_isdst != 0)))) + /* We can't possibly find a match, as we are oscillating + between two values. The requested time probably falls + within a spring-forward gap of size DT. Follow the common + practice in this case, which is to return a time that is DT + away from the requested time, preferring a time whose + tm_isdst differs from the requested value. (If no tm_isdst + was requested and only one of the two values has a nonzero + tm_isdst, prefer that value.) In practice, this is more + useful than returning -1. */ + goto offset_found; + + remaining_probes--; + if (remaining_probes == 0) + { + __set_errno (EOVERFLOW); + return -1; + } + + t1 = t2, t2 = t, t += dt, dst2 = tm.tm_isdst != 0; + } /* We have a match. Check whether tm.tm_isdst has the requested value, if any. */ @@ -441,25 +461,38 @@ __mktime_internal (struct tm *tp, if (! INT_ADD_WRAPV (t, delta * direction, &ot)) { struct tm otm; - ranged_convert (convert, &ot, &otm); + if (! ranged_convert (convert, &ot, &otm)) + return -1; if (! isdst_differ (isdst, otm.tm_isdst)) { /* We found the desired tm_isdst. Extrapolate back to the desired time. */ - t = guess_time_tm (year, yday, hour, min, sec, ot, &otm); - ranged_convert (convert, &t, &tm); - goto offset_found; + long_int gt = ot + tm_diff (year, yday, hour, min, sec, + &otm); + if (mktime_min <= gt && gt <= mktime_max) + { + if (convert_time (convert, gt, &tm)) + { + t = gt; + goto offset_found; + } + if (errno != EOVERFLOW) + return -1; + } } } } + + __set_errno (EOVERFLOW); + return -1; } offset_found: /* Set *OFFSET to the low-order bits of T - T0 - NEGATIVE_OFFSET_GUESS. This is just a heuristic to speed up the next mktime call, and correctness is unaffected if integer overflow occurs here. */ - INT_SUBTRACT_WRAPV (t, t0, &dt); - INT_SUBTRACT_WRAPV (dt, negative_offset_guess, offset); + INT_SUBTRACT_WRAPV (t, t0, offset); + INT_SUBTRACT_WRAPV (*offset, negative_offset_guess, offset); if (LEAP_SECONDS_POSSIBLE && sec_requested != tm.tm_sec) { @@ -469,8 +502,12 @@ __mktime_internal (struct tm *tp, sec_adjustment -= sec; sec_adjustment += sec_requested; if (INT_ADD_WRAPV (t, sec_adjustment, &t) - || ! (mktime_min <= t && t <= mktime_max) - || ! convert_time (convert, t, &tm)) + || ! (mktime_min <= t && t <= mktime_max)) + { + __set_errno (EOVERFLOW); + return -1; + } + if (! convert_time (convert, t, &tm)) return -1; } @@ -478,213 +515,52 @@ __mktime_internal (struct tm *tp, return t; } -#endif /* NEED_MKTIME_WORKING || NEED_MKTIME_INTERNAL || DEBUG_MKTIME */ +#endif /* _LIBC || NEED_MKTIME_WORKING || NEED_MKTIME_INTERNAL */ -#if NEED_MKTIME_WORKING || NEED_MKTIME_WINDOWS || DEBUG_MKTIME +#if defined _LIBC || NEED_MKTIME_WORKING || NEED_MKTIME_WINDOWS -# if NEED_MKTIME_WORKING || DEBUG_MKTIME -static mktime_offset_t localtime_offset; -# endif - -/* Convert *TP to a time_t value. */ -time_t -mktime (struct tm *tp) +/* Convert *TP to a __time64_t value. */ +__time64_t +__mktime64 (struct tm *tp) { -# if NEED_MKTIME_WINDOWS - /* Rectify the value of the environment variable TZ. - There are four possible kinds of such values: - - Traditional US time zone names, e.g. "PST8PDT". Syntax: see - <https://msdn.microsoft.com/en-us/library/90s5c885.aspx> - - Time zone names based on geography, that contain one or more - slashes, e.g. "Europe/Moscow". - - Time zone names based on geography, without slashes, e.g. - "Singapore". - - Time zone names that contain explicit DST rules. Syntax: see - <http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap08.html#tag_08_03> - The Microsoft CRT understands only the first kind. It produces incorrect - results if the value of TZ is of the other kinds. - But in a Cygwin environment, /etc/profile.d/tzset.sh sets TZ to a value - of the second kind for most geographies, or of the first kind in a few - other geographies. If it is of the second kind, neutralize it. For the - Microsoft CRT, an absent or empty TZ means the time zone that the user - has set in the Windows Control Panel. - If the value of TZ is of the third or fourth kind -- Cygwin programs - understand these syntaxes as well --, it does not matter whether we - neutralize it or not, since these values occur only when a Cygwin user - has set TZ explicitly; this case is 1. rare and 2. under the user's - responsibility. */ - const char *tz = getenv ("TZ"); - if (tz != NULL && strchr (tz, '/') != NULL) - _putenv ("TZ="); -# endif - -# if NEED_MKTIME_WORKING || DEBUG_MKTIME -# ifdef _LIBC /* POSIX.1 8.1.1 requires that whenever mktime() is called, the time zone names contained in the external variable 'tzname' shall be set as if the tzset() function had been called. */ __tzset (); -# elif HAVE_TZSET - tzset (); -# endif - return __mktime_internal (tp, __localtime_r, &localtime_offset); +# if defined _LIBC || NEED_MKTIME_WORKING + static mktime_offset_t localtime_offset; + return __mktime_internal (tp, __localtime64_r, &localtime_offset); # else # undef mktime return mktime (tp); # endif } +#endif /* _LIBC || NEED_MKTIME_WORKING || NEED_MKTIME_WINDOWS */ -#endif /* NEED_MKTIME_WORKING || NEED_MKTIME_WINDOWS || DEBUG_MKTIME */ - -#ifdef weak_alias -weak_alias (mktime, timelocal) -#endif - -#ifdef _LIBC -libc_hidden_def (mktime) -libc_hidden_weak (timelocal) -#endif - -#if DEBUG_MKTIME +#if defined _LIBC && __TIMESIZE != 64 -static int -not_equal_tm (const struct tm *a, const struct tm *b) -{ - return ((a->tm_sec ^ b->tm_sec) - | (a->tm_min ^ b->tm_min) - | (a->tm_hour ^ b->tm_hour) - | (a->tm_mday ^ b->tm_mday) - | (a->tm_mon ^ b->tm_mon) - | (a->tm_year ^ b->tm_year) - | (a->tm_yday ^ b->tm_yday) - | isdst_differ (a->tm_isdst, b->tm_isdst)); -} +libc_hidden_def (__mktime64) -static void -print_tm (const struct tm *tp) -{ - if (tp) - printf ("%04d-%02d-%02d %02d:%02d:%02d yday %03d wday %d isdst %d", - tp->tm_year + TM_YEAR_BASE, tp->tm_mon + 1, tp->tm_mday, - tp->tm_hour, tp->tm_min, tp->tm_sec, - tp->tm_yday, tp->tm_wday, tp->tm_isdst); - else - printf ("0"); -} - -static int -check_result (time_t tk, struct tm tmk, time_t tl, const struct tm *lt) -{ - if (tk != tl || !lt || not_equal_tm (&tmk, lt)) - { - printf ("mktime ("); - print_tm (lt); - printf (")\nyields ("); - print_tm (&tmk); - printf (") == %ld, should be %ld\n", (long int) tk, (long int) tl); - return 1; - } - - return 0; -} - -int -main (int argc, char **argv) +time_t +mktime (struct tm *tp) { - int status = 0; - struct tm tm, tmk, tml; - struct tm *lt; - time_t tk, tl, tl1; - char trailer; - - /* Sanity check, plus call tzset. */ - tl = 0; - if (! localtime (&tl)) + struct tm tm = *tp; + __time64_t t = __mktime64 (&tm); + if (in_time_t_range (t)) { - printf ("localtime (0) fails\n"); - status = 1; + *tp = tm; + return t; } - - if ((argc == 3 || argc == 4) - && (sscanf (argv[1], "%d-%d-%d%c", - &tm.tm_year, &tm.tm_mon, &tm.tm_mday, &trailer) - == 3) - && (sscanf (argv[2], "%d:%d:%d%c", - &tm.tm_hour, &tm.tm_min, &tm.tm_sec, &trailer) - == 3)) - { - tm.tm_year -= TM_YEAR_BASE; - tm.tm_mon--; - tm.tm_isdst = argc == 3 ? -1 : atoi (argv[3]); - tmk = tm; - tl = mktime (&tmk); - lt = localtime_r (&tl, &tml); - printf ("mktime returns %ld == ", (long int) tl); - print_tm (&tmk); - printf ("\n"); - status = check_result (tl, tmk, tl, lt); - } - else if (argc == 4 || (argc == 5 && strcmp (argv[4], "-") == 0)) + else { - time_t from = atol (argv[1]); - time_t by = atol (argv[2]); - time_t to = atol (argv[3]); - - if (argc == 4) - for (tl = from; by < 0 ? to <= tl : tl <= to; tl = tl1) - { - lt = localtime_r (&tl, &tml); - if (lt) - { - tmk = tml; - tk = mktime (&tmk); - status |= check_result (tk, tmk, tl, &tml); - } - else - { - printf ("localtime_r (%ld) yields 0\n", (long int) tl); - status = 1; - } - tl1 = tl + by; - if ((tl1 < tl) != (by < 0)) - break; - } - else - for (tl = from; by < 0 ? to <= tl : tl <= to; tl = tl1) - { - /* Null benchmark. */ - lt = localtime_r (&tl, &tml); - if (lt) - { - tmk = tml; - tk = tl; - status |= check_result (tk, tmk, tl, &tml); - } - else - { - printf ("localtime_r (%ld) yields 0\n", (long int) tl); - status = 1; - } - tl1 = tl + by; - if ((tl1 < tl) != (by < 0)) - break; - } + __set_errno (EOVERFLOW); + return -1; } - else - printf ("Usage:\ -\t%s YYYY-MM-DD HH:MM:SS [ISDST] # Test given time.\n\ -\t%s FROM BY TO # Test values FROM, FROM+BY, ..., TO.\n\ -\t%s FROM BY TO - # Do not test those values (for benchmark).\n", - argv[0], argv[0], argv[0]); - - return status; } -#endif /* DEBUG_MKTIME */ - -/* -Local Variables: -compile-command: "gcc -DDEBUG_MKTIME -I. -Wall -W -O2 -g mktime.c -o mktime" -End: -*/ +#endif + +weak_alias (mktime, timelocal) +libc_hidden_def (mktime) +libc_hidden_weak (timelocal) diff --git a/lib/nstrftime.c b/lib/nstrftime.c index 7df3869bbaf..bc84da5a0cb 100644 --- a/lib/nstrftime.c +++ b/lib/nstrftime.c @@ -91,6 +91,7 @@ extern char *tzname[]; # define UCHAR_T unsigned char # define L_(Str) Str # define NLW(Sym) Sym +# define ABALTMON_1 _NL_ABALTMON_1 # define MEMCPY(d, s, n) memcpy (d, s, n) # define STRLEN(s) strlen (s) @@ -179,7 +180,7 @@ extern char *tzname[]; if (digits == 0 && _n < _w) \ { \ size_t _delta = width - _n; \ - if (pad == L_('0')) \ + if (pad == L_('0') || pad == L_('+')) \ memset_zero (p, _delta); \ else \ memset_space (p, _delta); \ @@ -255,7 +256,7 @@ extern char *tzname[]; # undef _NL_CURRENT # define _NL_CURRENT(category, item) \ (current->values[_NL_ITEM_INDEX (item)].string) -# define LOCALE_PARAM , __locale_t loc +# define LOCALE_PARAM , locale_t loc # define LOCALE_ARG , loc # define HELPER_LOCALE_ARG , current #else @@ -417,7 +418,7 @@ iso_week_days (int yday, int wday) static size_t __strftime_internal (STREAM_OR_CHAR_T *, STRFTIME_ARG (size_t) const CHAR_T *, const struct tm *, - bool, bool * + bool, int, int, bool * extra_args_spec LOCALE_PARAM); /* Write information from TP into S according to the format @@ -432,8 +433,8 @@ my_strftime (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) const struct tm *tp extra_args_spec LOCALE_PARAM) { bool tzset_called = false; - return __strftime_internal (s, STRFTIME_ARG (maxsize) format, tp, - false, &tzset_called extra_args LOCALE_ARG); + return __strftime_internal (s, STRFTIME_ARG (maxsize) format, tp, false, + 0, -1, &tzset_called extra_args LOCALE_ARG); } #if defined _LIBC && ! FPRINTFTIME libc_hidden_def (my_strftime) @@ -445,7 +446,8 @@ libc_hidden_def (my_strftime) static size_t __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) const CHAR_T *format, - const struct tm *tp, bool upcase, bool *tzset_called + const struct tm *tp, bool upcase, + int yr_spec, int width, bool *tzset_called extra_args_spec LOCALE_PARAM) { #if defined _LIBC && defined USE_IN_EXTENDED_LOCALE_MODEL @@ -475,12 +477,19 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) # define f_month \ ((const CHAR_T *) (tp->tm_mon < 0 || tp->tm_mon > 11 \ ? "?" : _NL_CURRENT (LC_TIME, NLW(MON_1) + tp->tm_mon))) +# define a_altmonth \ + ((const CHAR_T *) (tp->tm_mon < 0 || tp->tm_mon > 11 \ + ? "?" : _NL_CURRENT (LC_TIME, NLW(ABALTMON_1) + tp->tm_mon))) +# define f_altmonth \ + ((const CHAR_T *) (tp->tm_mon < 0 || tp->tm_mon > 11 \ + ? "?" : _NL_CURRENT (LC_TIME, NLW(ALTMON_1) + tp->tm_mon))) # define ampm \ ((const CHAR_T *) _NL_CURRENT (LC_TIME, tp->tm_hour > 11 \ ? NLW(PM_STR) : NLW(AM_STR))) # define aw_len STRLEN (a_wkday) # define am_len STRLEN (a_month) +# define aam_len STRLEN (a_altmonth) # define ap_len STRLEN (ampm) #endif #if HAVE_TZNAME @@ -550,7 +559,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) if (hour12 == 0) hour12 = 12; - for (f = format; *f != '\0'; ++f) + for (f = format; *f != '\0'; width = -1, f++) { int pad = 0; /* Padding for number ('-', '_', or 0). */ int modifier; /* Field modifier ('E', 'O', or 0). */ @@ -568,12 +577,12 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) + (sizeof (int) < sizeof (time_t) ? INT_STRLEN_BOUND (time_t) : INT_STRLEN_BOUND (int))]; - int width = -1; bool to_lowcase = false; bool to_uppcase = upcase; size_t colons; bool change_case = false; int format_char; + int subwidth; #if DO_MULTIBYTE && !defined COMPILE_WIDE switch (*f) @@ -671,6 +680,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) /* This influences the number formats. */ case L_('_'): case L_('-'): + case L_('+'): case L_('0'): pad = *f; continue; @@ -689,7 +699,6 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) break; } - /* As a GNU extension we allow the field width to be specified. */ if (ISDIGIT (*f)) { width = 0; @@ -735,12 +744,16 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) } \ while (0) #define DO_SIGNED_NUMBER(d, negative, v) \ + DO_MAYBE_SIGNED_NUMBER (d, negative, v, do_signed_number) +#define DO_YEARISH(d, negative, v) \ + DO_MAYBE_SIGNED_NUMBER (d, negative, v, do_yearish) +#define DO_MAYBE_SIGNED_NUMBER(d, negative, v, label) \ do \ { \ digits = d; \ negative_number = negative; \ u_number_value = v; \ - goto do_signed_number; \ + goto label; \ } \ while (0) @@ -808,17 +821,20 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) to_uppcase = true; to_lowcase = false; } - if (modifier != 0) + if (modifier == L_('E')) goto bad_format; #ifdef _NL_CURRENT - cpy (am_len, a_month); + if (modifier == L_('O')) + cpy (aam_len, a_altmonth); + else + cpy (am_len, a_month); break; #else goto underlying_strftime; #endif case L_('B'): - if (modifier != 0) + if (modifier == L_('E')) goto bad_format; if (change_case) { @@ -826,7 +842,10 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) to_lowcase = false; } #ifdef _NL_CURRENT - cpy (STRLEN (f_month), f_month); + if (modifier == L_('O')) + cpy (STRLEN (f_altmonth), f_altmonth); + else + cpy (STRLEN (f_month), f_month); break; #else goto underlying_strftime; @@ -836,7 +855,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) if (modifier == L_('O')) goto bad_format; #ifdef _NL_CURRENT - if (! (modifier == 'E' + if (! (modifier == L_('E') && (*(subfmt = (const CHAR_T *) _NL_CURRENT (LC_TIME, NLW(ERA_D_T_FMT))) @@ -847,15 +866,17 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) #endif subformat: + subwidth = -1; + subformat_width: { size_t len = __strftime_internal (NULL, STRFTIME_ARG ((size_t) -1) - subfmt, - tp, to_uppcase, tzset_called + subfmt, tp, to_uppcase, + pad, subwidth, tzset_called extra_args LOCALE_ARG); add (len, __strftime_internal (p, STRFTIME_ARG (maxsize - i) - subfmt, - tp, to_uppcase, tzset_called + subfmt, tp, to_uppcase, + pad, subwidth, tzset_called extra_args LOCALE_ARG)); } break; @@ -916,7 +937,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) { int century = tp->tm_year / 100 + TM_YEAR_BASE / 100; century -= tp->tm_year % 100 < 0 && 0 < century; - DO_SIGNED_NUMBER (2, tp->tm_year < - TM_YEAR_BASE, century); + DO_YEARISH (2, tp->tm_year < - TM_YEAR_BASE, century); } case L_('x'): @@ -925,7 +946,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) #ifdef _NL_CURRENT if (! (modifier == L_('E') && (*(subfmt = - (const CHAR_T *)_NL_CURRENT (LC_TIME, NLW(ERA_D_FMT))) + (const CHAR_T *) _NL_CURRENT (LC_TIME, NLW(ERA_D_FMT))) != L_('\0')))) subfmt = (const CHAR_T *) _NL_CURRENT (LC_TIME, NLW(D_FMT)); goto subformat; @@ -957,9 +978,17 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) always_output_a_sign = true; goto do_number_body; + do_yearish: + if (pad == 0) + pad = yr_spec; + always_output_a_sign + = (pad == L_('+') + && ((digits == 2 ? 99 : 9999) < u_number_value + || digits < width)); + goto do_maybe_signed_number; + do_number_spacepad: - /* Force '_' flag unless overridden by '0' or '-' flag. */ - if (pad != L_('0') && pad != L_('-')) + if (pad == 0) pad = L_('_'); do_number: @@ -969,6 +998,8 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) do_signed_number: always_output_a_sign = false; + + do_maybe_signed_number: tz_colon_mask = 0; do_number_body: @@ -1073,8 +1104,19 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) case L_('F'): if (modifier != 0) goto bad_format; + if (pad == 0 && width < 0) + { + pad = L_('+'); + subwidth = 4; + } + else + { + subwidth = width - 6; + if (subwidth < 0) + subwidth = 0; + } subfmt = L_("%Y-%m-%d"); - goto subformat; + goto subformat_width; case L_('H'): if (modifier == L_('E')) @@ -1283,17 +1325,18 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) case L_('g'): { int yy = (tp->tm_year % 100 + year_adjust) % 100; - DO_NUMBER (2, (0 <= yy - ? yy - : tp->tm_year < -TM_YEAR_BASE - year_adjust - ? -yy - : yy + 100)); + DO_YEARISH (2, false, + (0 <= yy + ? yy + : tp->tm_year < -TM_YEAR_BASE - year_adjust + ? -yy + : yy + 100)); } case L_('G'): - DO_SIGNED_NUMBER (4, tp->tm_year < -TM_YEAR_BASE - year_adjust, - (tp->tm_year + (unsigned int) TM_YEAR_BASE - + year_adjust)); + DO_YEARISH (4, tp->tm_year < -TM_YEAR_BASE - year_adjust, + (tp->tm_year + (unsigned int) TM_YEAR_BASE + + year_adjust)); default: DO_NUMBER (2, days / 7 + 1); @@ -1313,7 +1356,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) DO_NUMBER (1, tp->tm_wday); case L_('Y'): - if (modifier == 'E') + if (modifier == L_('E')) { #if HAVE_STRUCT_ERA_ENTRY struct era_entry *era = _nl_get_era_entry (tp HELPER_LOCALE_ARG); @@ -1324,6 +1367,8 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) # else subfmt = era->era_format; # endif + if (pad == 0) + pad = yr_spec; goto subformat; } #else @@ -1333,8 +1378,8 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) if (modifier == L_('O')) goto bad_format; - DO_SIGNED_NUMBER (4, tp->tm_year < -TM_YEAR_BASE, - tp->tm_year + (unsigned int) TM_YEAR_BASE); + DO_YEARISH (4, tp->tm_year < -TM_YEAR_BASE, + tp->tm_year + (unsigned int) TM_YEAR_BASE); case L_('y'): if (modifier == L_('E')) @@ -1344,7 +1389,9 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) if (era) { int delta = tp->tm_year - era->start_date[0]; - DO_NUMBER (1, (era->offset + if (pad == 0) + pad = yr_spec; + DO_NUMBER (2, (era->offset + delta * era->absolute_direction)); } #else @@ -1356,7 +1403,7 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) int yy = tp->tm_year % 100; if (yy < 0) yy = tp->tm_year < - TM_YEAR_BASE ? -yy : yy + 100; - DO_NUMBER (2, yy); + DO_YEARISH (2, false, yy); } case L_('Z'): @@ -1424,28 +1471,10 @@ __strftime_internal (STREAM_OR_CHAR_T *s, STRFTIME_ARG (size_t maxsize) # endif ltm = *tp; + ltm.tm_wday = -1; lt = mktime_z (tz, <m); - - if (lt == (time_t) -1) - { - /* mktime returns -1 for errors, but -1 is also a - valid time_t value. Check whether an error really - occurred. */ - struct tm tm; - - if (! localtime_rz (tz, <, &tm) - || ((ltm.tm_sec ^ tm.tm_sec) - | (ltm.tm_min ^ tm.tm_min) - | (ltm.tm_hour ^ tm.tm_hour) - | (ltm.tm_mday ^ tm.tm_mday) - | (ltm.tm_mon ^ tm.tm_mon) - | (ltm.tm_year ^ tm.tm_year))) - break; - } - - if (! localtime_rz (0, <, >m)) + if (ltm.tm_wday < 0 || ! localtime_rz (0, <, >m)) break; - diff = tm_diff (<m, >m); } #endif diff --git a/lib/open.c b/lib/open.c index ba223bff1be..655260572d4 100644 --- a/lib/open.c +++ b/lib/open.c @@ -86,7 +86,7 @@ open (const char *filename, int flags, ...) flags &= ~O_NONBLOCK; #endif -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ if (strcmp (filename, "/dev/null") == 0) filename = "NUL"; #endif diff --git a/lib/pipe2.c b/lib/pipe2.c index c5747cd1874..15a5dec9852 100644 --- a/lib/pipe2.c +++ b/lib/pipe2.c @@ -29,7 +29,7 @@ # include "nonblocking.h" #endif -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ /* Native Windows API. */ # include <io.h> @@ -73,7 +73,7 @@ pipe2 (int fd[2], int flags) return -1; } -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ /* Native Windows API. */ if (_pipe (fd, 4096, flags & ~O_NONBLOCK) < 0) @@ -152,8 +152,7 @@ pipe2 (int fd[2], int flags) #endif -#if GNULIB_defined_O_NONBLOCK || \ - !((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) +#if GNULIB_defined_O_NONBLOCK || !(defined _WIN32 && ! defined __CYGWIN__) fail: { int saved_errno = errno; diff --git a/lib/pselect.c b/lib/pselect.c index dacb5332ec6..ae010aa02e4 100644 --- a/lib/pselect.c +++ b/lib/pselect.c @@ -83,9 +83,9 @@ pselect (int nfds, fd_set *restrict rfds, int rpl_pselect (int nfds, fd_set *restrict rfds, - fd_set *restrict wfds, fd_set *restrict xfds, + fd_set *restrict wfds, fd_set *restrict xfds, struct timespec const *restrict timeout, - sigset_t const *restrict sigmask) + sigset_t const *restrict sigmask) { int i; diff --git a/lib/putenv.c b/lib/putenv.c index 81a651a7549..81085e9d80a 100644 --- a/lib/putenv.c +++ b/lib/putenv.c @@ -34,7 +34,7 @@ #include <string.h> #include <unistd.h> -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#if defined _WIN32 && ! defined __CYGWIN__ # define WIN32_LEAN_AND_MEAN # include <windows.h> #endif @@ -153,7 +153,7 @@ putenv (char *string) *ep = string; break; } -# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# if defined _WIN32 && ! defined __CYGWIN__ if (putenv_result == 0) { /* _putenv propagated "NAME= " into the subprocess environment; diff --git a/lib/regcomp.c b/lib/regcomp.c new file mode 100644 index 00000000000..892139a02af --- /dev/null +++ b/lib/regcomp.c @@ -0,0 +1,3934 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#ifdef _LIBC +# include <locale/weight.h> +#endif + +static reg_errcode_t re_compile_internal (regex_t *preg, const char * pattern, + size_t length, reg_syntax_t syntax); +static void re_compile_fastmap_iter (regex_t *bufp, + const re_dfastate_t *init_state, + char *fastmap); +static reg_errcode_t init_dfa (re_dfa_t *dfa, size_t pat_len); +#ifdef RE_ENABLE_I18N +static void free_charset (re_charset_t *cset); +#endif /* RE_ENABLE_I18N */ +static void free_workarea_compile (regex_t *preg); +static reg_errcode_t create_initial_state (re_dfa_t *dfa); +#ifdef RE_ENABLE_I18N +static void optimize_utf8 (re_dfa_t *dfa); +#endif +static reg_errcode_t analyze (regex_t *preg); +static reg_errcode_t preorder (bin_tree_t *root, + reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra); +static reg_errcode_t postorder (bin_tree_t *root, + reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra); +static reg_errcode_t optimize_subexps (void *extra, bin_tree_t *node); +static reg_errcode_t lower_subexps (void *extra, bin_tree_t *node); +static bin_tree_t *lower_subexp (reg_errcode_t *err, regex_t *preg, + bin_tree_t *node); +static reg_errcode_t calc_first (void *extra, bin_tree_t *node); +static reg_errcode_t calc_next (void *extra, bin_tree_t *node); +static reg_errcode_t link_nfa_nodes (void *extra, bin_tree_t *node); +static Idx duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint); +static Idx search_duplicated_node (const re_dfa_t *dfa, Idx org_node, + unsigned int constraint); +static reg_errcode_t calc_eclosure (re_dfa_t *dfa); +static reg_errcode_t calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, + Idx node, bool root); +static reg_errcode_t calc_inveclosure (re_dfa_t *dfa); +static Idx fetch_number (re_string_t *input, re_token_t *token, + reg_syntax_t syntax); +static int peek_token (re_token_t *token, re_string_t *input, + reg_syntax_t syntax); +static bin_tree_t *parse (re_string_t *regexp, regex_t *preg, + reg_syntax_t syntax, reg_errcode_t *err); +static bin_tree_t *parse_reg_exp (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_branch (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_expression (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_sub_exp (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_dup_op (bin_tree_t *dup_elem, re_string_t *regexp, + re_dfa_t *dfa, re_token_t *token, + reg_syntax_t syntax, reg_errcode_t *err); +static bin_tree_t *parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, + re_token_t *token, reg_syntax_t syntax, + reg_errcode_t *err); +static reg_errcode_t parse_bracket_element (bracket_elem_t *elem, + re_string_t *regexp, + re_token_t *token, int token_len, + re_dfa_t *dfa, + reg_syntax_t syntax, + bool accept_hyphen); +static reg_errcode_t parse_bracket_symbol (bracket_elem_t *elem, + re_string_t *regexp, + re_token_t *token); +#ifdef RE_ENABLE_I18N +static reg_errcode_t build_equiv_class (bitset_t sbcset, + re_charset_t *mbcset, + Idx *equiv_class_alloc, + const unsigned char *name); +static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, + bitset_t sbcset, + re_charset_t *mbcset, + Idx *char_class_alloc, + const char *class_name, + reg_syntax_t syntax); +#else /* not RE_ENABLE_I18N */ +static reg_errcode_t build_equiv_class (bitset_t sbcset, + const unsigned char *name); +static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, + bitset_t sbcset, + const char *class_name, + reg_syntax_t syntax); +#endif /* not RE_ENABLE_I18N */ +static bin_tree_t *build_charclass_op (re_dfa_t *dfa, + RE_TRANSLATE_TYPE trans, + const char *class_name, + const char *extra, + bool non_match, reg_errcode_t *err); +static bin_tree_t *create_tree (re_dfa_t *dfa, + bin_tree_t *left, bin_tree_t *right, + re_token_type_t type); +static bin_tree_t *create_token_tree (re_dfa_t *dfa, + bin_tree_t *left, bin_tree_t *right, + const re_token_t *token); +static bin_tree_t *duplicate_tree (const bin_tree_t *src, re_dfa_t *dfa); +static void free_token (re_token_t *node); +static reg_errcode_t free_tree (void *extra, bin_tree_t *node); +static reg_errcode_t mark_opt_subexp (void *extra, bin_tree_t *node); + +/* This table gives an error message for each of the error codes listed + in regex.h. Obviously the order here has to be same as there. + POSIX doesn't require that we do anything for REG_NOERROR, + but why not be nice? */ + +static const char __re_error_msgid[] = + { +#define REG_NOERROR_IDX 0 + gettext_noop ("Success") /* REG_NOERROR */ + "\0" +#define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success") + gettext_noop ("No match") /* REG_NOMATCH */ + "\0" +#define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match") + gettext_noop ("Invalid regular expression") /* REG_BADPAT */ + "\0" +#define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression") + gettext_noop ("Invalid collation character") /* REG_ECOLLATE */ + "\0" +#define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character") + gettext_noop ("Invalid character class name") /* REG_ECTYPE */ + "\0" +#define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name") + gettext_noop ("Trailing backslash") /* REG_EESCAPE */ + "\0" +#define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash") + gettext_noop ("Invalid back reference") /* REG_ESUBREG */ + "\0" +#define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference") + gettext_noop ("Unmatched [, [^, [:, [., or [=") /* REG_EBRACK */ + "\0" +#define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [, [^, [:, [., or [=") + gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */ + "\0" +#define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(") + gettext_noop ("Unmatched \\{") /* REG_EBRACE */ + "\0" +#define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{") + gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */ + "\0" +#define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}") + gettext_noop ("Invalid range end") /* REG_ERANGE */ + "\0" +#define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end") + gettext_noop ("Memory exhausted") /* REG_ESPACE */ + "\0" +#define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted") + gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */ + "\0" +#define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression") + gettext_noop ("Premature end of regular expression") /* REG_EEND */ + "\0" +#define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression") + gettext_noop ("Regular expression too big") /* REG_ESIZE */ + "\0" +#define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big") + gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */ + }; + +static const size_t __re_error_msgid_idx[] = + { + REG_NOERROR_IDX, + REG_NOMATCH_IDX, + REG_BADPAT_IDX, + REG_ECOLLATE_IDX, + REG_ECTYPE_IDX, + REG_EESCAPE_IDX, + REG_ESUBREG_IDX, + REG_EBRACK_IDX, + REG_EPAREN_IDX, + REG_EBRACE_IDX, + REG_BADBR_IDX, + REG_ERANGE_IDX, + REG_ESPACE_IDX, + REG_BADRPT_IDX, + REG_EEND_IDX, + REG_ESIZE_IDX, + REG_ERPAREN_IDX + }; + +/* Entry points for GNU code. */ + +/* re_compile_pattern is the GNU regular expression compiler: it + compiles PATTERN (of length LENGTH) and puts the result in BUFP. + Returns 0 if the pattern was valid, otherwise an error string. + + Assumes the 'allocated' (and perhaps 'buffer') and 'translate' fields + are set in BUFP on entry. */ + +const char * +re_compile_pattern (const char *pattern, size_t length, + struct re_pattern_buffer *bufp) +{ + reg_errcode_t ret; + + /* And GNU code determines whether or not to get register information + by passing null for the REGS argument to re_match, etc., not by + setting no_sub, unless RE_NO_SUB is set. */ + bufp->no_sub = !!(re_syntax_options & RE_NO_SUB); + + /* Match anchors at newline. */ + bufp->newline_anchor = 1; + + ret = re_compile_internal (bufp, pattern, length, re_syntax_options); + + if (!ret) + return NULL; + return gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); +} +weak_alias (__re_compile_pattern, re_compile_pattern) + +/* Set by 're_set_syntax' to the current regexp syntax to recognize. Can + also be assigned to arbitrarily: each pattern buffer stores its own + syntax, so it can be changed between regex compilations. */ +/* This has no initializer because initialized variables in Emacs + become read-only after dumping. */ +reg_syntax_t re_syntax_options; + + +/* Specify the precise syntax of regexps for compilation. This provides + for compatibility for various utilities which historically have + different, incompatible syntaxes. + + The argument SYNTAX is a bit mask comprised of the various bits + defined in regex.h. We return the old syntax. */ + +reg_syntax_t +re_set_syntax (reg_syntax_t syntax) +{ + reg_syntax_t ret = re_syntax_options; + + re_syntax_options = syntax; + return ret; +} +weak_alias (__re_set_syntax, re_set_syntax) + +int +re_compile_fastmap (struct re_pattern_buffer *bufp) +{ + re_dfa_t *dfa = bufp->buffer; + char *fastmap = bufp->fastmap; + + memset (fastmap, '\0', sizeof (char) * SBC_MAX); + re_compile_fastmap_iter (bufp, dfa->init_state, fastmap); + if (dfa->init_state != dfa->init_state_word) + re_compile_fastmap_iter (bufp, dfa->init_state_word, fastmap); + if (dfa->init_state != dfa->init_state_nl) + re_compile_fastmap_iter (bufp, dfa->init_state_nl, fastmap); + if (dfa->init_state != dfa->init_state_begbuf) + re_compile_fastmap_iter (bufp, dfa->init_state_begbuf, fastmap); + bufp->fastmap_accurate = 1; + return 0; +} +weak_alias (__re_compile_fastmap, re_compile_fastmap) + +static inline void +__attribute__ ((always_inline)) +re_set_fastmap (char *fastmap, bool icase, int ch) +{ + fastmap[ch] = 1; + if (icase) + fastmap[tolower (ch)] = 1; +} + +/* Helper function for re_compile_fastmap. + Compile fastmap for the initial_state INIT_STATE. */ + +static void +re_compile_fastmap_iter (regex_t *bufp, const re_dfastate_t *init_state, + char *fastmap) +{ + re_dfa_t *dfa = bufp->buffer; + Idx node_cnt; + bool icase = (dfa->mb_cur_max == 1 && (bufp->syntax & RE_ICASE)); + for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) + { + Idx node = init_state->nodes.elems[node_cnt]; + re_token_type_t type = dfa->nodes[node].type; + + if (type == CHARACTER) + { + re_set_fastmap (fastmap, icase, dfa->nodes[node].opr.c); +#ifdef RE_ENABLE_I18N + if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) + { + unsigned char buf[MB_LEN_MAX]; + unsigned char *p; + wchar_t wc; + mbstate_t state; + + p = buf; + *p++ = dfa->nodes[node].opr.c; + while (++node < dfa->nodes_len + && dfa->nodes[node].type == CHARACTER + && dfa->nodes[node].mb_partial) + *p++ = dfa->nodes[node].opr.c; + memset (&state, '\0', sizeof (state)); + if (__mbrtowc (&wc, (const char *) buf, p - buf, + &state) == p - buf + && (__wcrtomb ((char *) buf, __towlower (wc), &state) + != (size_t) -1)) + re_set_fastmap (fastmap, false, buf[0]); + } +#endif + } + else if (type == SIMPLE_BRACKET) + { + int i, ch; + for (i = 0, ch = 0; i < BITSET_WORDS; ++i) + { + int j; + bitset_word_t w = dfa->nodes[node].opr.sbcset[i]; + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + if (w & ((bitset_word_t) 1 << j)) + re_set_fastmap (fastmap, icase, ch); + } + } +#ifdef RE_ENABLE_I18N + else if (type == COMPLEX_BRACKET) + { + re_charset_t *cset = dfa->nodes[node].opr.mbcset; + Idx i; + +# ifdef _LIBC + /* See if we have to try all bytes which start multiple collation + elements. + e.g. In da_DK, we want to catch 'a' since "aa" is a valid + collation element, and don't catch 'b' since 'b' is + the only collation element which starts from 'b' (and + it is caught by SIMPLE_BRACKET). */ + if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0 + && (cset->ncoll_syms || cset->nranges)) + { + const int32_t *table = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + for (i = 0; i < SBC_MAX; ++i) + if (table[i] < 0) + re_set_fastmap (fastmap, icase, i); + } +# endif /* _LIBC */ + + /* See if we have to start the match at all multibyte characters, + i.e. where we would not find an invalid sequence. This only + applies to multibyte character sets; for single byte character + sets, the SIMPLE_BRACKET again suffices. */ + if (dfa->mb_cur_max > 1 + && (cset->nchar_classes || cset->non_match || cset->nranges +# ifdef _LIBC + || cset->nequiv_classes +# endif /* _LIBC */ + )) + { + unsigned char c = 0; + do + { + mbstate_t mbs; + memset (&mbs, 0, sizeof (mbs)); + if (__mbrtowc (NULL, (char *) &c, 1, &mbs) == (size_t) -2) + re_set_fastmap (fastmap, false, (int) c); + } + while (++c != 0); + } + + else + { + /* ... Else catch all bytes which can start the mbchars. */ + for (i = 0; i < cset->nmbchars; ++i) + { + char buf[256]; + mbstate_t state; + memset (&state, '\0', sizeof (state)); + if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1) + re_set_fastmap (fastmap, icase, *(unsigned char *) buf); + if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) + { + if (__wcrtomb (buf, __towlower (cset->mbchars[i]), &state) + != (size_t) -1) + re_set_fastmap (fastmap, false, *(unsigned char *) buf); + } + } + } + } +#endif /* RE_ENABLE_I18N */ + else if (type == OP_PERIOD +#ifdef RE_ENABLE_I18N + || type == OP_UTF8_PERIOD +#endif /* RE_ENABLE_I18N */ + || type == END_OF_RE) + { + memset (fastmap, '\1', sizeof (char) * SBC_MAX); + if (type == END_OF_RE) + bufp->can_be_null = 1; + return; + } + } +} + +/* Entry point for POSIX code. */ +/* regcomp takes a regular expression as a string and compiles it. + + PREG is a regex_t *. We do not expect any fields to be initialized, + since POSIX says we shouldn't. Thus, we set + + 'buffer' to the compiled pattern; + 'used' to the length of the compiled pattern; + 'syntax' to RE_SYNTAX_POSIX_EXTENDED if the + REG_EXTENDED bit in CFLAGS is set; otherwise, to + RE_SYNTAX_POSIX_BASIC; + 'newline_anchor' to REG_NEWLINE being set in CFLAGS; + 'fastmap' to an allocated space for the fastmap; + 'fastmap_accurate' to zero; + 're_nsub' to the number of subexpressions in PATTERN. + + PATTERN is the address of the pattern string. + + CFLAGS is a series of bits which affect compilation. + + If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we + use POSIX basic syntax. + + If REG_NEWLINE is set, then . and [^...] don't match newline. + Also, regexec will try a match beginning after every newline. + + If REG_ICASE is set, then we considers upper- and lowercase + versions of letters to be equivalent when matching. + + If REG_NOSUB is set, then when PREG is passed to regexec, that + routine will report only success or failure, and nothing about the + registers. + + It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for + the return codes and their meanings.) */ + +int +regcomp (regex_t *__restrict preg, const char *__restrict pattern, int cflags) +{ + reg_errcode_t ret; + reg_syntax_t syntax = ((cflags & REG_EXTENDED) ? RE_SYNTAX_POSIX_EXTENDED + : RE_SYNTAX_POSIX_BASIC); + + preg->buffer = NULL; + preg->allocated = 0; + preg->used = 0; + + /* Try to allocate space for the fastmap. */ + preg->fastmap = re_malloc (char, SBC_MAX); + if (__glibc_unlikely (preg->fastmap == NULL)) + return REG_ESPACE; + + syntax |= (cflags & REG_ICASE) ? RE_ICASE : 0; + + /* If REG_NEWLINE is set, newlines are treated differently. */ + if (cflags & REG_NEWLINE) + { /* REG_NEWLINE implies neither . nor [^...] match newline. */ + syntax &= ~RE_DOT_NEWLINE; + syntax |= RE_HAT_LISTS_NOT_NEWLINE; + /* It also changes the matching behavior. */ + preg->newline_anchor = 1; + } + else + preg->newline_anchor = 0; + preg->no_sub = !!(cflags & REG_NOSUB); + preg->translate = NULL; + + ret = re_compile_internal (preg, pattern, strlen (pattern), syntax); + + /* POSIX doesn't distinguish between an unmatched open-group and an + unmatched close-group: both are REG_EPAREN. */ + if (ret == REG_ERPAREN) + ret = REG_EPAREN; + + /* We have already checked preg->fastmap != NULL. */ + if (__glibc_likely (ret == REG_NOERROR)) + /* Compute the fastmap now, since regexec cannot modify the pattern + buffer. This function never fails in this implementation. */ + (void) re_compile_fastmap (preg); + else + { + /* Some error occurred while compiling the expression. */ + re_free (preg->fastmap); + preg->fastmap = NULL; + } + + return (int) ret; +} +libc_hidden_def (__regcomp) +weak_alias (__regcomp, regcomp) + +/* Returns a message corresponding to an error code, ERRCODE, returned + from either regcomp or regexec. We don't use PREG here. */ + +size_t +regerror (int errcode, const regex_t *__restrict preg, char *__restrict errbuf, + size_t errbuf_size) +{ + const char *msg; + size_t msg_size; + int nerrcodes = sizeof __re_error_msgid_idx / sizeof __re_error_msgid_idx[0]; + + if (__glibc_unlikely (errcode < 0 || errcode >= nerrcodes)) + /* Only error codes returned by the rest of the code should be passed + to this routine. If we are given anything else, or if other regex + code generates an invalid error code, then the program has a bug. + Dump core so we can fix it. */ + abort (); + + msg = gettext (__re_error_msgid + __re_error_msgid_idx[errcode]); + + msg_size = strlen (msg) + 1; /* Includes the null. */ + + if (__glibc_likely (errbuf_size != 0)) + { + size_t cpy_size = msg_size; + if (__glibc_unlikely (msg_size > errbuf_size)) + { + cpy_size = errbuf_size - 1; + errbuf[cpy_size] = '\0'; + } + memcpy (errbuf, msg, cpy_size); + } + + return msg_size; +} +weak_alias (__regerror, regerror) + + +#ifdef RE_ENABLE_I18N +/* This static array is used for the map to single-byte characters when + UTF-8 is used. Otherwise we would allocate memory just to initialize + it the same all the time. UTF-8 is the preferred encoding so this is + a worthwhile optimization. */ +static const bitset_t utf8_sb_map = +{ + /* Set the first 128 bits. */ +# if defined __GNUC__ && !defined __STRICT_ANSI__ + [0 ... 0x80 / BITSET_WORD_BITS - 1] = BITSET_WORD_MAX +# else +# if 4 * BITSET_WORD_BITS < ASCII_CHARS +# error "bitset_word_t is narrower than 32 bits" +# elif 3 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, BITSET_WORD_MAX, BITSET_WORD_MAX, +# elif 2 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, BITSET_WORD_MAX, +# elif 1 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, +# endif + (BITSET_WORD_MAX + >> (SBC_MAX % BITSET_WORD_BITS == 0 + ? 0 + : BITSET_WORD_BITS - SBC_MAX % BITSET_WORD_BITS)) +# endif +}; +#endif + + +static void +free_dfa_content (re_dfa_t *dfa) +{ + Idx i, j; + + if (dfa->nodes) + for (i = 0; i < dfa->nodes_len; ++i) + free_token (dfa->nodes + i); + re_free (dfa->nexts); + for (i = 0; i < dfa->nodes_len; ++i) + { + if (dfa->eclosures != NULL) + re_node_set_free (dfa->eclosures + i); + if (dfa->inveclosures != NULL) + re_node_set_free (dfa->inveclosures + i); + if (dfa->edests != NULL) + re_node_set_free (dfa->edests + i); + } + re_free (dfa->edests); + re_free (dfa->eclosures); + re_free (dfa->inveclosures); + re_free (dfa->nodes); + + if (dfa->state_table) + for (i = 0; i <= dfa->state_hash_mask; ++i) + { + struct re_state_table_entry *entry = dfa->state_table + i; + for (j = 0; j < entry->num; ++j) + { + re_dfastate_t *state = entry->array[j]; + free_state (state); + } + re_free (entry->array); + } + re_free (dfa->state_table); +#ifdef RE_ENABLE_I18N + if (dfa->sb_char != utf8_sb_map) + re_free (dfa->sb_char); +#endif + re_free (dfa->subexp_map); +#ifdef DEBUG + re_free (dfa->re_str); +#endif + + re_free (dfa); +} + + +/* Free dynamically allocated space used by PREG. */ + +void +regfree (regex_t *preg) +{ + re_dfa_t *dfa = preg->buffer; + if (__glibc_likely (dfa != NULL)) + { + lock_fini (dfa->lock); + free_dfa_content (dfa); + } + preg->buffer = NULL; + preg->allocated = 0; + + re_free (preg->fastmap); + preg->fastmap = NULL; + + re_free (preg->translate); + preg->translate = NULL; +} +libc_hidden_def (__regfree) +weak_alias (__regfree, regfree) + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them unless specifically requested. */ + +#if defined _REGEX_RE_COMP || defined _LIBC + +/* BSD has one and only one pattern buffer. */ +static struct re_pattern_buffer re_comp_buf; + +char * +# ifdef _LIBC +/* Make these definitions weak in libc, so POSIX programs can redefine + these names if they don't use our functions, and still use + regcomp/regexec above without link errors. */ +weak_function +# endif +re_comp (const char *s) +{ + reg_errcode_t ret; + char *fastmap; + + if (!s) + { + if (!re_comp_buf.buffer) + return gettext ("No previous regular expression"); + return 0; + } + + if (re_comp_buf.buffer) + { + fastmap = re_comp_buf.fastmap; + re_comp_buf.fastmap = NULL; + __regfree (&re_comp_buf); + memset (&re_comp_buf, '\0', sizeof (re_comp_buf)); + re_comp_buf.fastmap = fastmap; + } + + if (re_comp_buf.fastmap == NULL) + { + re_comp_buf.fastmap = re_malloc (char, SBC_MAX); + if (re_comp_buf.fastmap == NULL) + return (char *) gettext (__re_error_msgid + + __re_error_msgid_idx[(int) REG_ESPACE]); + } + + /* Since 're_exec' always passes NULL for the 'regs' argument, we + don't need to initialize the pattern buffer fields which affect it. */ + + /* Match anchors at newlines. */ + re_comp_buf.newline_anchor = 1; + + ret = re_compile_internal (&re_comp_buf, s, strlen (s), re_syntax_options); + + if (!ret) + return NULL; + + /* Yes, we're discarding 'const' here if !HAVE_LIBINTL. */ + return (char *) gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); +} + +#ifdef _LIBC +libc_freeres_fn (free_mem) +{ + __regfree (&re_comp_buf); +} +#endif + +#endif /* _REGEX_RE_COMP */ + +/* Internal entry point. + Compile the regular expression PATTERN, whose length is LENGTH. + SYNTAX indicate regular expression's syntax. */ + +static reg_errcode_t +re_compile_internal (regex_t *preg, const char * pattern, size_t length, + reg_syntax_t syntax) +{ + reg_errcode_t err = REG_NOERROR; + re_dfa_t *dfa; + re_string_t regexp; + + /* Initialize the pattern buffer. */ + preg->fastmap_accurate = 0; + preg->syntax = syntax; + preg->not_bol = preg->not_eol = 0; + preg->used = 0; + preg->re_nsub = 0; + preg->can_be_null = 0; + preg->regs_allocated = REGS_UNALLOCATED; + + /* Initialize the dfa. */ + dfa = preg->buffer; + if (__glibc_unlikely (preg->allocated < sizeof (re_dfa_t))) + { + /* If zero allocated, but buffer is non-null, try to realloc + enough space. This loses if buffer's address is bogus, but + that is the user's responsibility. If ->buffer is NULL this + is a simple allocation. */ + dfa = re_realloc (preg->buffer, re_dfa_t, 1); + if (dfa == NULL) + return REG_ESPACE; + preg->allocated = sizeof (re_dfa_t); + preg->buffer = dfa; + } + preg->used = sizeof (re_dfa_t); + + err = init_dfa (dfa, length); + if (__glibc_unlikely (err == REG_NOERROR && lock_init (dfa->lock) != 0)) + err = REG_ESPACE; + if (__glibc_unlikely (err != REG_NOERROR)) + { + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + return err; + } +#ifdef DEBUG + /* Note: length+1 will not overflow since it is checked in init_dfa. */ + dfa->re_str = re_malloc (char, length + 1); + strncpy (dfa->re_str, pattern, length + 1); +#endif + + err = re_string_construct (®exp, pattern, length, preg->translate, + (syntax & RE_ICASE) != 0, dfa); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_compile_internal_free_return: + free_workarea_compile (preg); + re_string_destruct (®exp); + lock_fini (dfa->lock); + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + return err; + } + + /* Parse the regular expression, and build a structure tree. */ + preg->re_nsub = 0; + dfa->str_tree = parse (®exp, preg, syntax, &err); + if (__glibc_unlikely (dfa->str_tree == NULL)) + goto re_compile_internal_free_return; + + /* Analyze the tree and create the nfa. */ + err = analyze (preg); + if (__glibc_unlikely (err != REG_NOERROR)) + goto re_compile_internal_free_return; + +#ifdef RE_ENABLE_I18N + /* If possible, do searching in single byte encoding to speed things up. */ + if (dfa->is_utf8 && !(syntax & RE_ICASE) && preg->translate == NULL) + optimize_utf8 (dfa); +#endif + + /* Then create the initial state of the dfa. */ + err = create_initial_state (dfa); + + /* Release work areas. */ + free_workarea_compile (preg); + re_string_destruct (®exp); + + if (__glibc_unlikely (err != REG_NOERROR)) + { + lock_fini (dfa->lock); + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + } + + return err; +} + +/* Initialize DFA. We use the length of the regular expression PAT_LEN + as the initial length of some arrays. */ + +static reg_errcode_t +init_dfa (re_dfa_t *dfa, size_t pat_len) +{ + __re_size_t table_size; +#ifndef _LIBC + const char *codeset_name; +#endif +#ifdef RE_ENABLE_I18N + size_t max_i18n_object_size = MAX (sizeof (wchar_t), sizeof (wctype_t)); +#else + size_t max_i18n_object_size = 0; +#endif + size_t max_object_size = + MAX (sizeof (struct re_state_table_entry), + MAX (sizeof (re_token_t), + MAX (sizeof (re_node_set), + MAX (sizeof (regmatch_t), + max_i18n_object_size)))); + + memset (dfa, '\0', sizeof (re_dfa_t)); + + /* Force allocation of str_tree_storage the first time. */ + dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; + + /* Avoid overflows. The extra "/ 2" is for the table_size doubling + calculation below, and for similar doubling calculations + elsewhere. And it's <= rather than <, because some of the + doubling calculations add 1 afterwards. */ + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) / 2 + <= pat_len)) + return REG_ESPACE; + + dfa->nodes_alloc = pat_len + 1; + dfa->nodes = re_malloc (re_token_t, dfa->nodes_alloc); + + /* table_size = 2 ^ ceil(log pat_len) */ + for (table_size = 1; ; table_size <<= 1) + if (table_size > pat_len) + break; + + dfa->state_table = calloc (sizeof (struct re_state_table_entry), table_size); + dfa->state_hash_mask = table_size - 1; + + dfa->mb_cur_max = MB_CUR_MAX; +#ifdef _LIBC + if (dfa->mb_cur_max == 6 + && strcmp (_NL_CURRENT (LC_CTYPE, _NL_CTYPE_CODESET_NAME), "UTF-8") == 0) + dfa->is_utf8 = 1; + dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII) + != 0); +#else + codeset_name = nl_langinfo (CODESET); + if ((codeset_name[0] == 'U' || codeset_name[0] == 'u') + && (codeset_name[1] == 'T' || codeset_name[1] == 't') + && (codeset_name[2] == 'F' || codeset_name[2] == 'f') + && strcmp (codeset_name + 3 + (codeset_name[3] == '-'), "8") == 0) + dfa->is_utf8 = 1; + + /* We check exhaustively in the loop below if this charset is a + superset of ASCII. */ + dfa->map_notascii = 0; +#endif + +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + if (dfa->is_utf8) + dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map; + else + { + int i, j, ch; + + dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); + if (__glibc_unlikely (dfa->sb_char == NULL)) + return REG_ESPACE; + + /* Set the bits corresponding to single byte chars. */ + for (i = 0, ch = 0; i < BITSET_WORDS; ++i) + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + { + wint_t wch = __btowc (ch); + if (wch != WEOF) + dfa->sb_char[i] |= (bitset_word_t) 1 << j; +# ifndef _LIBC + if (isascii (ch) && wch != ch) + dfa->map_notascii = 1; +# endif + } + } + } +#endif + + if (__glibc_unlikely (dfa->nodes == NULL || dfa->state_table == NULL)) + return REG_ESPACE; + return REG_NOERROR; +} + +/* Initialize WORD_CHAR table, which indicate which character is + "word". In this case "word" means that it is the word construction + character used by some operators like "\<", "\>", etc. */ + +static void +init_word_char (re_dfa_t *dfa) +{ + int i = 0; + int j; + int ch = 0; + dfa->word_ops_used = 1; + if (__glibc_likely (dfa->map_notascii == 0)) + { + /* Avoid uint32_t and uint64_t as some non-GCC platforms lack + them, an issue when this code is used in Gnulib. */ + bitset_word_t bits0 = 0x00000000; + bitset_word_t bits1 = 0x03ff0000; + bitset_word_t bits2 = 0x87fffffe; + bitset_word_t bits3 = 0x07fffffe; + if (BITSET_WORD_BITS == 64) + { + /* Pacify gcc -Woverflow on 32-bit platformns. */ + dfa->word_char[0] = bits1 << 31 << 1 | bits0; + dfa->word_char[1] = bits3 << 31 << 1 | bits2; + i = 2; + } + else if (BITSET_WORD_BITS == 32) + { + dfa->word_char[0] = bits0; + dfa->word_char[1] = bits1; + dfa->word_char[2] = bits2; + dfa->word_char[3] = bits3; + i = 4; + } + else + goto general_case; + ch = 128; + + if (__glibc_likely (dfa->is_utf8)) + { + memset (&dfa->word_char[i], '\0', (SBC_MAX - ch) / 8); + return; + } + } + + general_case: + for (; i < BITSET_WORDS; ++i) + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + if (isalnum (ch) || ch == '_') + dfa->word_char[i] |= (bitset_word_t) 1 << j; +} + +/* Free the work area which are only used while compiling. */ + +static void +free_workarea_compile (regex_t *preg) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_storage_t *storage, *next; + for (storage = dfa->str_tree_storage; storage; storage = next) + { + next = storage->next; + re_free (storage); + } + dfa->str_tree_storage = NULL; + dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; + dfa->str_tree = NULL; + re_free (dfa->org_indices); + dfa->org_indices = NULL; +} + +/* Create initial states for all contexts. */ + +static reg_errcode_t +create_initial_state (re_dfa_t *dfa) +{ + Idx first, i; + reg_errcode_t err; + re_node_set init_nodes; + + /* Initial states have the epsilon closure of the node which is + the first node of the regular expression. */ + first = dfa->str_tree->first->node_idx; + dfa->init_node = first; + err = re_node_set_init_copy (&init_nodes, dfa->eclosures + first); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + /* The back-references which are in initial states can epsilon transit, + since in this case all of the subexpressions can be null. + Then we add epsilon closures of the nodes which are the next nodes of + the back-references. */ + if (dfa->nbackref > 0) + for (i = 0; i < init_nodes.nelem; ++i) + { + Idx node_idx = init_nodes.elems[i]; + re_token_type_t type = dfa->nodes[node_idx].type; + + Idx clexp_idx; + if (type != OP_BACK_REF) + continue; + for (clexp_idx = 0; clexp_idx < init_nodes.nelem; ++clexp_idx) + { + re_token_t *clexp_node; + clexp_node = dfa->nodes + init_nodes.elems[clexp_idx]; + if (clexp_node->type == OP_CLOSE_SUBEXP + && clexp_node->opr.idx == dfa->nodes[node_idx].opr.idx) + break; + } + if (clexp_idx == init_nodes.nelem) + continue; + + if (type == OP_BACK_REF) + { + Idx dest_idx = dfa->edests[node_idx].elems[0]; + if (!re_node_set_contains (&init_nodes, dest_idx)) + { + reg_errcode_t merge_err + = re_node_set_merge (&init_nodes, dfa->eclosures + dest_idx); + if (merge_err != REG_NOERROR) + return merge_err; + i = 0; + } + } + } + + /* It must be the first time to invoke acquire_state. */ + dfa->init_state = re_acquire_state_context (&err, dfa, &init_nodes, 0); + /* We don't check ERR here, since the initial state must not be NULL. */ + if (__glibc_unlikely (dfa->init_state == NULL)) + return err; + if (dfa->init_state->has_constraint) + { + dfa->init_state_word = re_acquire_state_context (&err, dfa, &init_nodes, + CONTEXT_WORD); + dfa->init_state_nl = re_acquire_state_context (&err, dfa, &init_nodes, + CONTEXT_NEWLINE); + dfa->init_state_begbuf = re_acquire_state_context (&err, dfa, + &init_nodes, + CONTEXT_NEWLINE + | CONTEXT_BEGBUF); + if (__glibc_unlikely (dfa->init_state_word == NULL + || dfa->init_state_nl == NULL + || dfa->init_state_begbuf == NULL)) + return err; + } + else + dfa->init_state_word = dfa->init_state_nl + = dfa->init_state_begbuf = dfa->init_state; + + re_node_set_free (&init_nodes); + return REG_NOERROR; +} + +#ifdef RE_ENABLE_I18N +/* If it is possible to do searching in single byte encoding instead of UTF-8 + to speed things up, set dfa->mb_cur_max to 1, clear is_utf8 and change + DFA nodes where needed. */ + +static void +optimize_utf8 (re_dfa_t *dfa) +{ + Idx node; + int i; + bool mb_chars = false; + bool has_period = false; + + for (node = 0; node < dfa->nodes_len; ++node) + switch (dfa->nodes[node].type) + { + case CHARACTER: + if (dfa->nodes[node].opr.c >= ASCII_CHARS) + mb_chars = true; + break; + case ANCHOR: + switch (dfa->nodes[node].opr.ctx_type) + { + case LINE_FIRST: + case LINE_LAST: + case BUF_FIRST: + case BUF_LAST: + break; + default: + /* Word anchors etc. cannot be handled. It's okay to test + opr.ctx_type since constraints (for all DFA nodes) are + created by ORing one or more opr.ctx_type values. */ + return; + } + break; + case OP_PERIOD: + has_period = true; + break; + case OP_BACK_REF: + case OP_ALT: + case END_OF_RE: + case OP_DUP_ASTERISK: + case OP_OPEN_SUBEXP: + case OP_CLOSE_SUBEXP: + break; + case COMPLEX_BRACKET: + return; + case SIMPLE_BRACKET: + /* Just double check. */ + { + int rshift = (ASCII_CHARS % BITSET_WORD_BITS == 0 + ? 0 + : BITSET_WORD_BITS - ASCII_CHARS % BITSET_WORD_BITS); + for (i = ASCII_CHARS / BITSET_WORD_BITS; i < BITSET_WORDS; ++i) + { + if (dfa->nodes[node].opr.sbcset[i] >> rshift != 0) + return; + rshift = 0; + } + } + break; + default: + abort (); + } + + if (mb_chars || has_period) + for (node = 0; node < dfa->nodes_len; ++node) + { + if (dfa->nodes[node].type == CHARACTER + && dfa->nodes[node].opr.c >= ASCII_CHARS) + dfa->nodes[node].mb_partial = 0; + else if (dfa->nodes[node].type == OP_PERIOD) + dfa->nodes[node].type = OP_UTF8_PERIOD; + } + + /* The search can be in single byte locale. */ + dfa->mb_cur_max = 1; + dfa->is_utf8 = 0; + dfa->has_mb_node = dfa->nbackref > 0 || has_period; +} +#endif + +/* Analyze the structure tree, and calculate "first", "next", "edest", + "eclosure", and "inveclosure". */ + +static reg_errcode_t +analyze (regex_t *preg) +{ + re_dfa_t *dfa = preg->buffer; + reg_errcode_t ret; + + /* Allocate arrays. */ + dfa->nexts = re_malloc (Idx, dfa->nodes_alloc); + dfa->org_indices = re_malloc (Idx, dfa->nodes_alloc); + dfa->edests = re_malloc (re_node_set, dfa->nodes_alloc); + dfa->eclosures = re_malloc (re_node_set, dfa->nodes_alloc); + if (__glibc_unlikely (dfa->nexts == NULL || dfa->org_indices == NULL + || dfa->edests == NULL || dfa->eclosures == NULL)) + return REG_ESPACE; + + dfa->subexp_map = re_malloc (Idx, preg->re_nsub); + if (dfa->subexp_map != NULL) + { + Idx i; + for (i = 0; i < preg->re_nsub; i++) + dfa->subexp_map[i] = i; + preorder (dfa->str_tree, optimize_subexps, dfa); + for (i = 0; i < preg->re_nsub; i++) + if (dfa->subexp_map[i] != i) + break; + if (i == preg->re_nsub) + { + re_free (dfa->subexp_map); + dfa->subexp_map = NULL; + } + } + + ret = postorder (dfa->str_tree, lower_subexps, preg); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + ret = postorder (dfa->str_tree, calc_first, dfa); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + preorder (dfa->str_tree, calc_next, dfa); + ret = preorder (dfa->str_tree, link_nfa_nodes, dfa); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + ret = calc_eclosure (dfa); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + + /* We only need this during the prune_impossible_nodes pass in regexec.c; + skip it if p_i_n will not run, as calc_inveclosure can be quadratic. */ + if ((!preg->no_sub && preg->re_nsub > 0 && dfa->has_plural_match) + || dfa->nbackref) + { + dfa->inveclosures = re_malloc (re_node_set, dfa->nodes_len); + if (__glibc_unlikely (dfa->inveclosures == NULL)) + return REG_ESPACE; + ret = calc_inveclosure (dfa); + } + + return ret; +} + +/* Our parse trees are very unbalanced, so we cannot use a stack to + implement parse tree visits. Instead, we use parent pointers and + some hairy code in these two functions. */ +static reg_errcode_t +postorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra) +{ + bin_tree_t *node, *prev; + + for (node = root; ; ) + { + /* Descend down the tree, preferably to the left (or to the right + if that's the only child). */ + while (node->left || node->right) + if (node->left) + node = node->left; + else + node = node->right; + + do + { + reg_errcode_t err = fn (extra, node); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + if (node->parent == NULL) + return REG_NOERROR; + prev = node; + node = node->parent; + } + /* Go up while we have a node that is reached from the right. */ + while (node->right == prev || node->right == NULL); + node = node->right; + } +} + +static reg_errcode_t +preorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra) +{ + bin_tree_t *node; + + for (node = root; ; ) + { + reg_errcode_t err = fn (extra, node); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + /* Go to the left node, or up and to the right. */ + if (node->left) + node = node->left; + else + { + bin_tree_t *prev = NULL; + while (node->right == prev || node->right == NULL) + { + prev = node; + node = node->parent; + if (!node) + return REG_NOERROR; + } + node = node->right; + } + } +} + +/* Optimization pass: if a SUBEXP is entirely contained, strip it and tell + re_search_internal to map the inner one's opr.idx to this one's. Adjust + backreferences as well. Requires a preorder visit. */ +static reg_errcode_t +optimize_subexps (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + + if (node->token.type == OP_BACK_REF && dfa->subexp_map) + { + int idx = node->token.opr.idx; + node->token.opr.idx = dfa->subexp_map[idx]; + dfa->used_bkref_map |= 1 << node->token.opr.idx; + } + + else if (node->token.type == SUBEXP + && node->left && node->left->token.type == SUBEXP) + { + Idx other_idx = node->left->token.opr.idx; + + node->left = node->left->left; + if (node->left) + node->left->parent = node; + + dfa->subexp_map[other_idx] = dfa->subexp_map[node->token.opr.idx]; + if (other_idx < BITSET_WORD_BITS) + dfa->used_bkref_map &= ~((bitset_word_t) 1 << other_idx); + } + + return REG_NOERROR; +} + +/* Lowering pass: Turn each SUBEXP node into the appropriate concatenation + of OP_OPEN_SUBEXP, the body of the SUBEXP (if any) and OP_CLOSE_SUBEXP. */ +static reg_errcode_t +lower_subexps (void *extra, bin_tree_t *node) +{ + regex_t *preg = (regex_t *) extra; + reg_errcode_t err = REG_NOERROR; + + if (node->left && node->left->token.type == SUBEXP) + { + node->left = lower_subexp (&err, preg, node->left); + if (node->left) + node->left->parent = node; + } + if (node->right && node->right->token.type == SUBEXP) + { + node->right = lower_subexp (&err, preg, node->right); + if (node->right) + node->right->parent = node; + } + + return err; +} + +static bin_tree_t * +lower_subexp (reg_errcode_t *err, regex_t *preg, bin_tree_t *node) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_t *body = node->left; + bin_tree_t *op, *cls, *tree1, *tree; + + if (preg->no_sub + /* We do not optimize empty subexpressions, because otherwise we may + have bad CONCAT nodes with NULL children. This is obviously not + very common, so we do not lose much. An example that triggers + this case is the sed "script" /\(\)/x. */ + && node->left != NULL + && (node->token.opr.idx >= BITSET_WORD_BITS + || !(dfa->used_bkref_map + & ((bitset_word_t) 1 << node->token.opr.idx)))) + return node->left; + + /* Convert the SUBEXP node to the concatenation of an + OP_OPEN_SUBEXP, the contents, and an OP_CLOSE_SUBEXP. */ + op = create_tree (dfa, NULL, NULL, OP_OPEN_SUBEXP); + cls = create_tree (dfa, NULL, NULL, OP_CLOSE_SUBEXP); + tree1 = body ? create_tree (dfa, body, cls, CONCAT) : cls; + tree = create_tree (dfa, op, tree1, CONCAT); + if (__glibc_unlikely (tree == NULL || tree1 == NULL + || op == NULL || cls == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + + op->token.opr.idx = cls->token.opr.idx = node->token.opr.idx; + op->token.opt_subexp = cls->token.opt_subexp = node->token.opt_subexp; + return tree; +} + +/* Pass 1 in building the NFA: compute FIRST and create unlinked automaton + nodes. Requires a postorder visit. */ +static reg_errcode_t +calc_first (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + if (node->token.type == CONCAT) + { + node->first = node->left->first; + node->node_idx = node->left->node_idx; + } + else + { + node->first = node; + node->node_idx = re_dfa_add_node (dfa, node->token); + if (__glibc_unlikely (node->node_idx == -1)) + return REG_ESPACE; + if (node->token.type == ANCHOR) + dfa->nodes[node->node_idx].constraint = node->token.opr.ctx_type; + } + return REG_NOERROR; +} + +/* Pass 2: compute NEXT on the tree. Preorder visit. */ +static reg_errcode_t +calc_next (void *extra, bin_tree_t *node) +{ + switch (node->token.type) + { + case OP_DUP_ASTERISK: + node->left->next = node; + break; + case CONCAT: + node->left->next = node->right->first; + node->right->next = node->next; + break; + default: + if (node->left) + node->left->next = node->next; + if (node->right) + node->right->next = node->next; + break; + } + return REG_NOERROR; +} + +/* Pass 3: link all DFA nodes to their NEXT node (any order will do). */ +static reg_errcode_t +link_nfa_nodes (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + Idx idx = node->node_idx; + reg_errcode_t err = REG_NOERROR; + + switch (node->token.type) + { + case CONCAT: + break; + + case END_OF_RE: + assert (node->next == NULL); + break; + + case OP_DUP_ASTERISK: + case OP_ALT: + { + Idx left, right; + dfa->has_plural_match = 1; + if (node->left != NULL) + left = node->left->first->node_idx; + else + left = node->next->node_idx; + if (node->right != NULL) + right = node->right->first->node_idx; + else + right = node->next->node_idx; + assert (left > -1); + assert (right > -1); + err = re_node_set_init_2 (dfa->edests + idx, left, right); + } + break; + + case ANCHOR: + case OP_OPEN_SUBEXP: + case OP_CLOSE_SUBEXP: + err = re_node_set_init_1 (dfa->edests + idx, node->next->node_idx); + break; + + case OP_BACK_REF: + dfa->nexts[idx] = node->next->node_idx; + if (node->token.type == OP_BACK_REF) + err = re_node_set_init_1 (dfa->edests + idx, dfa->nexts[idx]); + break; + + default: + assert (!IS_EPSILON_NODE (node->token.type)); + dfa->nexts[idx] = node->next->node_idx; + break; + } + + return err; +} + +/* Duplicate the epsilon closure of the node ROOT_NODE. + Note that duplicated nodes have constraint INIT_CONSTRAINT in addition + to their own constraint. */ + +static reg_errcode_t +duplicate_node_closure (re_dfa_t *dfa, Idx top_org_node, Idx top_clone_node, + Idx root_node, unsigned int init_constraint) +{ + Idx org_node, clone_node; + bool ok; + unsigned int constraint = init_constraint; + for (org_node = top_org_node, clone_node = top_clone_node;;) + { + Idx org_dest, clone_dest; + if (dfa->nodes[org_node].type == OP_BACK_REF) + { + /* If the back reference epsilon-transit, its destination must + also have the constraint. Then duplicate the epsilon closure + of the destination of the back reference, and store it in + edests of the back reference. */ + org_dest = dfa->nexts[org_node]; + re_node_set_empty (dfa->edests + clone_node); + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (__glibc_unlikely (clone_dest == -1)) + return REG_ESPACE; + dfa->nexts[clone_node] = dfa->nexts[org_node]; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + else if (dfa->edests[org_node].nelem == 0) + { + /* In case of the node can't epsilon-transit, don't duplicate the + destination and store the original destination as the + destination of the node. */ + dfa->nexts[clone_node] = dfa->nexts[org_node]; + break; + } + else if (dfa->edests[org_node].nelem == 1) + { + /* In case of the node can epsilon-transit, and it has only one + destination. */ + org_dest = dfa->edests[org_node].elems[0]; + re_node_set_empty (dfa->edests + clone_node); + /* If the node is root_node itself, it means the epsilon closure + has a loop. Then tie it to the destination of the root_node. */ + if (org_node == root_node && clone_node != org_node) + { + ok = re_node_set_insert (dfa->edests + clone_node, org_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + break; + } + /* In case the node has another constraint, append it. */ + constraint |= dfa->nodes[org_node].constraint; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (__glibc_unlikely (clone_dest == -1)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + else /* dfa->edests[org_node].nelem == 2 */ + { + /* In case of the node can epsilon-transit, and it has two + destinations. In the bin_tree_t and DFA, that's '|' and '*'. */ + org_dest = dfa->edests[org_node].elems[0]; + re_node_set_empty (dfa->edests + clone_node); + /* Search for a duplicated node which satisfies the constraint. */ + clone_dest = search_duplicated_node (dfa, org_dest, constraint); + if (clone_dest == -1) + { + /* There is no such duplicated node, create a new one. */ + reg_errcode_t err; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (__glibc_unlikely (clone_dest == -1)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + err = duplicate_node_closure (dfa, org_dest, clone_dest, + root_node, constraint); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + else + { + /* There is a duplicated node which satisfies the constraint, + use it to avoid infinite loop. */ + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + + org_dest = dfa->edests[org_node].elems[1]; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (__glibc_unlikely (clone_dest == -1)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + org_node = org_dest; + clone_node = clone_dest; + } + return REG_NOERROR; +} + +/* Search for a node which is duplicated from the node ORG_NODE, and + satisfies the constraint CONSTRAINT. */ + +static Idx +search_duplicated_node (const re_dfa_t *dfa, Idx org_node, + unsigned int constraint) +{ + Idx idx; + for (idx = dfa->nodes_len - 1; dfa->nodes[idx].duplicated && idx > 0; --idx) + { + if (org_node == dfa->org_indices[idx] + && constraint == dfa->nodes[idx].constraint) + return idx; /* Found. */ + } + return -1; /* Not found. */ +} + +/* Duplicate the node whose index is ORG_IDX and set the constraint CONSTRAINT. + Return the index of the new node, or -1 if insufficient storage is + available. */ + +static Idx +duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint) +{ + Idx dup_idx = re_dfa_add_node (dfa, dfa->nodes[org_idx]); + if (__glibc_likely (dup_idx != -1)) + { + dfa->nodes[dup_idx].constraint = constraint; + dfa->nodes[dup_idx].constraint |= dfa->nodes[org_idx].constraint; + dfa->nodes[dup_idx].duplicated = 1; + + /* Store the index of the original node. */ + dfa->org_indices[dup_idx] = org_idx; + } + return dup_idx; +} + +static reg_errcode_t +calc_inveclosure (re_dfa_t *dfa) +{ + Idx src, idx; + bool ok; + for (idx = 0; idx < dfa->nodes_len; ++idx) + re_node_set_init_empty (dfa->inveclosures + idx); + + for (src = 0; src < dfa->nodes_len; ++src) + { + Idx *elems = dfa->eclosures[src].elems; + for (idx = 0; idx < dfa->eclosures[src].nelem; ++idx) + { + ok = re_node_set_insert_last (dfa->inveclosures + elems[idx], src); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + } + + return REG_NOERROR; +} + +/* Calculate "eclosure" for all the node in DFA. */ + +static reg_errcode_t +calc_eclosure (re_dfa_t *dfa) +{ + Idx node_idx; + bool incomplete; +#ifdef DEBUG + assert (dfa->nodes_len > 0); +#endif + incomplete = false; + /* For each nodes, calculate epsilon closure. */ + for (node_idx = 0; ; ++node_idx) + { + reg_errcode_t err; + re_node_set eclosure_elem; + if (node_idx == dfa->nodes_len) + { + if (!incomplete) + break; + incomplete = false; + node_idx = 0; + } + +#ifdef DEBUG + assert (dfa->eclosures[node_idx].nelem != -1); +#endif + + /* If we have already calculated, skip it. */ + if (dfa->eclosures[node_idx].nelem != 0) + continue; + /* Calculate epsilon closure of 'node_idx'. */ + err = calc_eclosure_iter (&eclosure_elem, dfa, node_idx, true); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + if (dfa->eclosures[node_idx].nelem == 0) + { + incomplete = true; + re_node_set_free (&eclosure_elem); + } + } + return REG_NOERROR; +} + +/* Calculate epsilon closure of NODE. */ + +static reg_errcode_t +calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, Idx node, bool root) +{ + reg_errcode_t err; + Idx i; + re_node_set eclosure; + bool ok; + bool incomplete = false; + err = re_node_set_alloc (&eclosure, dfa->edests[node].nelem + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + /* This indicates that we are calculating this node now. + We reference this value to avoid infinite loop. */ + dfa->eclosures[node].nelem = -1; + + /* If the current node has constraints, duplicate all nodes + since they must inherit the constraints. */ + if (dfa->nodes[node].constraint + && dfa->edests[node].nelem + && !dfa->nodes[dfa->edests[node].elems[0]].duplicated) + { + err = duplicate_node_closure (dfa, node, node, node, + dfa->nodes[node].constraint); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + + /* Expand each epsilon destination nodes. */ + if (IS_EPSILON_NODE(dfa->nodes[node].type)) + for (i = 0; i < dfa->edests[node].nelem; ++i) + { + re_node_set eclosure_elem; + Idx edest = dfa->edests[node].elems[i]; + /* If calculating the epsilon closure of 'edest' is in progress, + return intermediate result. */ + if (dfa->eclosures[edest].nelem == -1) + { + incomplete = true; + continue; + } + /* If we haven't calculated the epsilon closure of 'edest' yet, + calculate now. Otherwise use calculated epsilon closure. */ + if (dfa->eclosures[edest].nelem == 0) + { + err = calc_eclosure_iter (&eclosure_elem, dfa, edest, false); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + else + eclosure_elem = dfa->eclosures[edest]; + /* Merge the epsilon closure of 'edest'. */ + err = re_node_set_merge (&eclosure, &eclosure_elem); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + /* If the epsilon closure of 'edest' is incomplete, + the epsilon closure of this node is also incomplete. */ + if (dfa->eclosures[edest].nelem == 0) + { + incomplete = true; + re_node_set_free (&eclosure_elem); + } + } + + /* An epsilon closure includes itself. */ + ok = re_node_set_insert (&eclosure, node); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + if (incomplete && !root) + dfa->eclosures[node].nelem = 0; + else + dfa->eclosures[node] = eclosure; + *new_set = eclosure; + return REG_NOERROR; +} + +/* Functions for token which are used in the parser. */ + +/* Fetch a token from INPUT. + We must not use this function inside bracket expressions. */ + +static void +fetch_token (re_token_t *result, re_string_t *input, reg_syntax_t syntax) +{ + re_string_skip_bytes (input, peek_token (result, input, syntax)); +} + +/* Peek a token from INPUT, and return the length of the token. + We must not use this function inside bracket expressions. */ + +static int +peek_token (re_token_t *token, re_string_t *input, reg_syntax_t syntax) +{ + unsigned char c; + + if (re_string_eoi (input)) + { + token->type = END_OF_RE; + return 0; + } + + c = re_string_peek_byte (input, 0); + token->opr.c = c; + + token->word_char = 0; +#ifdef RE_ENABLE_I18N + token->mb_partial = 0; + if (input->mb_cur_max > 1 + && !re_string_first_byte (input, re_string_cur_idx (input))) + { + token->type = CHARACTER; + token->mb_partial = 1; + return 1; + } +#endif + if (c == '\\') + { + unsigned char c2; + if (re_string_cur_idx (input) + 1 >= re_string_length (input)) + { + token->type = BACK_SLASH; + return 1; + } + + c2 = re_string_peek_byte_case (input, 1); + token->opr.c = c2; + token->type = CHARACTER; +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc = re_string_wchar_at (input, + re_string_cur_idx (input) + 1); + token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; + } + else +#endif + token->word_char = IS_WORD_CHAR (c2) != 0; + + switch (c2) + { + case '|': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_NO_BK_VBAR)) + token->type = OP_ALT; + break; + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + if (!(syntax & RE_NO_BK_REFS)) + { + token->type = OP_BACK_REF; + token->opr.idx = c2 - '1'; + } + break; + case '<': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_FIRST; + } + break; + case '>': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_LAST; + } + break; + case 'b': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_DELIM; + } + break; + case 'B': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = NOT_WORD_DELIM; + } + break; + case 'w': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_WORD; + break; + case 'W': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_NOTWORD; + break; + case 's': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_SPACE; + break; + case 'S': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_NOTSPACE; + break; + case '`': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = BUF_FIRST; + } + break; + case '\'': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = BUF_LAST; + } + break; + case '(': + if (!(syntax & RE_NO_BK_PARENS)) + token->type = OP_OPEN_SUBEXP; + break; + case ')': + if (!(syntax & RE_NO_BK_PARENS)) + token->type = OP_CLOSE_SUBEXP; + break; + case '+': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_PLUS; + break; + case '?': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_QUESTION; + break; + case '{': + if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) + token->type = OP_OPEN_DUP_NUM; + break; + case '}': + if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) + token->type = OP_CLOSE_DUP_NUM; + break; + default: + break; + } + return 2; + } + + token->type = CHARACTER; +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc = re_string_wchar_at (input, re_string_cur_idx (input)); + token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; + } + else +#endif + token->word_char = IS_WORD_CHAR (token->opr.c); + + switch (c) + { + case '\n': + if (syntax & RE_NEWLINE_ALT) + token->type = OP_ALT; + break; + case '|': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_NO_BK_VBAR)) + token->type = OP_ALT; + break; + case '*': + token->type = OP_DUP_ASTERISK; + break; + case '+': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_PLUS; + break; + case '?': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_QUESTION; + break; + case '{': + if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + token->type = OP_OPEN_DUP_NUM; + break; + case '}': + if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + token->type = OP_CLOSE_DUP_NUM; + break; + case '(': + if (syntax & RE_NO_BK_PARENS) + token->type = OP_OPEN_SUBEXP; + break; + case ')': + if (syntax & RE_NO_BK_PARENS) + token->type = OP_CLOSE_SUBEXP; + break; + case '[': + token->type = OP_OPEN_BRACKET; + break; + case '.': + token->type = OP_PERIOD; + break; + case '^': + if (!(syntax & (RE_CONTEXT_INDEP_ANCHORS | RE_CARET_ANCHORS_HERE)) + && re_string_cur_idx (input) != 0) + { + char prev = re_string_peek_byte (input, -1); + if (!(syntax & RE_NEWLINE_ALT) || prev != '\n') + break; + } + token->type = ANCHOR; + token->opr.ctx_type = LINE_FIRST; + break; + case '$': + if (!(syntax & RE_CONTEXT_INDEP_ANCHORS) + && re_string_cur_idx (input) + 1 != re_string_length (input)) + { + re_token_t next; + re_string_skip_bytes (input, 1); + peek_token (&next, input, syntax); + re_string_skip_bytes (input, -1); + if (next.type != OP_ALT && next.type != OP_CLOSE_SUBEXP) + break; + } + token->type = ANCHOR; + token->opr.ctx_type = LINE_LAST; + break; + default: + break; + } + return 1; +} + +/* Peek a token from INPUT, and return the length of the token. + We must not use this function out of bracket expressions. */ + +static int +peek_token_bracket (re_token_t *token, re_string_t *input, reg_syntax_t syntax) +{ + unsigned char c; + if (re_string_eoi (input)) + { + token->type = END_OF_RE; + return 0; + } + c = re_string_peek_byte (input, 0); + token->opr.c = c; + +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1 + && !re_string_first_byte (input, re_string_cur_idx (input))) + { + token->type = CHARACTER; + return 1; + } +#endif /* RE_ENABLE_I18N */ + + if (c == '\\' && (syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) + && re_string_cur_idx (input) + 1 < re_string_length (input)) + { + /* In this case, '\' escape a character. */ + unsigned char c2; + re_string_skip_bytes (input, 1); + c2 = re_string_peek_byte (input, 0); + token->opr.c = c2; + token->type = CHARACTER; + return 1; + } + if (c == '[') /* '[' is a special char in a bracket exps. */ + { + unsigned char c2; + int token_len; + if (re_string_cur_idx (input) + 1 < re_string_length (input)) + c2 = re_string_peek_byte (input, 1); + else + c2 = 0; + token->opr.c = c2; + token_len = 2; + switch (c2) + { + case '.': + token->type = OP_OPEN_COLL_ELEM; + break; + + case '=': + token->type = OP_OPEN_EQUIV_CLASS; + break; + + case ':': + if (syntax & RE_CHAR_CLASSES) + { + token->type = OP_OPEN_CHAR_CLASS; + break; + } + FALLTHROUGH; + default: + token->type = CHARACTER; + token->opr.c = c; + token_len = 1; + break; + } + return token_len; + } + switch (c) + { + case '-': + token->type = OP_CHARSET_RANGE; + break; + case ']': + token->type = OP_CLOSE_BRACKET; + break; + case '^': + token->type = OP_NON_MATCH_LIST; + break; + default: + token->type = CHARACTER; + } + return 1; +} + +/* Functions for parser. */ + +/* Entry point of the parser. + Parse the regular expression REGEXP and return the structure tree. + If an error occurs, ERR is set by error code, and return NULL. + This function build the following tree, from regular expression <reg_exp>: + CAT + / \ + / \ + <reg_exp> EOR + + CAT means concatenation. + EOR means end of regular expression. */ + +static bin_tree_t * +parse (re_string_t *regexp, regex_t *preg, reg_syntax_t syntax, + reg_errcode_t *err) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_t *tree, *eor, *root; + re_token_t current_token; + dfa->syntax = syntax; + fetch_token (¤t_token, regexp, syntax | RE_CARET_ANCHORS_HERE); + tree = parse_reg_exp (regexp, preg, ¤t_token, syntax, 0, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + eor = create_tree (dfa, NULL, NULL, END_OF_RE); + if (tree != NULL) + root = create_tree (dfa, tree, eor, CONCAT); + else + root = eor; + if (__glibc_unlikely (eor == NULL || root == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + return root; +} + +/* This function build the following tree, from regular expression + <branch1>|<branch2>: + ALT + / \ + / \ + <branch1> <branch2> + + ALT means alternative, which represents the operator '|'. */ + +static bin_tree_t * +parse_reg_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_t *tree, *branch = NULL; + bitset_word_t initial_bkref_map = dfa->completed_bkref_map; + tree = parse_branch (regexp, preg, token, syntax, nest, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + + while (token->type == OP_ALT) + { + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); + if (token->type != OP_ALT && token->type != END_OF_RE + && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) + { + bitset_word_t accumulated_bkref_map = dfa->completed_bkref_map; + dfa->completed_bkref_map = initial_bkref_map; + branch = parse_branch (regexp, preg, token, syntax, nest, err); + if (__glibc_unlikely (*err != REG_NOERROR && branch == NULL)) + { + if (tree != NULL) + postorder (tree, free_tree, NULL); + return NULL; + } + dfa->completed_bkref_map |= accumulated_bkref_map; + } + else + branch = NULL; + tree = create_tree (dfa, tree, branch, OP_ALT); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + } + return tree; +} + +/* This function build the following tree, from regular expression + <exp1><exp2>: + CAT + / \ + / \ + <exp1> <exp2> + + CAT means concatenation. */ + +static bin_tree_t * +parse_branch (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + bin_tree_t *tree, *expr; + re_dfa_t *dfa = preg->buffer; + tree = parse_expression (regexp, preg, token, syntax, nest, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + + while (token->type != OP_ALT && token->type != END_OF_RE + && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) + { + expr = parse_expression (regexp, preg, token, syntax, nest, err); + if (__glibc_unlikely (*err != REG_NOERROR && expr == NULL)) + { + if (tree != NULL) + postorder (tree, free_tree, NULL); + return NULL; + } + if (tree != NULL && expr != NULL) + { + bin_tree_t *newtree = create_tree (dfa, tree, expr, CONCAT); + if (newtree == NULL) + { + postorder (expr, free_tree, NULL); + postorder (tree, free_tree, NULL); + *err = REG_ESPACE; + return NULL; + } + tree = newtree; + } + else if (tree == NULL) + tree = expr; + /* Otherwise expr == NULL, we don't need to create new tree. */ + } + return tree; +} + +/* This function build the following tree, from regular expression a*: + * + | + a +*/ + +static bin_tree_t * +parse_expression (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_t *tree; + switch (token->type) + { + case CHARACTER: + tree = create_token_tree (dfa, NULL, NULL, token); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + while (!re_string_eoi (regexp) + && !re_string_first_byte (regexp, re_string_cur_idx (regexp))) + { + bin_tree_t *mbc_remain; + fetch_token (token, regexp, syntax); + mbc_remain = create_token_tree (dfa, NULL, NULL, token); + tree = create_tree (dfa, tree, mbc_remain, CONCAT); + if (__glibc_unlikely (mbc_remain == NULL || tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + } + } +#endif + break; + + case OP_OPEN_SUBEXP: + tree = parse_sub_exp (regexp, preg, token, syntax, nest + 1, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + break; + + case OP_OPEN_BRACKET: + tree = parse_bracket_exp (regexp, dfa, token, syntax, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + break; + + case OP_BACK_REF: + if (!__glibc_likely (dfa->completed_bkref_map & (1 << token->opr.idx))) + { + *err = REG_ESUBREG; + return NULL; + } + dfa->used_bkref_map |= 1 << token->opr.idx; + tree = create_token_tree (dfa, NULL, NULL, token); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + ++dfa->nbackref; + dfa->has_mb_node = 1; + break; + + case OP_OPEN_DUP_NUM: + if (syntax & RE_CONTEXT_INVALID_DUP) + { + *err = REG_BADRPT; + return NULL; + } + FALLTHROUGH; + case OP_DUP_ASTERISK: + case OP_DUP_PLUS: + case OP_DUP_QUESTION: + if (syntax & RE_CONTEXT_INVALID_OPS) + { + *err = REG_BADRPT; + return NULL; + } + else if (syntax & RE_CONTEXT_INDEP_OPS) + { + fetch_token (token, regexp, syntax); + return parse_expression (regexp, preg, token, syntax, nest, err); + } + FALLTHROUGH; + case OP_CLOSE_SUBEXP: + if ((token->type == OP_CLOSE_SUBEXP) + && !(syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)) + { + *err = REG_ERPAREN; + return NULL; + } + FALLTHROUGH; + case OP_CLOSE_DUP_NUM: + /* We treat it as a normal character. */ + + /* Then we can these characters as normal characters. */ + token->type = CHARACTER; + /* mb_partial and word_char bits should be initialized already + by peek_token. */ + tree = create_token_tree (dfa, NULL, NULL, token); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + break; + + case ANCHOR: + if ((token->opr.ctx_type + & (WORD_DELIM | NOT_WORD_DELIM | WORD_FIRST | WORD_LAST)) + && dfa->word_ops_used == 0) + init_word_char (dfa); + if (token->opr.ctx_type == WORD_DELIM + || token->opr.ctx_type == NOT_WORD_DELIM) + { + bin_tree_t *tree_first, *tree_last; + if (token->opr.ctx_type == WORD_DELIM) + { + token->opr.ctx_type = WORD_FIRST; + tree_first = create_token_tree (dfa, NULL, NULL, token); + token->opr.ctx_type = WORD_LAST; + } + else + { + token->opr.ctx_type = INSIDE_WORD; + tree_first = create_token_tree (dfa, NULL, NULL, token); + token->opr.ctx_type = INSIDE_NOTWORD; + } + tree_last = create_token_tree (dfa, NULL, NULL, token); + tree = create_tree (dfa, tree_first, tree_last, OP_ALT); + if (__glibc_unlikely (tree_first == NULL || tree_last == NULL + || tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + } + else + { + tree = create_token_tree (dfa, NULL, NULL, token); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + } + /* We must return here, since ANCHORs can't be followed + by repetition operators. + eg. RE"^*" is invalid or "<ANCHOR(^)><CHAR(*)>", + it must not be "<ANCHOR(^)><REPEAT(*)>". */ + fetch_token (token, regexp, syntax); + return tree; + + case OP_PERIOD: + tree = create_token_tree (dfa, NULL, NULL, token); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + if (dfa->mb_cur_max > 1) + dfa->has_mb_node = 1; + break; + + case OP_WORD: + case OP_NOTWORD: + tree = build_charclass_op (dfa, regexp->trans, + "alnum", + "_", + token->type == OP_NOTWORD, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + break; + + case OP_SPACE: + case OP_NOTSPACE: + tree = build_charclass_op (dfa, regexp->trans, + "space", + "", + token->type == OP_NOTSPACE, err); + if (__glibc_unlikely (*err != REG_NOERROR && tree == NULL)) + return NULL; + break; + + case OP_ALT: + case END_OF_RE: + return NULL; + + case BACK_SLASH: + *err = REG_EESCAPE; + return NULL; + + default: + /* Must not happen? */ +#ifdef DEBUG + assert (0); +#endif + return NULL; + } + fetch_token (token, regexp, syntax); + + while (token->type == OP_DUP_ASTERISK || token->type == OP_DUP_PLUS + || token->type == OP_DUP_QUESTION || token->type == OP_OPEN_DUP_NUM) + { + bin_tree_t *dup_tree = parse_dup_op (tree, regexp, dfa, token, + syntax, err); + if (__glibc_unlikely (*err != REG_NOERROR && dup_tree == NULL)) + { + if (tree != NULL) + postorder (tree, free_tree, NULL); + return NULL; + } + tree = dup_tree; + /* In BRE consecutive duplications are not allowed. */ + if ((syntax & RE_CONTEXT_INVALID_DUP) + && (token->type == OP_DUP_ASTERISK + || token->type == OP_OPEN_DUP_NUM)) + { + if (tree != NULL) + postorder (tree, free_tree, NULL); + *err = REG_BADRPT; + return NULL; + } + } + + return tree; +} + +/* This function build the following tree, from regular expression + (<reg_exp>): + SUBEXP + | + <reg_exp> +*/ + +static bin_tree_t * +parse_sub_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = preg->buffer; + bin_tree_t *tree; + size_t cur_nsub; + cur_nsub = preg->re_nsub++; + + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); + + /* The subexpression may be a null string. */ + if (token->type == OP_CLOSE_SUBEXP) + tree = NULL; + else + { + tree = parse_reg_exp (regexp, preg, token, syntax, nest, err); + if (__glibc_unlikely (*err == REG_NOERROR + && token->type != OP_CLOSE_SUBEXP)) + { + if (tree != NULL) + postorder (tree, free_tree, NULL); + *err = REG_EPAREN; + } + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + } + + if (cur_nsub <= '9' - '1') + dfa->completed_bkref_map |= 1 << cur_nsub; + + tree = create_tree (dfa, tree, NULL, SUBEXP); + if (__glibc_unlikely (tree == NULL)) + { + *err = REG_ESPACE; + return NULL; + } + tree->token.opr.idx = cur_nsub; + return tree; +} + +/* This function parse repetition operators like "*", "+", "{1,3}" etc. */ + +static bin_tree_t * +parse_dup_op (bin_tree_t *elem, re_string_t *regexp, re_dfa_t *dfa, + re_token_t *token, reg_syntax_t syntax, reg_errcode_t *err) +{ + bin_tree_t *tree = NULL, *old_tree = NULL; + Idx i, start, end, start_idx = re_string_cur_idx (regexp); + re_token_t start_token = *token; + + if (token->type == OP_OPEN_DUP_NUM) + { + end = 0; + start = fetch_number (regexp, token, syntax); + if (start == -1) + { + if (token->type == CHARACTER && token->opr.c == ',') + start = 0; /* We treat "{,m}" as "{0,m}". */ + else + { + *err = REG_BADBR; /* <re>{} is invalid. */ + return NULL; + } + } + if (__glibc_likely (start != -2)) + { + /* We treat "{n}" as "{n,n}". */ + end = ((token->type == OP_CLOSE_DUP_NUM) ? start + : ((token->type == CHARACTER && token->opr.c == ',') + ? fetch_number (regexp, token, syntax) : -2)); + } + if (__glibc_unlikely (start == -2 || end == -2)) + { + /* Invalid sequence. */ + if (__glibc_unlikely (!(syntax & RE_INVALID_INTERVAL_ORD))) + { + if (token->type == END_OF_RE) + *err = REG_EBRACE; + else + *err = REG_BADBR; + + return NULL; + } + + /* If the syntax bit is set, rollback. */ + re_string_set_index (regexp, start_idx); + *token = start_token; + token->type = CHARACTER; + /* mb_partial and word_char bits should be already initialized by + peek_token. */ + return elem; + } + + if (__glibc_unlikely ((end != -1 && start > end) + || token->type != OP_CLOSE_DUP_NUM)) + { + /* First number greater than second. */ + *err = REG_BADBR; + return NULL; + } + + if (__glibc_unlikely (RE_DUP_MAX < (end == -1 ? start : end))) + { + *err = REG_ESIZE; + return NULL; + } + } + else + { + start = (token->type == OP_DUP_PLUS) ? 1 : 0; + end = (token->type == OP_DUP_QUESTION) ? 1 : -1; + } + + fetch_token (token, regexp, syntax); + + if (__glibc_unlikely (elem == NULL)) + return NULL; + if (__glibc_unlikely (start == 0 && end == 0)) + { + postorder (elem, free_tree, NULL); + return NULL; + } + + /* Extract "<re>{n,m}" to "<re><re>...<re><re>{0,<m-n>}". */ + if (__glibc_unlikely (start > 0)) + { + tree = elem; + for (i = 2; i <= start; ++i) + { + elem = duplicate_tree (elem, dfa); + tree = create_tree (dfa, tree, elem, CONCAT); + if (__glibc_unlikely (elem == NULL || tree == NULL)) + goto parse_dup_op_espace; + } + + if (start == end) + return tree; + + /* Duplicate ELEM before it is marked optional. */ + elem = duplicate_tree (elem, dfa); + if (__glibc_unlikely (elem == NULL)) + goto parse_dup_op_espace; + old_tree = tree; + } + else + old_tree = NULL; + + if (elem->token.type == SUBEXP) + { + uintptr_t subidx = elem->token.opr.idx; + postorder (elem, mark_opt_subexp, (void *) subidx); + } + + tree = create_tree (dfa, elem, NULL, + (end == -1 ? OP_DUP_ASTERISK : OP_ALT)); + if (__glibc_unlikely (tree == NULL)) + goto parse_dup_op_espace; + + /* This loop is actually executed only when end != -1, + to rewrite <re>{0,n} as (<re>(<re>...<re>?)?)?... We have + already created the start+1-th copy. */ + if (TYPE_SIGNED (Idx) || end != -1) + for (i = start + 2; i <= end; ++i) + { + elem = duplicate_tree (elem, dfa); + tree = create_tree (dfa, tree, elem, CONCAT); + if (__glibc_unlikely (elem == NULL || tree == NULL)) + goto parse_dup_op_espace; + + tree = create_tree (dfa, tree, NULL, OP_ALT); + if (__glibc_unlikely (tree == NULL)) + goto parse_dup_op_espace; + } + + if (old_tree) + tree = create_tree (dfa, old_tree, tree, CONCAT); + + return tree; + + parse_dup_op_espace: + *err = REG_ESPACE; + return NULL; +} + +/* Size of the names for collating symbol/equivalence_class/character_class. + I'm not sure, but maybe enough. */ +#define BRACKET_NAME_BUF_SIZE 32 + +#ifndef _LIBC + +# ifdef RE_ENABLE_I18N +/* Convert the byte B to the corresponding wide character. In a + unibyte locale, treat B as itself. In a multibyte locale, return + WEOF if B is an encoding error. */ +static wint_t +parse_byte (unsigned char b, re_charset_t *mbcset) +{ + return mbcset == NULL ? b : __btowc (b); +} +# endif + + /* Local function for parse_bracket_exp only used in case of NOT _LIBC. + Build the range expression which starts from START_ELEM, and ends + at END_ELEM. The result are written to MBCSET and SBCSET. + RANGE_ALLOC is the allocated size of mbcset->range_starts, and + mbcset->range_ends, is a pointer argument since we may + update it. */ + +static reg_errcode_t +# ifdef RE_ENABLE_I18N +build_range_exp (const reg_syntax_t syntax, + bitset_t sbcset, + re_charset_t *mbcset, + Idx *range_alloc, + const bracket_elem_t *start_elem, + const bracket_elem_t *end_elem) +# else /* not RE_ENABLE_I18N */ +build_range_exp (const reg_syntax_t syntax, + bitset_t sbcset, + const bracket_elem_t *start_elem, + const bracket_elem_t *end_elem) +# endif /* not RE_ENABLE_I18N */ +{ + unsigned int start_ch, end_ch; + /* Equivalence Classes and Character Classes can't be a range start/end. */ + if (__glibc_unlikely (start_elem->type == EQUIV_CLASS + || start_elem->type == CHAR_CLASS + || end_elem->type == EQUIV_CLASS + || end_elem->type == CHAR_CLASS)) + return REG_ERANGE; + + /* We can handle no multi character collating elements without libc + support. */ + if (__glibc_unlikely ((start_elem->type == COLL_SYM + && strlen ((char *) start_elem->opr.name) > 1) + || (end_elem->type == COLL_SYM + && strlen ((char *) end_elem->opr.name) > 1))) + return REG_ECOLLATE; + +# ifdef RE_ENABLE_I18N + { + wchar_t wc; + wint_t start_wc; + wint_t end_wc; + + start_ch = ((start_elem->type == SB_CHAR) ? start_elem->opr.ch + : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] + : 0)); + end_ch = ((end_elem->type == SB_CHAR) ? end_elem->opr.ch + : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] + : 0)); + start_wc = ((start_elem->type == SB_CHAR || start_elem->type == COLL_SYM) + ? parse_byte (start_ch, mbcset) : start_elem->opr.wch); + end_wc = ((end_elem->type == SB_CHAR || end_elem->type == COLL_SYM) + ? parse_byte (end_ch, mbcset) : end_elem->opr.wch); + if (start_wc == WEOF || end_wc == WEOF) + return REG_ECOLLATE; + else if (__glibc_unlikely ((syntax & RE_NO_EMPTY_RANGES) + && start_wc > end_wc)) + return REG_ERANGE; + + /* Got valid collation sequence values, add them as a new entry. + However, for !_LIBC we have no collation elements: if the + character set is single byte, the single byte character set + that we build below suffices. parse_bracket_exp passes + no MBCSET if dfa->mb_cur_max == 1. */ + if (mbcset) + { + /* Check the space of the arrays. */ + if (__glibc_unlikely (*range_alloc == mbcset->nranges)) + { + /* There is not enough space, need realloc. */ + wchar_t *new_array_start, *new_array_end; + Idx new_nranges; + + /* +1 in case of mbcset->nranges is 0. */ + new_nranges = 2 * mbcset->nranges + 1; + /* Use realloc since mbcset->range_starts and mbcset->range_ends + are NULL if *range_alloc == 0. */ + new_array_start = re_realloc (mbcset->range_starts, wchar_t, + new_nranges); + new_array_end = re_realloc (mbcset->range_ends, wchar_t, + new_nranges); + + if (__glibc_unlikely (new_array_start == NULL + || new_array_end == NULL)) + { + re_free (new_array_start); + re_free (new_array_end); + return REG_ESPACE; + } + + mbcset->range_starts = new_array_start; + mbcset->range_ends = new_array_end; + *range_alloc = new_nranges; + } + + mbcset->range_starts[mbcset->nranges] = start_wc; + mbcset->range_ends[mbcset->nranges++] = end_wc; + } + + /* Build the table for single byte characters. */ + for (wc = 0; wc < SBC_MAX; ++wc) + { + if (start_wc <= wc && wc <= end_wc) + bitset_set (sbcset, wc); + } + } +# else /* not RE_ENABLE_I18N */ + { + unsigned int ch; + start_ch = ((start_elem->type == SB_CHAR ) ? start_elem->opr.ch + : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] + : 0)); + end_ch = ((end_elem->type == SB_CHAR ) ? end_elem->opr.ch + : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] + : 0)); + if (start_ch > end_ch) + return REG_ERANGE; + /* Build the table for single byte characters. */ + for (ch = 0; ch < SBC_MAX; ++ch) + if (start_ch <= ch && ch <= end_ch) + bitset_set (sbcset, ch); + } +# endif /* not RE_ENABLE_I18N */ + return REG_NOERROR; +} +#endif /* not _LIBC */ + +#ifndef _LIBC +/* Helper function for parse_bracket_exp only used in case of NOT _LIBC.. + Build the collating element which is represented by NAME. + The result are written to MBCSET and SBCSET. + COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a + pointer argument since we may update it. */ + +static reg_errcode_t +# ifdef RE_ENABLE_I18N +build_collating_symbol (bitset_t sbcset, re_charset_t *mbcset, + Idx *coll_sym_alloc, const unsigned char *name) +# else /* not RE_ENABLE_I18N */ +build_collating_symbol (bitset_t sbcset, const unsigned char *name) +# endif /* not RE_ENABLE_I18N */ +{ + size_t name_len = strlen ((const char *) name); + if (__glibc_unlikely (name_len != 1)) + return REG_ECOLLATE; + else + { + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } +} +#endif /* not _LIBC */ + +/* This function parse bracket expression like "[abc]", "[a-c]", + "[[.a-a.]]" etc. */ + +static bin_tree_t * +parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, re_token_t *token, + reg_syntax_t syntax, reg_errcode_t *err) +{ +#ifdef _LIBC + const unsigned char *collseqmb; + const char *collseqwc; + uint32_t nrules; + int32_t table_size; + const int32_t *symb_table; + const unsigned char *extra; + + /* Local function for parse_bracket_exp used in _LIBC environment. + Seek the collating symbol entry corresponding to NAME. + Return the index of the symbol in the SYMB_TABLE, + or -1 if not found. */ + + auto inline int32_t + __attribute__ ((always_inline)) + seek_collating_symbol_entry (const unsigned char *name, size_t name_len) + { + int32_t elem; + + for (elem = 0; elem < table_size; elem++) + if (symb_table[2 * elem] != 0) + { + int32_t idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element name. */ + idx += 1 + extra[idx]; + if (/* Compare the length of the name. */ + name_len == extra[idx] + /* Compare the name. */ + && memcmp (name, &extra[idx + 1], name_len) == 0) + /* Yep, this is the entry. */ + return elem; + } + return -1; + } + + /* Local function for parse_bracket_exp used in _LIBC environment. + Look up the collation sequence value of BR_ELEM. + Return the value if succeeded, UINT_MAX otherwise. */ + + auto inline unsigned int + __attribute__ ((always_inline)) + lookup_collation_sequence_value (bracket_elem_t *br_elem) + { + if (br_elem->type == SB_CHAR) + { + /* + if (MB_CUR_MAX == 1) + */ + if (nrules == 0) + return collseqmb[br_elem->opr.ch]; + else + { + wint_t wc = __btowc (br_elem->opr.ch); + return __collseq_table_lookup (collseqwc, wc); + } + } + else if (br_elem->type == MB_CHAR) + { + if (nrules != 0) + return __collseq_table_lookup (collseqwc, br_elem->opr.wch); + } + else if (br_elem->type == COLL_SYM) + { + size_t sym_name_len = strlen ((char *) br_elem->opr.name); + if (nrules != 0) + { + int32_t elem, idx; + elem = seek_collating_symbol_entry (br_elem->opr.name, + sym_name_len); + if (elem != -1) + { + /* We found the entry. */ + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element name. */ + idx += 1 + extra[idx]; + /* Skip the byte sequence of the collating element. */ + idx += 1 + extra[idx]; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + /* Skip the multibyte collation sequence value. */ + idx += sizeof (unsigned int); + /* Skip the wide char sequence of the collating element. */ + idx += sizeof (unsigned int) * + (1 + *(unsigned int *) (extra + idx)); + /* Return the collation sequence value. */ + return *(unsigned int *) (extra + idx); + } + else if (sym_name_len == 1) + { + /* No valid character. Match it as a single byte + character. */ + return collseqmb[br_elem->opr.name[0]]; + } + } + else if (sym_name_len == 1) + return collseqmb[br_elem->opr.name[0]]; + } + return UINT_MAX; + } + + /* Local function for parse_bracket_exp used in _LIBC environment. + Build the range expression which starts from START_ELEM, and ends + at END_ELEM. The result are written to MBCSET and SBCSET. + RANGE_ALLOC is the allocated size of mbcset->range_starts, and + mbcset->range_ends, is a pointer argument since we may + update it. */ + + auto inline reg_errcode_t + __attribute__ ((always_inline)) + build_range_exp (bitset_t sbcset, re_charset_t *mbcset, int *range_alloc, + bracket_elem_t *start_elem, bracket_elem_t *end_elem) + { + unsigned int ch; + uint32_t start_collseq; + uint32_t end_collseq; + + /* Equivalence Classes and Character Classes can't be a range + start/end. */ + if (__glibc_unlikely (start_elem->type == EQUIV_CLASS + || start_elem->type == CHAR_CLASS + || end_elem->type == EQUIV_CLASS + || end_elem->type == CHAR_CLASS)) + return REG_ERANGE; + + /* FIXME: Implement rational ranges here, too. */ + start_collseq = lookup_collation_sequence_value (start_elem); + end_collseq = lookup_collation_sequence_value (end_elem); + /* Check start/end collation sequence values. */ + if (__glibc_unlikely (start_collseq == UINT_MAX + || end_collseq == UINT_MAX)) + return REG_ECOLLATE; + if (__glibc_unlikely ((syntax & RE_NO_EMPTY_RANGES) + && start_collseq > end_collseq)) + return REG_ERANGE; + + /* Got valid collation sequence values, add them as a new entry. + However, if we have no collation elements, and the character set + is single byte, the single byte character set that we + build below suffices. */ + if (nrules > 0 || dfa->mb_cur_max > 1) + { + /* Check the space of the arrays. */ + if (__glibc_unlikely (*range_alloc == mbcset->nranges)) + { + /* There is not enough space, need realloc. */ + uint32_t *new_array_start; + uint32_t *new_array_end; + Idx new_nranges; + + /* +1 in case of mbcset->nranges is 0. */ + new_nranges = 2 * mbcset->nranges + 1; + new_array_start = re_realloc (mbcset->range_starts, uint32_t, + new_nranges); + new_array_end = re_realloc (mbcset->range_ends, uint32_t, + new_nranges); + + if (__glibc_unlikely (new_array_start == NULL + || new_array_end == NULL)) + return REG_ESPACE; + + mbcset->range_starts = new_array_start; + mbcset->range_ends = new_array_end; + *range_alloc = new_nranges; + } + + mbcset->range_starts[mbcset->nranges] = start_collseq; + mbcset->range_ends[mbcset->nranges++] = end_collseq; + } + + /* Build the table for single byte characters. */ + for (ch = 0; ch < SBC_MAX; ch++) + { + uint32_t ch_collseq; + /* + if (MB_CUR_MAX == 1) + */ + if (nrules == 0) + ch_collseq = collseqmb[ch]; + else + ch_collseq = __collseq_table_lookup (collseqwc, __btowc (ch)); + if (start_collseq <= ch_collseq && ch_collseq <= end_collseq) + bitset_set (sbcset, ch); + } + return REG_NOERROR; + } + + /* Local function for parse_bracket_exp used in _LIBC environment. + Build the collating element which is represented by NAME. + The result are written to MBCSET and SBCSET. + COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a + pointer argument since we may update it. */ + + auto inline reg_errcode_t + __attribute__ ((always_inline)) + build_collating_symbol (bitset_t sbcset, re_charset_t *mbcset, + Idx *coll_sym_alloc, const unsigned char *name) + { + int32_t elem, idx; + size_t name_len = strlen ((const char *) name); + if (nrules != 0) + { + elem = seek_collating_symbol_entry (name, name_len); + if (elem != -1) + { + /* We found the entry. */ + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element name. */ + idx += 1 + extra[idx]; + } + else if (name_len == 1) + { + /* No valid character, treat it as a normal + character. */ + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } + else + return REG_ECOLLATE; + + /* Got valid collation sequence, add it as a new entry. */ + /* Check the space of the arrays. */ + if (__glibc_unlikely (*coll_sym_alloc == mbcset->ncoll_syms)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->ncoll_syms is 0. */ + Idx new_coll_sym_alloc = 2 * mbcset->ncoll_syms + 1; + /* Use realloc since mbcset->coll_syms is NULL + if *alloc == 0. */ + int32_t *new_coll_syms = re_realloc (mbcset->coll_syms, int32_t, + new_coll_sym_alloc); + if (__glibc_unlikely (new_coll_syms == NULL)) + return REG_ESPACE; + mbcset->coll_syms = new_coll_syms; + *coll_sym_alloc = new_coll_sym_alloc; + } + mbcset->coll_syms[mbcset->ncoll_syms++] = idx; + return REG_NOERROR; + } + else + { + if (__glibc_unlikely (name_len != 1)) + return REG_ECOLLATE; + else + { + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } + } + } +#endif + + re_token_t br_token; + re_bitset_ptr_t sbcset; +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; + Idx coll_sym_alloc = 0, range_alloc = 0, mbchar_alloc = 0; + Idx equiv_class_alloc = 0, char_class_alloc = 0; +#endif /* not RE_ENABLE_I18N */ + bool non_match = false; + bin_tree_t *work_tree; + int token_len; + bool first_round = true; +#ifdef _LIBC + collseqmb = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB); + nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules) + { + /* + if (MB_CUR_MAX > 1) + */ + collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC); + table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB); + symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_SYMB_TABLEMB); + extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_SYMB_EXTRAMB); + } +#endif + sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); +#ifdef RE_ENABLE_I18N + mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); +#endif /* RE_ENABLE_I18N */ +#ifdef RE_ENABLE_I18N + if (__glibc_unlikely (sbcset == NULL || mbcset == NULL)) +#else + if (__glibc_unlikely (sbcset == NULL)) +#endif /* RE_ENABLE_I18N */ + { + re_free (sbcset); +#ifdef RE_ENABLE_I18N + re_free (mbcset); +#endif + *err = REG_ESPACE; + return NULL; + } + + token_len = peek_token_bracket (token, regexp, syntax); + if (__glibc_unlikely (token->type == END_OF_RE)) + { + *err = REG_BADPAT; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_NON_MATCH_LIST) + { +#ifdef RE_ENABLE_I18N + mbcset->non_match = 1; +#endif /* not RE_ENABLE_I18N */ + non_match = true; + if (syntax & RE_HAT_LISTS_NOT_NEWLINE) + bitset_set (sbcset, '\n'); + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + token_len = peek_token_bracket (token, regexp, syntax); + if (__glibc_unlikely (token->type == END_OF_RE)) + { + *err = REG_BADPAT; + goto parse_bracket_exp_free_return; + } + } + + /* We treat the first ']' as a normal character. */ + if (token->type == OP_CLOSE_BRACKET) + token->type = CHARACTER; + + while (1) + { + bracket_elem_t start_elem, end_elem; + unsigned char start_name_buf[BRACKET_NAME_BUF_SIZE]; + unsigned char end_name_buf[BRACKET_NAME_BUF_SIZE]; + reg_errcode_t ret; + int token_len2 = 0; + bool is_range_exp = false; + re_token_t token2; + + start_elem.opr.name = start_name_buf; + start_elem.type = COLL_SYM; + ret = parse_bracket_element (&start_elem, regexp, token, token_len, dfa, + syntax, first_round); + if (__glibc_unlikely (ret != REG_NOERROR)) + { + *err = ret; + goto parse_bracket_exp_free_return; + } + first_round = false; + + /* Get information about the next token. We need it in any case. */ + token_len = peek_token_bracket (token, regexp, syntax); + + /* Do not check for ranges if we know they are not allowed. */ + if (start_elem.type != CHAR_CLASS && start_elem.type != EQUIV_CLASS) + { + if (__glibc_unlikely (token->type == END_OF_RE)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_CHARSET_RANGE) + { + re_string_skip_bytes (regexp, token_len); /* Skip '-'. */ + token_len2 = peek_token_bracket (&token2, regexp, syntax); + if (__glibc_unlikely (token2.type == END_OF_RE)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token2.type == OP_CLOSE_BRACKET) + { + /* We treat the last '-' as a normal character. */ + re_string_skip_bytes (regexp, -token_len); + token->type = CHARACTER; + } + else + is_range_exp = true; + } + } + + if (is_range_exp == true) + { + end_elem.opr.name = end_name_buf; + end_elem.type = COLL_SYM; + ret = parse_bracket_element (&end_elem, regexp, &token2, token_len2, + dfa, syntax, true); + if (__glibc_unlikely (ret != REG_NOERROR)) + { + *err = ret; + goto parse_bracket_exp_free_return; + } + + token_len = peek_token_bracket (token, regexp, syntax); + +#ifdef _LIBC + *err = build_range_exp (sbcset, mbcset, &range_alloc, + &start_elem, &end_elem); +#else +# ifdef RE_ENABLE_I18N + *err = build_range_exp (syntax, sbcset, + dfa->mb_cur_max > 1 ? mbcset : NULL, + &range_alloc, &start_elem, &end_elem); +# else + *err = build_range_exp (syntax, sbcset, &start_elem, &end_elem); +# endif +#endif /* RE_ENABLE_I18N */ + if (__glibc_unlikely (*err != REG_NOERROR)) + goto parse_bracket_exp_free_return; + } + else + { + switch (start_elem.type) + { + case SB_CHAR: + bitset_set (sbcset, start_elem.opr.ch); + break; +#ifdef RE_ENABLE_I18N + case MB_CHAR: + /* Check whether the array has enough space. */ + if (__glibc_unlikely (mbchar_alloc == mbcset->nmbchars)) + { + wchar_t *new_mbchars; + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nmbchars is 0. */ + mbchar_alloc = 2 * mbcset->nmbchars + 1; + /* Use realloc since array is NULL if *alloc == 0. */ + new_mbchars = re_realloc (mbcset->mbchars, wchar_t, + mbchar_alloc); + if (__glibc_unlikely (new_mbchars == NULL)) + goto parse_bracket_exp_espace; + mbcset->mbchars = new_mbchars; + } + mbcset->mbchars[mbcset->nmbchars++] = start_elem.opr.wch; + break; +#endif /* RE_ENABLE_I18N */ + case EQUIV_CLASS: + *err = build_equiv_class (sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &equiv_class_alloc, +#endif /* RE_ENABLE_I18N */ + start_elem.opr.name); + if (__glibc_unlikely (*err != REG_NOERROR)) + goto parse_bracket_exp_free_return; + break; + case COLL_SYM: + *err = build_collating_symbol (sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &coll_sym_alloc, +#endif /* RE_ENABLE_I18N */ + start_elem.opr.name); + if (__glibc_unlikely (*err != REG_NOERROR)) + goto parse_bracket_exp_free_return; + break; + case CHAR_CLASS: + *err = build_charclass (regexp->trans, sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &char_class_alloc, +#endif /* RE_ENABLE_I18N */ + (const char *) start_elem.opr.name, + syntax); + if (__glibc_unlikely (*err != REG_NOERROR)) + goto parse_bracket_exp_free_return; + break; + default: + assert (0); + break; + } + } + if (__glibc_unlikely (token->type == END_OF_RE)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_CLOSE_BRACKET) + break; + } + + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + + /* If it is non-matching list. */ + if (non_match) + bitset_not (sbcset); + +#ifdef RE_ENABLE_I18N + /* Ensure only single byte characters are set. */ + if (dfa->mb_cur_max > 1) + bitset_mask (sbcset, dfa->sb_char); + + if (mbcset->nmbchars || mbcset->ncoll_syms || mbcset->nequiv_classes + || mbcset->nranges || (dfa->mb_cur_max > 1 && (mbcset->nchar_classes + || mbcset->non_match))) + { + bin_tree_t *mbc_tree; + int sbc_idx; + /* Build a tree for complex bracket. */ + dfa->has_mb_node = 1; + br_token.type = COMPLEX_BRACKET; + br_token.opr.mbcset = mbcset; + mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (__glibc_unlikely (mbc_tree == NULL)) + goto parse_bracket_exp_espace; + for (sbc_idx = 0; sbc_idx < BITSET_WORDS; ++sbc_idx) + if (sbcset[sbc_idx]) + break; + /* If there are no bits set in sbcset, there is no point + of having both SIMPLE_BRACKET and COMPLEX_BRACKET. */ + if (sbc_idx < BITSET_WORDS) + { + /* Build a tree for simple bracket. */ + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + work_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (__glibc_unlikely (work_tree == NULL)) + goto parse_bracket_exp_espace; + + /* Then join them by ALT node. */ + work_tree = create_tree (dfa, work_tree, mbc_tree, OP_ALT); + if (__glibc_unlikely (work_tree == NULL)) + goto parse_bracket_exp_espace; + } + else + { + re_free (sbcset); + work_tree = mbc_tree; + } + } + else +#endif /* not RE_ENABLE_I18N */ + { +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif + /* Build a tree for simple bracket. */ + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + work_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (__glibc_unlikely (work_tree == NULL)) + goto parse_bracket_exp_espace; + } + return work_tree; + + parse_bracket_exp_espace: + *err = REG_ESPACE; + parse_bracket_exp_free_return: + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + return NULL; +} + +/* Parse an element in the bracket expression. */ + +static reg_errcode_t +parse_bracket_element (bracket_elem_t *elem, re_string_t *regexp, + re_token_t *token, int token_len, re_dfa_t *dfa, + reg_syntax_t syntax, bool accept_hyphen) +{ +#ifdef RE_ENABLE_I18N + int cur_char_size; + cur_char_size = re_string_char_size_at (regexp, re_string_cur_idx (regexp)); + if (cur_char_size > 1) + { + elem->type = MB_CHAR; + elem->opr.wch = re_string_wchar_at (regexp, re_string_cur_idx (regexp)); + re_string_skip_bytes (regexp, cur_char_size); + return REG_NOERROR; + } +#endif /* RE_ENABLE_I18N */ + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + if (token->type == OP_OPEN_COLL_ELEM || token->type == OP_OPEN_CHAR_CLASS + || token->type == OP_OPEN_EQUIV_CLASS) + return parse_bracket_symbol (elem, regexp, token); + if (__glibc_unlikely (token->type == OP_CHARSET_RANGE) && !accept_hyphen) + { + /* A '-' must only appear as anything but a range indicator before + the closing bracket. Everything else is an error. */ + re_token_t token2; + (void) peek_token_bracket (&token2, regexp, syntax); + if (token2.type != OP_CLOSE_BRACKET) + /* The actual error value is not standardized since this whole + case is undefined. But ERANGE makes good sense. */ + return REG_ERANGE; + } + elem->type = SB_CHAR; + elem->opr.ch = token->opr.c; + return REG_NOERROR; +} + +/* Parse a bracket symbol in the bracket expression. Bracket symbols are + such as [:<character_class>:], [.<collating_element>.], and + [=<equivalent_class>=]. */ + +static reg_errcode_t +parse_bracket_symbol (bracket_elem_t *elem, re_string_t *regexp, + re_token_t *token) +{ + unsigned char ch, delim = token->opr.c; + int i = 0; + if (re_string_eoi(regexp)) + return REG_EBRACK; + for (;; ++i) + { + if (i >= BRACKET_NAME_BUF_SIZE) + return REG_EBRACK; + if (token->type == OP_OPEN_CHAR_CLASS) + ch = re_string_fetch_byte_case (regexp); + else + ch = re_string_fetch_byte (regexp); + if (re_string_eoi(regexp)) + return REG_EBRACK; + if (ch == delim && re_string_peek_byte (regexp, 0) == ']') + break; + elem->opr.name[i] = ch; + } + re_string_skip_bytes (regexp, 1); + elem->opr.name[i] = '\0'; + switch (token->type) + { + case OP_OPEN_COLL_ELEM: + elem->type = COLL_SYM; + break; + case OP_OPEN_EQUIV_CLASS: + elem->type = EQUIV_CLASS; + break; + case OP_OPEN_CHAR_CLASS: + elem->type = CHAR_CLASS; + break; + default: + break; + } + return REG_NOERROR; +} + + /* Helper function for parse_bracket_exp. + Build the equivalence class which is represented by NAME. + The result are written to MBCSET and SBCSET. + EQUIV_CLASS_ALLOC is the allocated size of mbcset->equiv_classes, + is a pointer argument since we may update it. */ + +static reg_errcode_t +#ifdef RE_ENABLE_I18N +build_equiv_class (bitset_t sbcset, re_charset_t *mbcset, + Idx *equiv_class_alloc, const unsigned char *name) +#else /* not RE_ENABLE_I18N */ +build_equiv_class (bitset_t sbcset, const unsigned char *name) +#endif /* not RE_ENABLE_I18N */ +{ +#ifdef _LIBC + uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules != 0) + { + const int32_t *table, *indirect; + const unsigned char *weights, *extra, *cp; + unsigned char char_buf[2]; + int32_t idx1, idx2; + unsigned int ch; + size_t len; + /* Calculate the index for equivalence class. */ + cp = name; + table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + weights = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_WEIGHTMB); + extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_INDIRECTMB); + idx1 = findidx (table, indirect, extra, &cp, -1); + if (__glibc_unlikely (idx1 == 0 || *cp != '\0')) + /* This isn't a valid character. */ + return REG_ECOLLATE; + + /* Build single byte matching table for this equivalence class. */ + len = weights[idx1 & 0xffffff]; + for (ch = 0; ch < SBC_MAX; ++ch) + { + char_buf[0] = ch; + cp = char_buf; + idx2 = findidx (table, indirect, extra, &cp, 1); +/* + idx2 = table[ch]; +*/ + if (idx2 == 0) + /* This isn't a valid character. */ + continue; + /* Compare only if the length matches and the collation rule + index is the same. */ + if (len == weights[idx2 & 0xffffff] && (idx1 >> 24) == (idx2 >> 24) + && memcmp (weights + (idx1 & 0xffffff) + 1, + weights + (idx2 & 0xffffff) + 1, len) == 0) + bitset_set (sbcset, ch); + } + /* Check whether the array has enough space. */ + if (__glibc_unlikely (*equiv_class_alloc == mbcset->nequiv_classes)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nequiv_classes is 0. */ + Idx new_equiv_class_alloc = 2 * mbcset->nequiv_classes + 1; + /* Use realloc since the array is NULL if *alloc == 0. */ + int32_t *new_equiv_classes = re_realloc (mbcset->equiv_classes, + int32_t, + new_equiv_class_alloc); + if (__glibc_unlikely (new_equiv_classes == NULL)) + return REG_ESPACE; + mbcset->equiv_classes = new_equiv_classes; + *equiv_class_alloc = new_equiv_class_alloc; + } + mbcset->equiv_classes[mbcset->nequiv_classes++] = idx1; + } + else +#endif /* _LIBC */ + { + if (__glibc_unlikely (strlen ((const char *) name) != 1)) + return REG_ECOLLATE; + bitset_set (sbcset, *name); + } + return REG_NOERROR; +} + + /* Helper function for parse_bracket_exp. + Build the character class which is represented by NAME. + The result are written to MBCSET and SBCSET. + CHAR_CLASS_ALLOC is the allocated size of mbcset->char_classes, + is a pointer argument since we may update it. */ + +static reg_errcode_t +#ifdef RE_ENABLE_I18N +build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, + re_charset_t *mbcset, Idx *char_class_alloc, + const char *class_name, reg_syntax_t syntax) +#else /* not RE_ENABLE_I18N */ +build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, + const char *class_name, reg_syntax_t syntax) +#endif /* not RE_ENABLE_I18N */ +{ + int i; + const char *name = class_name; + + /* In case of REG_ICASE "upper" and "lower" match the both of + upper and lower cases. */ + if ((syntax & RE_ICASE) + && (strcmp (name, "upper") == 0 || strcmp (name, "lower") == 0)) + name = "alpha"; + +#ifdef RE_ENABLE_I18N + /* Check the space of the arrays. */ + if (__glibc_unlikely (*char_class_alloc == mbcset->nchar_classes)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nchar_classes is 0. */ + Idx new_char_class_alloc = 2 * mbcset->nchar_classes + 1; + /* Use realloc since array is NULL if *alloc == 0. */ + wctype_t *new_char_classes = re_realloc (mbcset->char_classes, wctype_t, + new_char_class_alloc); + if (__glibc_unlikely (new_char_classes == NULL)) + return REG_ESPACE; + mbcset->char_classes = new_char_classes; + *char_class_alloc = new_char_class_alloc; + } + mbcset->char_classes[mbcset->nchar_classes++] = __wctype (name); +#endif /* RE_ENABLE_I18N */ + +#define BUILD_CHARCLASS_LOOP(ctype_func) \ + do { \ + if (__glibc_unlikely (trans != NULL)) \ + { \ + for (i = 0; i < SBC_MAX; ++i) \ + if (ctype_func (i)) \ + bitset_set (sbcset, trans[i]); \ + } \ + else \ + { \ + for (i = 0; i < SBC_MAX; ++i) \ + if (ctype_func (i)) \ + bitset_set (sbcset, i); \ + } \ + } while (0) + + if (strcmp (name, "alnum") == 0) + BUILD_CHARCLASS_LOOP (isalnum); + else if (strcmp (name, "cntrl") == 0) + BUILD_CHARCLASS_LOOP (iscntrl); + else if (strcmp (name, "lower") == 0) + BUILD_CHARCLASS_LOOP (islower); + else if (strcmp (name, "space") == 0) + BUILD_CHARCLASS_LOOP (isspace); + else if (strcmp (name, "alpha") == 0) + BUILD_CHARCLASS_LOOP (isalpha); + else if (strcmp (name, "digit") == 0) + BUILD_CHARCLASS_LOOP (isdigit); + else if (strcmp (name, "print") == 0) + BUILD_CHARCLASS_LOOP (isprint); + else if (strcmp (name, "upper") == 0) + BUILD_CHARCLASS_LOOP (isupper); + else if (strcmp (name, "blank") == 0) + BUILD_CHARCLASS_LOOP (isblank); + else if (strcmp (name, "graph") == 0) + BUILD_CHARCLASS_LOOP (isgraph); + else if (strcmp (name, "punct") == 0) + BUILD_CHARCLASS_LOOP (ispunct); + else if (strcmp (name, "xdigit") == 0) + BUILD_CHARCLASS_LOOP (isxdigit); + else + return REG_ECTYPE; + + return REG_NOERROR; +} + +static bin_tree_t * +build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans, + const char *class_name, + const char *extra, bool non_match, + reg_errcode_t *err) +{ + re_bitset_ptr_t sbcset; +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; + Idx alloc = 0; +#endif /* not RE_ENABLE_I18N */ + reg_errcode_t ret; + re_token_t br_token; + bin_tree_t *tree; + + sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); + if (__glibc_unlikely (sbcset == NULL)) + { + *err = REG_ESPACE; + return NULL; + } +#ifdef RE_ENABLE_I18N + mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); + if (__glibc_unlikely (mbcset == NULL)) + { + re_free (sbcset); + *err = REG_ESPACE; + return NULL; + } + mbcset->non_match = non_match; +#endif /* RE_ENABLE_I18N */ + + /* We don't care the syntax in this case. */ + ret = build_charclass (trans, sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &alloc, +#endif /* RE_ENABLE_I18N */ + class_name, 0); + + if (__glibc_unlikely (ret != REG_NOERROR)) + { + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + *err = ret; + return NULL; + } + /* \w match '_' also. */ + for (; *extra; extra++) + bitset_set (sbcset, *extra); + + /* If it is non-matching list. */ + if (non_match) + bitset_not (sbcset); + +#ifdef RE_ENABLE_I18N + /* Ensure only single byte characters are set. */ + if (dfa->mb_cur_max > 1) + bitset_mask (sbcset, dfa->sb_char); +#endif + + /* Build a tree for simple bracket. */ +#if defined GCC_LINT || defined lint + memset (&br_token, 0, sizeof br_token); +#endif + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (__glibc_unlikely (tree == NULL)) + goto build_word_op_espace; + +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + bin_tree_t *mbc_tree; + /* Build a tree for complex bracket. */ + br_token.type = COMPLEX_BRACKET; + br_token.opr.mbcset = mbcset; + dfa->has_mb_node = 1; + mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (__glibc_unlikely (mbc_tree == NULL)) + goto build_word_op_espace; + /* Then join them by ALT node. */ + tree = create_tree (dfa, tree, mbc_tree, OP_ALT); + if (__glibc_likely (mbc_tree != NULL)) + return tree; + } + else + { + free_charset (mbcset); + return tree; + } +#else /* not RE_ENABLE_I18N */ + return tree; +#endif /* not RE_ENABLE_I18N */ + + build_word_op_espace: + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + *err = REG_ESPACE; + return NULL; +} + +/* This is intended for the expressions like "a{1,3}". + Fetch a number from 'input', and return the number. + Return -1 if the number field is empty like "{,1}". + Return RE_DUP_MAX + 1 if the number field is too large. + Return -2 if an error occurred. */ + +static Idx +fetch_number (re_string_t *input, re_token_t *token, reg_syntax_t syntax) +{ + Idx num = -1; + unsigned char c; + while (1) + { + fetch_token (token, input, syntax); + c = token->opr.c; + if (__glibc_unlikely (token->type == END_OF_RE)) + return -2; + if (token->type == OP_CLOSE_DUP_NUM || c == ',') + break; + num = ((token->type != CHARACTER || c < '0' || '9' < c || num == -2) + ? -2 + : num == -1 + ? c - '0' + : MIN (RE_DUP_MAX + 1, num * 10 + c - '0')); + } + return num; +} + +#ifdef RE_ENABLE_I18N +static void +free_charset (re_charset_t *cset) +{ + re_free (cset->mbchars); +# ifdef _LIBC + re_free (cset->coll_syms); + re_free (cset->equiv_classes); +# endif + re_free (cset->range_starts); + re_free (cset->range_ends); + re_free (cset->char_classes); + re_free (cset); +} +#endif /* RE_ENABLE_I18N */ + +/* Functions for binary tree operation. */ + +/* Create a tree node. */ + +static bin_tree_t * +create_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, + re_token_type_t type) +{ + re_token_t t; +#if defined GCC_LINT || defined lint + memset (&t, 0, sizeof t); +#endif + t.type = type; + return create_token_tree (dfa, left, right, &t); +} + +static bin_tree_t * +create_token_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, + const re_token_t *token) +{ + bin_tree_t *tree; + if (__glibc_unlikely (dfa->str_tree_storage_idx == BIN_TREE_STORAGE_SIZE)) + { + bin_tree_storage_t *storage = re_malloc (bin_tree_storage_t, 1); + + if (storage == NULL) + return NULL; + storage->next = dfa->str_tree_storage; + dfa->str_tree_storage = storage; + dfa->str_tree_storage_idx = 0; + } + tree = &dfa->str_tree_storage->data[dfa->str_tree_storage_idx++]; + + tree->parent = NULL; + tree->left = left; + tree->right = right; + tree->token = *token; + tree->token.duplicated = 0; + tree->token.opt_subexp = 0; + tree->first = NULL; + tree->next = NULL; + tree->node_idx = -1; + + if (left != NULL) + left->parent = tree; + if (right != NULL) + right->parent = tree; + return tree; +} + +/* Mark the tree SRC as an optional subexpression. + To be called from preorder or postorder. */ + +static reg_errcode_t +mark_opt_subexp (void *extra, bin_tree_t *node) +{ + Idx idx = (uintptr_t) extra; + if (node->token.type == SUBEXP && node->token.opr.idx == idx) + node->token.opt_subexp = 1; + + return REG_NOERROR; +} + +/* Free the allocated memory inside NODE. */ + +static void +free_token (re_token_t *node) +{ +#ifdef RE_ENABLE_I18N + if (node->type == COMPLEX_BRACKET && node->duplicated == 0) + free_charset (node->opr.mbcset); + else +#endif /* RE_ENABLE_I18N */ + if (node->type == SIMPLE_BRACKET && node->duplicated == 0) + re_free (node->opr.sbcset); +} + +/* Worker function for tree walking. Free the allocated memory inside NODE + and its children. */ + +static reg_errcode_t +free_tree (void *extra, bin_tree_t *node) +{ + free_token (&node->token); + return REG_NOERROR; +} + + +/* Duplicate the node SRC, and return new node. This is a preorder + visit similar to the one implemented by the generic visitor, but + we need more infrastructure to maintain two parallel trees --- so, + it's easier to duplicate. */ + +static bin_tree_t * +duplicate_tree (const bin_tree_t *root, re_dfa_t *dfa) +{ + const bin_tree_t *node; + bin_tree_t *dup_root; + bin_tree_t **p_new = &dup_root, *dup_node = root->parent; + + for (node = root; ; ) + { + /* Create a new tree and link it back to the current parent. */ + *p_new = create_token_tree (dfa, NULL, NULL, &node->token); + if (*p_new == NULL) + return NULL; + (*p_new)->parent = dup_node; + (*p_new)->token.duplicated = 1; + dup_node = *p_new; + + /* Go to the left node, or up and to the right. */ + if (node->left) + { + node = node->left; + p_new = &dup_node->left; + } + else + { + const bin_tree_t *prev = NULL; + while (node->right == prev || node->right == NULL) + { + prev = node; + node = node->parent; + dup_node = dup_node->parent; + if (!node) + return dup_root; + } + node = node->right; + p_new = &dup_node->right; + } + } +} diff --git a/lib/regex.c b/lib/regex.c new file mode 100644 index 00000000000..eab7a48b240 --- /dev/null +++ b/lib/regex.c @@ -0,0 +1,81 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#ifndef _LIBC +# include <libc-config.h> + +# if __GNUC_PREREQ (4, 6) +# pragma GCC diagnostic ignored "-Wsuggest-attribute=pure" +# endif +# if __GNUC_PREREQ (4, 3) +# pragma GCC diagnostic ignored "-Wold-style-definition" +# pragma GCC diagnostic ignored "-Wtype-limits" +# endif +#endif + +/* Make sure no one compiles this code with a C++ compiler. */ +#if defined __cplusplus && defined _LIBC +# error "This is C code, use a C compiler" +#endif + +#ifdef _LIBC +/* We have to keep the namespace clean. */ +# define regfree(preg) __regfree (preg) +# define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef) +# define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags) +# define regerror(errcode, preg, errbuf, errbuf_size) \ + __regerror(errcode, preg, errbuf, errbuf_size) +# define re_set_registers(bu, re, nu, st, en) \ + __re_set_registers (bu, re, nu, st, en) +# define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \ + __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) +# define re_match(bufp, string, size, pos, regs) \ + __re_match (bufp, string, size, pos, regs) +# define re_search(bufp, string, size, startpos, range, regs) \ + __re_search (bufp, string, size, startpos, range, regs) +# define re_compile_pattern(pattern, length, bufp) \ + __re_compile_pattern (pattern, length, bufp) +# define re_set_syntax(syntax) __re_set_syntax (syntax) +# define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \ + __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop) +# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp) + +# include "../locale/localeinfo.h" +#endif + +/* On some systems, limits.h sets RE_DUP_MAX to a lower value than + GNU regex allows. Include it before <regex.h>, which correctly + #undefs RE_DUP_MAX and sets it to the right value. */ +#include <limits.h> + +#include <regex.h> +#include "regex_internal.h" + +#include "regex_internal.c" +#include "regcomp.c" +#include "regexec.c" + +/* Binary backward compatibility. */ +#if _LIBC +# include <shlib-compat.h> +# if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3) +link_warning (re_max_failures, "the 're_max_failures' variable is obsolete and will go away.") +int re_max_failures = 2000; +# endif +#endif diff --git a/lib/regex.h b/lib/regex.h new file mode 100644 index 00000000000..77ac1a559c4 --- /dev/null +++ b/lib/regex.h @@ -0,0 +1,658 @@ +/* Definitions for data structures and routines for the regular + expression library. + Copyright (C) 1985, 1989-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#ifndef _REGEX_H +#define _REGEX_H 1 + +#include <sys/types.h> + +/* Allow the use in C++ code. */ +#ifdef __cplusplus +extern "C" { +#endif + +/* Define __USE_GNU to declare GNU extensions that violate the + POSIX name space rules. */ +#ifdef _GNU_SOURCE +# define __USE_GNU 1 +#endif + +#ifdef _REGEX_LARGE_OFFSETS + +/* Use types and values that are wide enough to represent signed and + unsigned byte offsets in memory. This currently works only when + the regex code is used outside of the GNU C library; it is not yet + supported within glibc itself, and glibc users should not define + _REGEX_LARGE_OFFSETS. */ + +/* The type of object sizes. */ +typedef size_t __re_size_t; + +/* The type of object sizes, in places where the traditional code + uses unsigned long int. */ +typedef size_t __re_long_size_t; + +#else + +/* The traditional GNU regex implementation mishandles strings longer + than INT_MAX. */ +typedef unsigned int __re_size_t; +typedef unsigned long int __re_long_size_t; + +#endif + +/* The following two types have to be signed and unsigned integer type + wide enough to hold a value of a pointer. For most ANSI compilers + ptrdiff_t and size_t should be likely OK. Still size of these two + types is 2 for Microsoft C. Ugh... */ +typedef long int s_reg_t; +typedef unsigned long int active_reg_t; + +/* The following bits are used to determine the regexp syntax we + recognize. The set/not-set meanings are chosen so that Emacs syntax + remains the value 0. The bits are given in alphabetical order, and + the definitions shifted by one from the previous bit; thus, when we + add or remove a bit, only one other definition need change. */ +typedef unsigned long int reg_syntax_t; + +#ifdef __USE_GNU +/* If this bit is not set, then \ inside a bracket expression is literal. + If set, then such a \ quotes the following character. */ +# define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) + +/* If this bit is not set, then + and ? are operators, and \+ and \? are + literals. + If set, then \+ and \? are operators and + and ? are literals. */ +# define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) + +/* If this bit is set, then character classes are supported. They are: + [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], + [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. + If not set, then character classes are not supported. */ +# define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) + +/* If this bit is set, then ^ and $ are always anchors (outside bracket + expressions, of course). + If this bit is not set, then it depends: + ^ is an anchor if it is at the beginning of a regular + expression or after an open-group or an alternation operator; + $ is an anchor if it is at the end of a regular expression, or + before a close-group or an alternation operator. + + This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because + POSIX draft 11.2 says that * etc. in leading positions is undefined. + We already implemented a previous draft which made those constructs + invalid, though, so we haven't changed the code back. */ +# define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) + +/* If this bit is set, then special characters are always special + regardless of where they are in the pattern. + If this bit is not set, then special characters are special only in + some contexts; otherwise they are ordinary. Specifically, + * + ? and intervals are only special when not after the beginning, + open-group, or alternation operator. */ +# define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) + +/* If this bit is set, then *, +, ?, and { cannot be first in an re or + immediately after an alternation or begin-group operator. */ +# define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) + +/* If this bit is set, then . matches newline. + If not set, then it doesn't. */ +# define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) + +/* If this bit is set, then . doesn't match NUL. + If not set, then it does. */ +# define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) + +/* If this bit is set, nonmatching lists [^...] do not match newline. + If not set, they do. */ +# define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) + +/* If this bit is set, either \{...\} or {...} defines an + interval, depending on RE_NO_BK_BRACES. + If not set, \{, \}, {, and } are literals. */ +# define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) + +/* If this bit is set, +, ? and | aren't recognized as operators. + If not set, they are. */ +# define RE_LIMITED_OPS (RE_INTERVALS << 1) + +/* If this bit is set, newline is an alternation operator. + If not set, newline is literal. */ +# define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) + +/* If this bit is set, then '{...}' defines an interval, and \{ and \} + are literals. + If not set, then '\{...\}' defines an interval. */ +# define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) + +/* If this bit is set, (...) defines a group, and \( and \) are literals. + If not set, \(...\) defines a group, and ( and ) are literals. */ +# define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) + +/* If this bit is set, then \<digit> matches <digit>. + If not set, then \<digit> is a back-reference. */ +# define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) + +/* If this bit is set, then | is an alternation operator, and \| is literal. + If not set, then \| is an alternation operator, and | is literal. */ +# define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) + +/* If this bit is set, then an ending range point collating higher + than the starting range point, as in [z-a], is invalid. + If not set, then when ending range point collates higher than the + starting range point, the range is ignored. */ +# define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) + +/* If this bit is set, then an unmatched ) is ordinary. + If not set, then an unmatched ) is invalid. */ +# define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) + +/* If this bit is set, succeed as soon as we match the whole pattern, + without further backtracking. */ +# define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) + +/* If this bit is set, do not process the GNU regex operators. + If not set, then the GNU regex operators are recognized. */ +# define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) + +/* If this bit is set, turn on internal regex debugging. + If not set, and debugging was on, turn it off. + This only works if regex.c is compiled -DDEBUG. + We define this bit always, so that all that's needed to turn on + debugging is to recompile regex.c; the calling code can always have + this bit set, and it won't affect anything in the normal case. */ +# define RE_DEBUG (RE_NO_GNU_OPS << 1) + +/* If this bit is set, a syntactically invalid interval is treated as + a string of ordinary characters. For example, the ERE 'a{1' is + treated as 'a\{1'. */ +# define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1) + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +# define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1) + +/* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only + for ^, because it is difficult to scan the regex backwards to find + whether ^ should be special. */ +# define RE_CARET_ANCHORS_HERE (RE_ICASE << 1) + +/* If this bit is set, then \{ cannot be first in a regex or + immediately after an alternation, open-group or \} operator. */ +# define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1) + +/* If this bit is set, then no_sub will be set to 1 during + re_compile_pattern. */ +# define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1) +#endif + +/* This global variable defines the particular regexp syntax to use (for + some interfaces). When a regexp is compiled, the syntax used is + stored in the pattern buffer, so changing this does not affect + already-compiled regexps. */ +extern reg_syntax_t re_syntax_options; + +#ifdef __USE_GNU +/* Define combinations of the above bits for the standard possibilities. + (The [[[ comments delimit what gets put into the Texinfo file, so + don't delete them!) */ +/* [[[begin syntaxes]]] */ +# define RE_SYNTAX_EMACS 0 + +# define RE_SYNTAX_AWK \ + (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ + | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CHAR_CLASSES \ + | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) + +# define RE_SYNTAX_GNU_AWK \ + ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ + | RE_INVALID_INTERVAL_ORD) \ + & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \ + | RE_CONTEXT_INVALID_OPS )) + +# define RE_SYNTAX_POSIX_AWK \ + (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ + | RE_INTERVALS | RE_NO_GNU_OPS \ + | RE_INVALID_INTERVAL_ORD) + +# define RE_SYNTAX_GREP \ + ((RE_SYNTAX_POSIX_BASIC | RE_NEWLINE_ALT) \ + & ~(RE_CONTEXT_INVALID_DUP | RE_DOT_NOT_NULL)) + +# define RE_SYNTAX_EGREP \ + ((RE_SYNTAX_POSIX_EXTENDED | RE_INVALID_INTERVAL_ORD | RE_NEWLINE_ALT) \ + & ~(RE_CONTEXT_INVALID_OPS | RE_DOT_NOT_NULL)) + +/* POSIX grep -E behavior is no longer incompatible with GNU. */ +# define RE_SYNTAX_POSIX_EGREP \ + RE_SYNTAX_EGREP + +/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ +# define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC + +# define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC + +/* Syntax bits common to both basic and extended POSIX regex syntax. */ +# define _RE_SYNTAX_POSIX_COMMON \ + (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ + | RE_INTERVALS | RE_NO_EMPTY_RANGES) + +# define RE_SYNTAX_POSIX_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP) + +/* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes + RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this + isn't minimal, since other operators, such as \`, aren't disabled. */ +# define RE_SYNTAX_POSIX_MINIMAL_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) + +# define RE_SYNTAX_POSIX_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ + | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD) + +/* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is + removed and RE_NO_BK_REFS is added. */ +# define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) +/* [[[end syntaxes]]] */ + +/* Maximum number of duplicates an interval can allow. POSIX-conforming + systems might define this in <limits.h>, but we want our + value, so remove any previous define. */ +# ifdef _REGEX_INCLUDE_LIMITS_H +# include <limits.h> +# endif +# ifdef RE_DUP_MAX +# undef RE_DUP_MAX +# endif + +/* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored + the counter as a 2-byte signed integer. This is no longer true, so + RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to + ((SIZE_MAX - 9) / 10) if _REGEX_LARGE_OFFSETS is defined. + However, there would be a huge performance problem if someone + actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains + its historical value. */ +# define RE_DUP_MAX (0x7fff) +#endif + + +/* POSIX 'cflags' bits (i.e., information for 'regcomp'). */ + +/* If this bit is set, then use extended regular expression syntax. + If not set, then use basic regular expression syntax. */ +#define REG_EXTENDED 1 + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +#define REG_ICASE (1 << 1) + +/* If this bit is set, then anchors do not match at newline + characters in the string. + If not set, then anchors do match at newlines. */ +#define REG_NEWLINE (1 << 2) + +/* If this bit is set, then report only success or fail in regexec. + If not set, then returns differ between not matching and errors. */ +#define REG_NOSUB (1 << 3) + + +/* POSIX 'eflags' bits (i.e., information for regexec). */ + +/* If this bit is set, then the beginning-of-line operator doesn't match + the beginning of the string (presumably because it's not the + beginning of a line). + If not set, then the beginning-of-line operator does match the + beginning of the string. */ +#define REG_NOTBOL 1 + +/* Like REG_NOTBOL, except for the end-of-line. */ +#define REG_NOTEOL (1 << 1) + +/* Use PMATCH[0] to delimit the start and end of the search in the + buffer. */ +#define REG_STARTEND (1 << 2) + + +/* If any error codes are removed, changed, or added, update the + '__re_error_msgid' table in regcomp.c. */ + +typedef enum +{ + _REG_ENOSYS = -1, /* This will never happen for this implementation. */ + _REG_NOERROR = 0, /* Success. */ + _REG_NOMATCH, /* Didn't find a match (for regexec). */ + + /* POSIX regcomp return error codes. (In the order listed in the + standard.) */ + _REG_BADPAT, /* Invalid pattern. */ + _REG_ECOLLATE, /* Invalid collating element. */ + _REG_ECTYPE, /* Invalid character class name. */ + _REG_EESCAPE, /* Trailing backslash. */ + _REG_ESUBREG, /* Invalid back reference. */ + _REG_EBRACK, /* Unmatched left bracket. */ + _REG_EPAREN, /* Parenthesis imbalance. */ + _REG_EBRACE, /* Unmatched \{. */ + _REG_BADBR, /* Invalid contents of \{\}. */ + _REG_ERANGE, /* Invalid range end. */ + _REG_ESPACE, /* Ran out of memory. */ + _REG_BADRPT, /* No preceding re for repetition op. */ + + /* Error codes we've added. */ + _REG_EEND, /* Premature end. */ + _REG_ESIZE, /* Too large (e.g., repeat count too large). */ + _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ +} reg_errcode_t; + +#if defined _XOPEN_SOURCE || defined __USE_XOPEN2K +# define REG_ENOSYS _REG_ENOSYS +#endif +#define REG_NOERROR _REG_NOERROR +#define REG_NOMATCH _REG_NOMATCH +#define REG_BADPAT _REG_BADPAT +#define REG_ECOLLATE _REG_ECOLLATE +#define REG_ECTYPE _REG_ECTYPE +#define REG_EESCAPE _REG_EESCAPE +#define REG_ESUBREG _REG_ESUBREG +#define REG_EBRACK _REG_EBRACK +#define REG_EPAREN _REG_EPAREN +#define REG_EBRACE _REG_EBRACE +#define REG_BADBR _REG_BADBR +#define REG_ERANGE _REG_ERANGE +#define REG_ESPACE _REG_ESPACE +#define REG_BADRPT _REG_BADRPT +#define REG_EEND _REG_EEND +#define REG_ESIZE _REG_ESIZE +#define REG_ERPAREN _REG_ERPAREN + +/* This data structure represents a compiled pattern. Before calling + the pattern compiler, the fields 'buffer', 'allocated', 'fastmap', + and 'translate' can be set. After the pattern has been compiled, + the fields 're_nsub', 'not_bol' and 'not_eol' are available. All + other fields are private to the regex routines. */ + +#ifndef RE_TRANSLATE_TYPE +# define __RE_TRANSLATE_TYPE unsigned char * +# ifdef __USE_GNU +# define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE +# endif +#endif + +#ifdef __USE_GNU +# define __REPB_PREFIX(name) name +#else +# define __REPB_PREFIX(name) __##name +#endif + +struct re_pattern_buffer +{ + /* Space that holds the compiled pattern. The type + 'struct re_dfa_t' is private and is not declared here. */ + struct re_dfa_t *__REPB_PREFIX(buffer); + + /* Number of bytes to which 'buffer' points. */ + __re_long_size_t __REPB_PREFIX(allocated); + + /* Number of bytes actually used in 'buffer'. */ + __re_long_size_t __REPB_PREFIX(used); + + /* Syntax setting with which the pattern was compiled. */ + reg_syntax_t __REPB_PREFIX(syntax); + + /* Pointer to a fastmap, if any, otherwise zero. re_search uses the + fastmap, if there is one, to skip over impossible starting points + for matches. */ + char *__REPB_PREFIX(fastmap); + + /* Either a translate table to apply to all characters before + comparing them, or zero for no translation. The translation is + applied to a pattern when it is compiled and to a string when it + is matched. */ + __RE_TRANSLATE_TYPE __REPB_PREFIX(translate); + + /* Number of subexpressions found by the compiler. */ + size_t re_nsub; + + /* Zero if this pattern cannot match the empty string, one else. + Well, in truth it's used only in 're_search_2', to see whether or + not we should use the fastmap, so we don't set this absolutely + perfectly; see 're_compile_fastmap' (the "duplicate" case). */ + unsigned __REPB_PREFIX(can_be_null) : 1; + + /* If REGS_UNALLOCATED, allocate space in the 'regs' structure + for 'max (RE_NREGS, re_nsub + 1)' groups. + If REGS_REALLOCATE, reallocate space if necessary. + If REGS_FIXED, use what's there. */ +#ifdef __USE_GNU +# define REGS_UNALLOCATED 0 +# define REGS_REALLOCATE 1 +# define REGS_FIXED 2 +#endif + unsigned __REPB_PREFIX(regs_allocated) : 2; + + /* Set to zero when 're_compile_pattern' compiles a pattern; set to + one by 're_compile_fastmap' if it updates the fastmap. */ + unsigned __REPB_PREFIX(fastmap_accurate) : 1; + + /* If set, 're_match_2' does not return information about + subexpressions. */ + unsigned __REPB_PREFIX(no_sub) : 1; + + /* If set, a beginning-of-line anchor doesn't match at the beginning + of the string. */ + unsigned __REPB_PREFIX(not_bol) : 1; + + /* Similarly for an end-of-line anchor. */ + unsigned __REPB_PREFIX(not_eol) : 1; + + /* If true, an anchor at a newline matches. */ + unsigned __REPB_PREFIX(newline_anchor) : 1; +}; + +typedef struct re_pattern_buffer regex_t; + +/* Type for byte offsets within the string. POSIX mandates this. */ +#ifdef _REGEX_LARGE_OFFSETS +/* POSIX 1003.1-2008 requires that regoff_t be at least as wide as + ptrdiff_t and ssize_t. We don't know of any hosts where ptrdiff_t + is wider than ssize_t, so ssize_t is safe. ptrdiff_t is not + visible here, so use ssize_t. */ +typedef ssize_t regoff_t; +#else +/* The traditional GNU regex implementation mishandles strings longer + than INT_MAX. */ +typedef int regoff_t; +#endif + + +#ifdef __USE_GNU +/* This is the structure we store register match data in. See + regex.texinfo for a full description of what registers match. */ +struct re_registers +{ + __re_size_t num_regs; + regoff_t *start; + regoff_t *end; +}; + + +/* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, + 're_match_2' returns information about at least this many registers + the first time a 'regs' structure is passed. */ +# ifndef RE_NREGS +# define RE_NREGS 30 +# endif +#endif + + +/* POSIX specification for registers. Aside from the different names than + 're_registers', POSIX uses an array of structures, instead of a + structure of arrays. */ +typedef struct +{ + regoff_t rm_so; /* Byte offset from string's start to substring's start. */ + regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ +} regmatch_t; + +/* Declarations for routines. */ + +#ifdef __USE_GNU +/* Sets the current default syntax to SYNTAX, and return the old syntax. + You can also simply assign to the 're_syntax_options' variable. */ +extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax); + +/* Compile the regular expression PATTERN, with length LENGTH + and syntax given by the global 're_syntax_options', into the buffer + BUFFER. Return NULL if successful, and an error string if not. + + To free the allocated storage, you must call 'regfree' on BUFFER. + Note that the translate table must either have been initialized by + 'regcomp', with a malloc'ed value, or set to NULL before calling + 'regfree'. */ +extern const char *re_compile_pattern (const char *__pattern, size_t __length, + struct re_pattern_buffer *__buffer); + + +/* Compile a fastmap for the compiled pattern in BUFFER; used to + accelerate searches. Return 0 if successful and -2 if was an + internal error. */ +extern int re_compile_fastmap (struct re_pattern_buffer *__buffer); + + +/* Search in the string STRING (with length LENGTH) for the pattern + compiled into BUFFER. Start searching at position START, for RANGE + characters. Return the starting position of the match, -1 for no + match, or -2 for an internal error. Also return register + information in REGS (if REGS and BUFFER->no_sub are nonzero). */ +extern regoff_t re_search (struct re_pattern_buffer *__buffer, + const char *__String, regoff_t __length, + regoff_t __start, regoff_t __range, + struct re_registers *__regs); + + +/* Like 're_search', but search in the concatenation of STRING1 and + STRING2. Also, stop searching at index START + STOP. */ +extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer, + const char *__string1, regoff_t __length1, + const char *__string2, regoff_t __length2, + regoff_t __start, regoff_t __range, + struct re_registers *__regs, + regoff_t __stop); + + +/* Like 're_search', but return how many characters in STRING the regexp + in BUFFER matched, starting at position START. */ +extern regoff_t re_match (struct re_pattern_buffer *__buffer, + const char *__String, regoff_t __length, + regoff_t __start, struct re_registers *__regs); + + +/* Relates to 're_match' as 're_search_2' relates to 're_search'. */ +extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer, + const char *__string1, regoff_t __length1, + const char *__string2, regoff_t __length2, + regoff_t __start, struct re_registers *__regs, + regoff_t __stop); + + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using BUFFER and REGS will use this memory + for recording register information. STARTS and ENDS must be + allocated with malloc, and must each be at least 'NUM_REGS * sizeof + (regoff_t)' bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + BUFFER will allocate its own register data, without + freeing the old data. */ +extern void re_set_registers (struct re_pattern_buffer *__buffer, + struct re_registers *__regs, + __re_size_t __num_regs, + regoff_t *__starts, regoff_t *__ends); +#endif /* Use GNU */ + +#if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_MISC) +# ifndef _CRAY +/* 4.2 bsd compatibility. */ +extern char *re_comp (const char *); +extern int re_exec (const char *); +# endif +#endif + +/* For plain 'restrict', use glibc's __restrict if defined. + Otherwise, GCC 2.95 and later have "__restrict"; C99 compilers have + "restrict", and "configure" may have defined "restrict". + Other compilers use __restrict, __restrict__, and _Restrict, and + 'configure' might #define 'restrict' to those words, so pick a + different name. */ +#ifndef _Restrict_ +# if defined __restrict || 2 < __GNUC__ + (95 <= __GNUC_MINOR__) +# define _Restrict_ __restrict +# elif 199901L <= __STDC_VERSION__ || defined restrict +# define _Restrict_ restrict +# else +# define _Restrict_ +# endif +#endif +/* For [restrict], use glibc's __restrict_arr if available. + Otherwise, GCC 3.1 (not in C++ mode) and C99 support [restrict]. */ +#ifndef _Restrict_arr_ +# ifdef __restrict_arr +# define _Restrict_arr_ __restrict_arr +# elif ((199901L <= __STDC_VERSION__ || 3 < __GNUC__ + (1 <= __GNUC_MINOR__)) \ + && !defined __GNUG__) +# define _Restrict_arr_ _Restrict_ +# else +# define _Restrict_arr_ +# endif +#endif + +/* POSIX compatibility. */ +extern int regcomp (regex_t *_Restrict_ __preg, + const char *_Restrict_ __pattern, + int __cflags); + +extern int regexec (const regex_t *_Restrict_ __preg, + const char *_Restrict_ __String, size_t __nmatch, + regmatch_t __pmatch[_Restrict_arr_], + int __eflags); + +extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg, + char *_Restrict_ __errbuf, size_t __errbuf_size); + +extern void regfree (regex_t *__preg); + + +#ifdef __cplusplus +} +#endif /* C++ */ + +#endif /* regex.h */ diff --git a/lib/regex_internal.c b/lib/regex_internal.c new file mode 100644 index 00000000000..b592f06725c --- /dev/null +++ b/lib/regex_internal.c @@ -0,0 +1,1746 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +static void re_string_construct_common (const char *str, Idx len, + re_string_t *pstr, + RE_TRANSLATE_TYPE trans, bool icase, + const re_dfa_t *dfa); +static re_dfastate_t *create_ci_newstate (const re_dfa_t *dfa, + const re_node_set *nodes, + re_hashval_t hash); +static re_dfastate_t *create_cd_newstate (const re_dfa_t *dfa, + const re_node_set *nodes, + unsigned int context, + re_hashval_t hash); +static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, + Idx new_buf_len); +#ifdef RE_ENABLE_I18N +static void build_wcs_buffer (re_string_t *pstr); +static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr); +#endif /* RE_ENABLE_I18N */ +static void build_upper_buffer (re_string_t *pstr); +static void re_string_translate_buffer (re_string_t *pstr); +static unsigned int re_string_context_at (const re_string_t *input, Idx idx, + int eflags) __attribute__ ((pure)); + +/* Functions for string operation. */ + +/* This function allocate the buffers. It is necessary to call + re_string_reconstruct before using the object. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, + RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) +{ + reg_errcode_t ret; + Idx init_buf_len; + + /* Ensure at least one character fits into the buffers. */ + if (init_len < dfa->mb_cur_max) + init_len = dfa->mb_cur_max; + init_buf_len = (len + 1 < init_len) ? len + 1: init_len; + re_string_construct_common (str, len, pstr, trans, icase, dfa); + + ret = re_string_realloc_buffers (pstr, init_buf_len); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + + pstr->word_char = dfa->word_char; + pstr->word_ops_used = dfa->word_ops_used; + pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; + pstr->valid_len = (pstr->mbs_allocated || dfa->mb_cur_max > 1) ? 0 : len; + pstr->valid_raw_len = pstr->valid_len; + return REG_NOERROR; +} + +/* This function allocate the buffers, and initialize them. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_string_construct (re_string_t *pstr, const char *str, Idx len, + RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) +{ + reg_errcode_t ret; + memset (pstr, '\0', sizeof (re_string_t)); + re_string_construct_common (str, len, pstr, trans, icase, dfa); + + if (len > 0) + { + ret = re_string_realloc_buffers (pstr, len + 1); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + } + pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; + + if (icase) + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + while (1) + { + ret = build_wcs_upper_buffer (pstr); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + if (pstr->valid_raw_len >= len) + break; + if (pstr->bufs_len > pstr->valid_len + dfa->mb_cur_max) + break; + ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + } + } + else +#endif /* RE_ENABLE_I18N */ + build_upper_buffer (pstr); + } + else + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + build_wcs_buffer (pstr); + else +#endif /* RE_ENABLE_I18N */ + { + if (trans != NULL) + re_string_translate_buffer (pstr); + else + { + pstr->valid_len = pstr->bufs_len; + pstr->valid_raw_len = pstr->bufs_len; + } + } + } + + return REG_NOERROR; +} + +/* Helper functions for re_string_allocate, and re_string_construct. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_string_realloc_buffers (re_string_t *pstr, Idx new_buf_len) +{ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + wint_t *new_wcs; + + /* Avoid overflow in realloc. */ + const size_t max_object_size = MAX (sizeof (wint_t), sizeof (Idx)); + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) + < new_buf_len)) + return REG_ESPACE; + + new_wcs = re_realloc (pstr->wcs, wint_t, new_buf_len); + if (__glibc_unlikely (new_wcs == NULL)) + return REG_ESPACE; + pstr->wcs = new_wcs; + if (pstr->offsets != NULL) + { + Idx *new_offsets = re_realloc (pstr->offsets, Idx, new_buf_len); + if (__glibc_unlikely (new_offsets == NULL)) + return REG_ESPACE; + pstr->offsets = new_offsets; + } + } +#endif /* RE_ENABLE_I18N */ + if (pstr->mbs_allocated) + { + unsigned char *new_mbs = re_realloc (pstr->mbs, unsigned char, + new_buf_len); + if (__glibc_unlikely (new_mbs == NULL)) + return REG_ESPACE; + pstr->mbs = new_mbs; + } + pstr->bufs_len = new_buf_len; + return REG_NOERROR; +} + + +static void +re_string_construct_common (const char *str, Idx len, re_string_t *pstr, + RE_TRANSLATE_TYPE trans, bool icase, + const re_dfa_t *dfa) +{ + pstr->raw_mbs = (const unsigned char *) str; + pstr->len = len; + pstr->raw_len = len; + pstr->trans = trans; + pstr->icase = icase; + pstr->mbs_allocated = (trans != NULL || icase); + pstr->mb_cur_max = dfa->mb_cur_max; + pstr->is_utf8 = dfa->is_utf8; + pstr->map_notascii = dfa->map_notascii; + pstr->stop = pstr->len; + pstr->raw_stop = pstr->stop; +} + +#ifdef RE_ENABLE_I18N + +/* Build wide character buffer PSTR->WCS. + If the byte sequence of the string are: + <mb1>(0), <mb1>(1), <mb2>(0), <mb2>(1), <sb3> + Then wide character buffer will be: + <wc1> , WEOF , <wc2> , WEOF , <wc3> + We use WEOF for padding, they indicate that the position isn't + a first byte of a multibyte character. + + Note that this function assumes PSTR->VALID_LEN elements are already + built and starts from PSTR->VALID_LEN. */ + +static void +build_wcs_buffer (re_string_t *pstr) +{ +#ifdef _LIBC + unsigned char buf[MB_LEN_MAX]; + assert (MB_LEN_MAX >= pstr->mb_cur_max); +#else + unsigned char buf[64]; +#endif + mbstate_t prev_st; + Idx byte_idx, end_idx, remain_len; + size_t mbclen; + + /* Build the buffers from pstr->valid_len to either pstr->len or + pstr->bufs_len. */ + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + for (byte_idx = pstr->valid_len; byte_idx < end_idx;) + { + wchar_t wc; + const char *p; + + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + /* Apply the translation if we need. */ + if (__glibc_unlikely (pstr->trans != NULL)) + { + int i, ch; + + for (i = 0; i < pstr->mb_cur_max && i < remain_len; ++i) + { + ch = pstr->raw_mbs [pstr->raw_mbs_idx + byte_idx + i]; + buf[i] = pstr->mbs[byte_idx + i] = pstr->trans[ch]; + } + p = (const char *) buf; + } + else + p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx; + mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); + if (__glibc_unlikely (mbclen == (size_t) -1 || mbclen == 0 + || (mbclen == (size_t) -2 + && pstr->bufs_len >= pstr->len))) + { + /* We treat these cases as a singlebyte character. */ + mbclen = 1; + wc = (wchar_t) pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; + if (__glibc_unlikely (pstr->trans != NULL)) + wc = pstr->trans[wc]; + pstr->cur_state = prev_st; + } + else if (__glibc_unlikely (mbclen == (size_t) -2)) + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + + /* Write wide character and padding. */ + pstr->wcs[byte_idx++] = wc; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = byte_idx; +} + +/* Build wide character buffer PSTR->WCS like build_wcs_buffer, + but for REG_ICASE. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +build_wcs_upper_buffer (re_string_t *pstr) +{ + mbstate_t prev_st; + Idx src_idx, byte_idx, end_idx, remain_len; + size_t mbclen; +#ifdef _LIBC + char buf[MB_LEN_MAX]; + assert (MB_LEN_MAX >= pstr->mb_cur_max); +#else + char buf[64]; +#endif + + byte_idx = pstr->valid_len; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + /* The following optimization assumes that ASCII characters can be + mapped to wide characters with a simple cast. */ + if (! pstr->map_notascii && pstr->trans == NULL && !pstr->offsets_needed) + { + while (byte_idx < end_idx) + { + wchar_t wc; + + if (isascii (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]) + && mbsinit (&pstr->cur_state)) + { + /* In case of a singlebyte character. */ + pstr->mbs[byte_idx] + = toupper (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]); + /* The next step uses the assumption that wchar_t is encoded + ASCII-safe: all ASCII values can be converted like this. */ + pstr->wcs[byte_idx] = (wchar_t) pstr->mbs[byte_idx]; + ++byte_idx; + continue; + } + + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + mbclen = __mbrtowc (&wc, + ((const char *) pstr->raw_mbs + pstr->raw_mbs_idx + + byte_idx), remain_len, &pstr->cur_state); + if (__glibc_likely (0 < mbclen && mbclen < (size_t) -2)) + { + wchar_t wcu = __towupper (wc); + if (wcu != wc) + { + size_t mbcdlen; + + mbcdlen = __wcrtomb (buf, wcu, &prev_st); + if (__glibc_likely (mbclen == mbcdlen)) + memcpy (pstr->mbs + byte_idx, buf, mbclen); + else + { + src_idx = byte_idx; + goto offsets_needed; + } + } + else + memcpy (pstr->mbs + byte_idx, + pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx, mbclen); + pstr->wcs[byte_idx++] = wcu; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + else if (mbclen == (size_t) -1 || mbclen == 0 + || (mbclen == (size_t) -2 && pstr->bufs_len >= pstr->len)) + { + /* It is an invalid character, an incomplete character + at the end of the string, or '\0'. Just use the byte. */ + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; + pstr->mbs[byte_idx] = ch; + /* And also cast it to wide char. */ + pstr->wcs[byte_idx++] = (wchar_t) ch; + if (__glibc_unlikely (mbclen == (size_t) -1)) + pstr->cur_state = prev_st; + } + else + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = byte_idx; + return REG_NOERROR; + } + else + for (src_idx = pstr->valid_raw_len; byte_idx < end_idx;) + { + wchar_t wc; + const char *p; + offsets_needed: + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + if (__glibc_unlikely (pstr->trans != NULL)) + { + int i, ch; + + for (i = 0; i < pstr->mb_cur_max && i < remain_len; ++i) + { + ch = pstr->raw_mbs [pstr->raw_mbs_idx + src_idx + i]; + buf[i] = pstr->trans[ch]; + } + p = (const char *) buf; + } + else + p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + src_idx; + mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); + if (__glibc_likely (0 < mbclen && mbclen < (size_t) -2)) + { + wchar_t wcu = __towupper (wc); + if (wcu != wc) + { + size_t mbcdlen; + + mbcdlen = __wcrtomb ((char *) buf, wcu, &prev_st); + if (__glibc_likely (mbclen == mbcdlen)) + memcpy (pstr->mbs + byte_idx, buf, mbclen); + else if (mbcdlen != (size_t) -1) + { + size_t i; + + if (byte_idx + mbcdlen > pstr->bufs_len) + { + pstr->cur_state = prev_st; + break; + } + + if (pstr->offsets == NULL) + { + pstr->offsets = re_malloc (Idx, pstr->bufs_len); + + if (pstr->offsets == NULL) + return REG_ESPACE; + } + if (!pstr->offsets_needed) + { + for (i = 0; i < (size_t) byte_idx; ++i) + pstr->offsets[i] = i; + pstr->offsets_needed = 1; + } + + memcpy (pstr->mbs + byte_idx, buf, mbcdlen); + pstr->wcs[byte_idx] = wcu; + pstr->offsets[byte_idx] = src_idx; + for (i = 1; i < mbcdlen; ++i) + { + pstr->offsets[byte_idx + i] + = src_idx + (i < mbclen ? i : mbclen - 1); + pstr->wcs[byte_idx + i] = WEOF; + } + pstr->len += mbcdlen - mbclen; + if (pstr->raw_stop > src_idx) + pstr->stop += mbcdlen - mbclen; + end_idx = (pstr->bufs_len > pstr->len) + ? pstr->len : pstr->bufs_len; + byte_idx += mbcdlen; + src_idx += mbclen; + continue; + } + else + memcpy (pstr->mbs + byte_idx, p, mbclen); + } + else + memcpy (pstr->mbs + byte_idx, p, mbclen); + + if (__glibc_unlikely (pstr->offsets_needed != 0)) + { + size_t i; + for (i = 0; i < mbclen; ++i) + pstr->offsets[byte_idx + i] = src_idx + i; + } + src_idx += mbclen; + + pstr->wcs[byte_idx++] = wcu; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + else if (mbclen == (size_t) -1 || mbclen == 0 + || (mbclen == (size_t) -2 && pstr->bufs_len >= pstr->len)) + { + /* It is an invalid character or '\0'. Just use the byte. */ + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + src_idx]; + + if (__glibc_unlikely (pstr->trans != NULL)) + ch = pstr->trans [ch]; + pstr->mbs[byte_idx] = ch; + + if (__glibc_unlikely (pstr->offsets_needed != 0)) + pstr->offsets[byte_idx] = src_idx; + ++src_idx; + + /* And also cast it to wide char. */ + pstr->wcs[byte_idx++] = (wchar_t) ch; + if (__glibc_unlikely (mbclen == (size_t) -1)) + pstr->cur_state = prev_st; + } + else + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = src_idx; + return REG_NOERROR; +} + +/* Skip characters until the index becomes greater than NEW_RAW_IDX. + Return the index. */ + +static Idx +re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) +{ + mbstate_t prev_st; + Idx rawbuf_idx; + size_t mbclen; + wint_t wc = WEOF; + + /* Skip the characters which are not necessary to check. */ + for (rawbuf_idx = pstr->raw_mbs_idx + pstr->valid_raw_len; + rawbuf_idx < new_raw_idx;) + { + wchar_t wc2; + Idx remain_len = pstr->raw_len - rawbuf_idx; + prev_st = pstr->cur_state; + mbclen = __mbrtowc (&wc2, (const char *) pstr->raw_mbs + rawbuf_idx, + remain_len, &pstr->cur_state); + if (__glibc_unlikely (mbclen == (size_t) -2 || mbclen == (size_t) -1 + || mbclen == 0)) + { + /* We treat these cases as a single byte character. */ + if (mbclen == 0 || remain_len == 0) + wc = L'\0'; + else + wc = *(unsigned char *) (pstr->raw_mbs + rawbuf_idx); + mbclen = 1; + pstr->cur_state = prev_st; + } + else + wc = wc2; + /* Then proceed the next character. */ + rawbuf_idx += mbclen; + } + *last_wc = wc; + return rawbuf_idx; +} +#endif /* RE_ENABLE_I18N */ + +/* Build the buffer PSTR->MBS, and apply the translation if we need. + This function is used in case of REG_ICASE. */ + +static void +build_upper_buffer (re_string_t *pstr) +{ + Idx char_idx, end_idx; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + for (char_idx = pstr->valid_len; char_idx < end_idx; ++char_idx) + { + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + char_idx]; + if (__glibc_unlikely (pstr->trans != NULL)) + ch = pstr->trans[ch]; + pstr->mbs[char_idx] = toupper (ch); + } + pstr->valid_len = char_idx; + pstr->valid_raw_len = char_idx; +} + +/* Apply TRANS to the buffer in PSTR. */ + +static void +re_string_translate_buffer (re_string_t *pstr) +{ + Idx buf_idx, end_idx; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + for (buf_idx = pstr->valid_len; buf_idx < end_idx; ++buf_idx) + { + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + buf_idx]; + pstr->mbs[buf_idx] = pstr->trans[ch]; + } + + pstr->valid_len = buf_idx; + pstr->valid_raw_len = buf_idx; +} + +/* This function re-construct the buffers. + Concretely, convert to wide character in case of pstr->mb_cur_max > 1, + convert to upper case in case of REG_ICASE, apply translation. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) +{ + Idx offset; + + if (__glibc_unlikely (pstr->raw_mbs_idx <= idx)) + offset = idx - pstr->raw_mbs_idx; + else + { + /* Reset buffer. */ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); +#endif /* RE_ENABLE_I18N */ + pstr->len = pstr->raw_len; + pstr->stop = pstr->raw_stop; + pstr->valid_len = 0; + pstr->raw_mbs_idx = 0; + pstr->valid_raw_len = 0; + pstr->offsets_needed = 0; + pstr->tip_context = ((eflags & REG_NOTBOL) ? CONTEXT_BEGBUF + : CONTEXT_NEWLINE | CONTEXT_BEGBUF); + if (!pstr->mbs_allocated) + pstr->mbs = (unsigned char *) pstr->raw_mbs; + offset = idx; + } + + if (__glibc_likely (offset != 0)) + { + /* Should the already checked characters be kept? */ + if (__glibc_likely (offset < pstr->valid_raw_len)) + { + /* Yes, move them to the front of the buffer. */ +#ifdef RE_ENABLE_I18N + if (__glibc_unlikely (pstr->offsets_needed)) + { + Idx low = 0, high = pstr->valid_len, mid; + do + { + mid = (high + low) / 2; + if (pstr->offsets[mid] > offset) + high = mid; + else if (pstr->offsets[mid] < offset) + low = mid + 1; + else + break; + } + while (low < high); + if (pstr->offsets[mid] < offset) + ++mid; + pstr->tip_context = re_string_context_at (pstr, mid - 1, + eflags); + /* This can be quite complicated, so handle specially + only the common and easy case where the character with + different length representation of lower and upper + case is present at or after offset. */ + if (pstr->valid_len > offset + && mid == offset && pstr->offsets[mid] == offset) + { + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); + memmove (pstr->mbs, pstr->mbs + offset, pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; + for (low = 0; low < pstr->valid_len; low++) + pstr->offsets[low] = pstr->offsets[low + offset] - offset; + } + else + { + /* Otherwise, just find out how long the partial multibyte + character at offset is and fill it with WEOF/255. */ + pstr->len = pstr->raw_len - idx + offset; + pstr->stop = pstr->raw_stop - idx + offset; + pstr->offsets_needed = 0; + while (mid > 0 && pstr->offsets[mid - 1] == offset) + --mid; + while (mid < pstr->valid_len) + if (pstr->wcs[mid] != WEOF) + break; + else + ++mid; + if (mid == pstr->valid_len) + pstr->valid_len = 0; + else + { + pstr->valid_len = pstr->offsets[mid] - offset; + if (pstr->valid_len) + { + for (low = 0; low < pstr->valid_len; ++low) + pstr->wcs[low] = WEOF; + memset (pstr->mbs, 255, pstr->valid_len); + } + } + pstr->valid_raw_len = pstr->valid_len; + } + } + else +#endif + { + pstr->tip_context = re_string_context_at (pstr, offset - 1, + eflags); +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); +#endif /* RE_ENABLE_I18N */ + if (__glibc_unlikely (pstr->mbs_allocated)) + memmove (pstr->mbs, pstr->mbs + offset, + pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; +#if defined DEBUG && DEBUG + assert (pstr->valid_len > 0); +#endif + } + } + else + { +#ifdef RE_ENABLE_I18N + /* No, skip all characters until IDX. */ + Idx prev_valid_len = pstr->valid_len; + + if (__glibc_unlikely (pstr->offsets_needed)) + { + pstr->len = pstr->raw_len - idx + offset; + pstr->stop = pstr->raw_stop - idx + offset; + pstr->offsets_needed = 0; + } +#endif + pstr->valid_len = 0; +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + Idx wcs_idx; + wint_t wc = WEOF; + + if (pstr->is_utf8) + { + const unsigned char *raw, *p, *end; + + /* Special case UTF-8. Multi-byte chars start with any + byte other than 0x80 - 0xbf. */ + raw = pstr->raw_mbs + pstr->raw_mbs_idx; + end = raw + (offset - pstr->mb_cur_max); + if (end < pstr->raw_mbs) + end = pstr->raw_mbs; + p = raw + offset - 1; +#ifdef _LIBC + /* We know the wchar_t encoding is UCS4, so for the simple + case, ASCII characters, skip the conversion step. */ + if (isascii (*p) && __glibc_likely (pstr->trans == NULL)) + { + memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); + /* pstr->valid_len = 0; */ + wc = (wchar_t) *p; + } + else +#endif + for (; p >= end; --p) + if ((*p & 0xc0) != 0x80) + { + mbstate_t cur_state; + wchar_t wc2; + Idx mlen = raw + pstr->len - p; + unsigned char buf[6]; + size_t mbclen; + + const unsigned char *pp = p; + if (__glibc_unlikely (pstr->trans != NULL)) + { + int i = mlen < 6 ? mlen : 6; + while (--i >= 0) + buf[i] = pstr->trans[p[i]]; + pp = buf; + } + /* XXX Don't use mbrtowc, we know which conversion + to use (UTF-8 -> UCS4). */ + memset (&cur_state, 0, sizeof (cur_state)); + mbclen = __mbrtowc (&wc2, (const char *) pp, mlen, + &cur_state); + if (raw + offset - p <= mbclen + && mbclen < (size_t) -2) + { + memset (&pstr->cur_state, '\0', + sizeof (mbstate_t)); + pstr->valid_len = mbclen - (raw + offset - p); + wc = wc2; + } + break; + } + } + + if (wc == WEOF) + pstr->valid_len = re_string_skip_chars (pstr, idx, &wc) - idx; + if (wc == WEOF) + pstr->tip_context + = re_string_context_at (pstr, prev_valid_len - 1, eflags); + else + pstr->tip_context = ((__glibc_unlikely (pstr->word_ops_used != 0) + && IS_WIDE_WORD_CHAR (wc)) + ? CONTEXT_WORD + : ((IS_WIDE_NEWLINE (wc) + && pstr->newline_anchor) + ? CONTEXT_NEWLINE : 0)); + if (__glibc_unlikely (pstr->valid_len)) + { + for (wcs_idx = 0; wcs_idx < pstr->valid_len; ++wcs_idx) + pstr->wcs[wcs_idx] = WEOF; + if (pstr->mbs_allocated) + memset (pstr->mbs, 255, pstr->valid_len); + } + pstr->valid_raw_len = pstr->valid_len; + } + else +#endif /* RE_ENABLE_I18N */ + { + int c = pstr->raw_mbs[pstr->raw_mbs_idx + offset - 1]; + pstr->valid_raw_len = 0; + if (pstr->trans) + c = pstr->trans[c]; + pstr->tip_context = (bitset_contain (pstr->word_char, c) + ? CONTEXT_WORD + : ((IS_NEWLINE (c) && pstr->newline_anchor) + ? CONTEXT_NEWLINE : 0)); + } + } + if (!__glibc_unlikely (pstr->mbs_allocated)) + pstr->mbs += offset; + } + pstr->raw_mbs_idx = idx; + pstr->len -= offset; + pstr->stop -= offset; + + /* Then build the buffers. */ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + if (pstr->icase) + { + reg_errcode_t ret = build_wcs_upper_buffer (pstr); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + } + else + build_wcs_buffer (pstr); + } + else +#endif /* RE_ENABLE_I18N */ + if (__glibc_unlikely (pstr->mbs_allocated)) + { + if (pstr->icase) + build_upper_buffer (pstr); + else if (pstr->trans != NULL) + re_string_translate_buffer (pstr); + } + else + pstr->valid_len = pstr->len; + + pstr->cur_idx = 0; + return REG_NOERROR; +} + +static unsigned char +__attribute__ ((pure)) +re_string_peek_byte_case (const re_string_t *pstr, Idx idx) +{ + int ch; + Idx off; + + /* Handle the common (easiest) cases first. */ + if (__glibc_likely (!pstr->mbs_allocated)) + return re_string_peek_byte (pstr, idx); + +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1 + && ! re_string_is_single_byte_char (pstr, pstr->cur_idx + idx)) + return re_string_peek_byte (pstr, idx); +#endif + + off = pstr->cur_idx + idx; +#ifdef RE_ENABLE_I18N + if (pstr->offsets_needed) + off = pstr->offsets[off]; +#endif + + ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; + +#ifdef RE_ENABLE_I18N + /* Ensure that e.g. for tr_TR.UTF-8 BACKSLASH DOTLESS SMALL LETTER I + this function returns CAPITAL LETTER I instead of first byte of + DOTLESS SMALL LETTER I. The latter would confuse the parser, + since peek_byte_case doesn't advance cur_idx in any way. */ + if (pstr->offsets_needed && !isascii (ch)) + return re_string_peek_byte (pstr, idx); +#endif + + return ch; +} + +static unsigned char +re_string_fetch_byte_case (re_string_t *pstr) +{ + if (__glibc_likely (!pstr->mbs_allocated)) + return re_string_fetch_byte (pstr); + +#ifdef RE_ENABLE_I18N + if (pstr->offsets_needed) + { + Idx off; + int ch; + + /* For tr_TR.UTF-8 [[:islower:]] there is + [[: CAPITAL LETTER I WITH DOT lower:]] in mbs. Skip + in that case the whole multi-byte character and return + the original letter. On the other side, with + [[: DOTLESS SMALL LETTER I return [[:I, as doing + anything else would complicate things too much. */ + + if (!re_string_first_byte (pstr, pstr->cur_idx)) + return re_string_fetch_byte (pstr); + + off = pstr->offsets[pstr->cur_idx]; + ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; + + if (! isascii (ch)) + return re_string_fetch_byte (pstr); + + re_string_skip_bytes (pstr, + re_string_char_size_at (pstr, pstr->cur_idx)); + return ch; + } +#endif + + return pstr->raw_mbs[pstr->raw_mbs_idx + pstr->cur_idx++]; +} + +static void +re_string_destruct (re_string_t *pstr) +{ +#ifdef RE_ENABLE_I18N + re_free (pstr->wcs); + re_free (pstr->offsets); +#endif /* RE_ENABLE_I18N */ + if (pstr->mbs_allocated) + re_free (pstr->mbs); +} + +/* Return the context at IDX in INPUT. */ + +static unsigned int +re_string_context_at (const re_string_t *input, Idx idx, int eflags) +{ + int c; + if (__glibc_unlikely (idx < 0)) + /* In this case, we use the value stored in input->tip_context, + since we can't know the character in input->mbs[-1] here. */ + return input->tip_context; + if (__glibc_unlikely (idx == input->len)) + return ((eflags & REG_NOTEOL) ? CONTEXT_ENDBUF + : CONTEXT_NEWLINE | CONTEXT_ENDBUF); +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc; + Idx wc_idx = idx; + while(input->wcs[wc_idx] == WEOF) + { +#if defined DEBUG && DEBUG + /* It must not happen. */ + assert (wc_idx >= 0); +#endif + --wc_idx; + if (wc_idx < 0) + return input->tip_context; + } + wc = input->wcs[wc_idx]; + if (__glibc_unlikely (input->word_ops_used != 0) + && IS_WIDE_WORD_CHAR (wc)) + return CONTEXT_WORD; + return (IS_WIDE_NEWLINE (wc) && input->newline_anchor + ? CONTEXT_NEWLINE : 0); + } + else +#endif + { + c = re_string_byte_at (input, idx); + if (bitset_contain (input->word_char, c)) + return CONTEXT_WORD; + return IS_NEWLINE (c) && input->newline_anchor ? CONTEXT_NEWLINE : 0; + } +} + +/* Functions for set operation. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_alloc (re_node_set *set, Idx size) +{ + set->alloc = size; + set->nelem = 0; + set->elems = re_malloc (Idx, size); + if (__glibc_unlikely (set->elems == NULL) + && (MALLOC_0_IS_NONNULL || size != 0)) + return REG_ESPACE; + return REG_NOERROR; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_init_1 (re_node_set *set, Idx elem) +{ + set->alloc = 1; + set->nelem = 1; + set->elems = re_malloc (Idx, 1); + if (__glibc_unlikely (set->elems == NULL)) + { + set->alloc = set->nelem = 0; + return REG_ESPACE; + } + set->elems[0] = elem; + return REG_NOERROR; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_init_2 (re_node_set *set, Idx elem1, Idx elem2) +{ + set->alloc = 2; + set->elems = re_malloc (Idx, 2); + if (__glibc_unlikely (set->elems == NULL)) + return REG_ESPACE; + if (elem1 == elem2) + { + set->nelem = 1; + set->elems[0] = elem1; + } + else + { + set->nelem = 2; + if (elem1 < elem2) + { + set->elems[0] = elem1; + set->elems[1] = elem2; + } + else + { + set->elems[0] = elem2; + set->elems[1] = elem1; + } + } + return REG_NOERROR; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_init_copy (re_node_set *dest, const re_node_set *src) +{ + dest->nelem = src->nelem; + if (src->nelem > 0) + { + dest->alloc = dest->nelem; + dest->elems = re_malloc (Idx, dest->alloc); + if (__glibc_unlikely (dest->elems == NULL)) + { + dest->alloc = dest->nelem = 0; + return REG_ESPACE; + } + memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); + } + else + re_node_set_init_empty (dest); + return REG_NOERROR; +} + +/* Calculate the intersection of the sets SRC1 and SRC2. And merge it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. + Note: We assume dest->elems is NULL, when dest->alloc is 0. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, + const re_node_set *src2) +{ + Idx i1, i2, is, id, delta, sbase; + if (src1->nelem == 0 || src2->nelem == 0) + return REG_NOERROR; + + /* We need dest->nelem + 2 * elems_in_intersection; this is a + conservative estimate. */ + if (src1->nelem + src2->nelem + dest->nelem > dest->alloc) + { + Idx new_alloc = src1->nelem + src2->nelem + dest->alloc; + Idx *new_elems = re_realloc (dest->elems, Idx, new_alloc); + if (__glibc_unlikely (new_elems == NULL)) + return REG_ESPACE; + dest->elems = new_elems; + dest->alloc = new_alloc; + } + + /* Find the items in the intersection of SRC1 and SRC2, and copy + into the top of DEST those that are not already in DEST itself. */ + sbase = dest->nelem + src1->nelem + src2->nelem; + i1 = src1->nelem - 1; + i2 = src2->nelem - 1; + id = dest->nelem - 1; + for (;;) + { + if (src1->elems[i1] == src2->elems[i2]) + { + /* Try to find the item in DEST. Maybe we could binary search? */ + while (id >= 0 && dest->elems[id] > src1->elems[i1]) + --id; + + if (id < 0 || dest->elems[id] != src1->elems[i1]) + dest->elems[--sbase] = src1->elems[i1]; + + if (--i1 < 0 || --i2 < 0) + break; + } + + /* Lower the highest of the two items. */ + else if (src1->elems[i1] < src2->elems[i2]) + { + if (--i2 < 0) + break; + } + else + { + if (--i1 < 0) + break; + } + } + + id = dest->nelem - 1; + is = dest->nelem + src1->nelem + src2->nelem - 1; + delta = is - sbase + 1; + + /* Now copy. When DELTA becomes zero, the remaining + DEST elements are already in place; this is more or + less the same loop that is in re_node_set_merge. */ + dest->nelem += delta; + if (delta > 0 && id >= 0) + for (;;) + { + if (dest->elems[is] > dest->elems[id]) + { + /* Copy from the top. */ + dest->elems[id + delta--] = dest->elems[is--]; + if (delta == 0) + break; + } + else + { + /* Slide from the bottom. */ + dest->elems[id + delta] = dest->elems[id]; + if (--id < 0) + break; + } + } + + /* Copy remaining SRC elements. */ + memcpy (dest->elems, dest->elems + sbase, delta * sizeof (Idx)); + + return REG_NOERROR; +} + +/* Calculate the union set of the sets SRC1 and SRC2. And store it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_init_union (re_node_set *dest, const re_node_set *src1, + const re_node_set *src2) +{ + Idx i1, i2, id; + if (src1 != NULL && src1->nelem > 0 && src2 != NULL && src2->nelem > 0) + { + dest->alloc = src1->nelem + src2->nelem; + dest->elems = re_malloc (Idx, dest->alloc); + if (__glibc_unlikely (dest->elems == NULL)) + return REG_ESPACE; + } + else + { + if (src1 != NULL && src1->nelem > 0) + return re_node_set_init_copy (dest, src1); + else if (src2 != NULL && src2->nelem > 0) + return re_node_set_init_copy (dest, src2); + else + re_node_set_init_empty (dest); + return REG_NOERROR; + } + for (i1 = i2 = id = 0 ; i1 < src1->nelem && i2 < src2->nelem ;) + { + if (src1->elems[i1] > src2->elems[i2]) + { + dest->elems[id++] = src2->elems[i2++]; + continue; + } + if (src1->elems[i1] == src2->elems[i2]) + ++i2; + dest->elems[id++] = src1->elems[i1++]; + } + if (i1 < src1->nelem) + { + memcpy (dest->elems + id, src1->elems + i1, + (src1->nelem - i1) * sizeof (Idx)); + id += src1->nelem - i1; + } + else if (i2 < src2->nelem) + { + memcpy (dest->elems + id, src2->elems + i2, + (src2->nelem - i2) * sizeof (Idx)); + id += src2->nelem - i2; + } + dest->nelem = id; + return REG_NOERROR; +} + +/* Calculate the union set of the sets DEST and SRC. And store it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_node_set_merge (re_node_set *dest, const re_node_set *src) +{ + Idx is, id, sbase, delta; + if (src == NULL || src->nelem == 0) + return REG_NOERROR; + if (dest->alloc < 2 * src->nelem + dest->nelem) + { + Idx new_alloc = 2 * (src->nelem + dest->alloc); + Idx *new_buffer = re_realloc (dest->elems, Idx, new_alloc); + if (__glibc_unlikely (new_buffer == NULL)) + return REG_ESPACE; + dest->elems = new_buffer; + dest->alloc = new_alloc; + } + + if (__glibc_unlikely (dest->nelem == 0)) + { + dest->nelem = src->nelem; + memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); + return REG_NOERROR; + } + + /* Copy into the top of DEST the items of SRC that are not + found in DEST. Maybe we could binary search in DEST? */ + for (sbase = dest->nelem + 2 * src->nelem, + is = src->nelem - 1, id = dest->nelem - 1; is >= 0 && id >= 0; ) + { + if (dest->elems[id] == src->elems[is]) + is--, id--; + else if (dest->elems[id] < src->elems[is]) + dest->elems[--sbase] = src->elems[is--]; + else /* if (dest->elems[id] > src->elems[is]) */ + --id; + } + + if (is >= 0) + { + /* If DEST is exhausted, the remaining items of SRC must be unique. */ + sbase -= is + 1; + memcpy (dest->elems + sbase, src->elems, (is + 1) * sizeof (Idx)); + } + + id = dest->nelem - 1; + is = dest->nelem + 2 * src->nelem - 1; + delta = is - sbase + 1; + if (delta == 0) + return REG_NOERROR; + + /* Now copy. When DELTA becomes zero, the remaining + DEST elements are already in place. */ + dest->nelem += delta; + for (;;) + { + if (dest->elems[is] > dest->elems[id]) + { + /* Copy from the top. */ + dest->elems[id + delta--] = dest->elems[is--]; + if (delta == 0) + break; + } + else + { + /* Slide from the bottom. */ + dest->elems[id + delta] = dest->elems[id]; + if (--id < 0) + { + /* Copy remaining SRC elements. */ + memcpy (dest->elems, dest->elems + sbase, + delta * sizeof (Idx)); + break; + } + } + } + + return REG_NOERROR; +} + +/* Insert the new element ELEM to the re_node_set* SET. + SET should not already have ELEM. + Return true if successful. */ + +static bool +__attribute_warn_unused_result__ +re_node_set_insert (re_node_set *set, Idx elem) +{ + Idx idx; + /* In case the set is empty. */ + if (set->alloc == 0) + return __glibc_likely (re_node_set_init_1 (set, elem) == REG_NOERROR); + + if (__glibc_unlikely (set->nelem) == 0) + { + /* We already guaranteed above that set->alloc != 0. */ + set->elems[0] = elem; + ++set->nelem; + return true; + } + + /* Realloc if we need. */ + if (set->alloc == set->nelem) + { + Idx *new_elems; + set->alloc = set->alloc * 2; + new_elems = re_realloc (set->elems, Idx, set->alloc); + if (__glibc_unlikely (new_elems == NULL)) + return false; + set->elems = new_elems; + } + + /* Move the elements which follows the new element. Test the + first element separately to skip a check in the inner loop. */ + if (elem < set->elems[0]) + { + idx = 0; + for (idx = set->nelem; idx > 0; idx--) + set->elems[idx] = set->elems[idx - 1]; + } + else + { + for (idx = set->nelem; set->elems[idx - 1] > elem; idx--) + set->elems[idx] = set->elems[idx - 1]; + } + + /* Insert the new element. */ + set->elems[idx] = elem; + ++set->nelem; + return true; +} + +/* Insert the new element ELEM to the re_node_set* SET. + SET should not already have any element greater than or equal to ELEM. + Return true if successful. */ + +static bool +__attribute_warn_unused_result__ +re_node_set_insert_last (re_node_set *set, Idx elem) +{ + /* Realloc if we need. */ + if (set->alloc == set->nelem) + { + Idx *new_elems; + set->alloc = (set->alloc + 1) * 2; + new_elems = re_realloc (set->elems, Idx, set->alloc); + if (__glibc_unlikely (new_elems == NULL)) + return false; + set->elems = new_elems; + } + + /* Insert the new element. */ + set->elems[set->nelem++] = elem; + return true; +} + +/* Compare two node sets SET1 and SET2. + Return true if SET1 and SET2 are equivalent. */ + +static bool +__attribute__ ((pure)) +re_node_set_compare (const re_node_set *set1, const re_node_set *set2) +{ + Idx i; + if (set1 == NULL || set2 == NULL || set1->nelem != set2->nelem) + return false; + for (i = set1->nelem ; --i >= 0 ; ) + if (set1->elems[i] != set2->elems[i]) + return false; + return true; +} + +/* Return (idx + 1) if SET contains the element ELEM, return 0 otherwise. */ + +static Idx +__attribute__ ((pure)) +re_node_set_contains (const re_node_set *set, Idx elem) +{ + __re_size_t idx, right, mid; + if (set->nelem <= 0) + return 0; + + /* Binary search the element. */ + idx = 0; + right = set->nelem - 1; + while (idx < right) + { + mid = (idx + right) / 2; + if (set->elems[mid] < elem) + idx = mid + 1; + else + right = mid; + } + return set->elems[idx] == elem ? idx + 1 : 0; +} + +static void +re_node_set_remove_at (re_node_set *set, Idx idx) +{ + if (idx < 0 || idx >= set->nelem) + return; + --set->nelem; + for (; idx < set->nelem; idx++) + set->elems[idx] = set->elems[idx + 1]; +} + + +/* Add the token TOKEN to dfa->nodes, and return the index of the token. + Or return -1 if an error occurred. */ + +static Idx +re_dfa_add_node (re_dfa_t *dfa, re_token_t token) +{ + if (__glibc_unlikely (dfa->nodes_len >= dfa->nodes_alloc)) + { + size_t new_nodes_alloc = dfa->nodes_alloc * 2; + Idx *new_nexts, *new_indices; + re_node_set *new_edests, *new_eclosures; + re_token_t *new_nodes; + + /* Avoid overflows in realloc. */ + const size_t max_object_size = MAX (sizeof (re_token_t), + MAX (sizeof (re_node_set), + sizeof (Idx))); + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) + < new_nodes_alloc)) + return -1; + + new_nodes = re_realloc (dfa->nodes, re_token_t, new_nodes_alloc); + if (__glibc_unlikely (new_nodes == NULL)) + return -1; + dfa->nodes = new_nodes; + new_nexts = re_realloc (dfa->nexts, Idx, new_nodes_alloc); + new_indices = re_realloc (dfa->org_indices, Idx, new_nodes_alloc); + new_edests = re_realloc (dfa->edests, re_node_set, new_nodes_alloc); + new_eclosures = re_realloc (dfa->eclosures, re_node_set, new_nodes_alloc); + if (__glibc_unlikely (new_nexts == NULL || new_indices == NULL + || new_edests == NULL || new_eclosures == NULL)) + { + re_free (new_nexts); + re_free (new_indices); + re_free (new_edests); + re_free (new_eclosures); + return -1; + } + dfa->nexts = new_nexts; + dfa->org_indices = new_indices; + dfa->edests = new_edests; + dfa->eclosures = new_eclosures; + dfa->nodes_alloc = new_nodes_alloc; + } + dfa->nodes[dfa->nodes_len] = token; + dfa->nodes[dfa->nodes_len].constraint = 0; +#ifdef RE_ENABLE_I18N + dfa->nodes[dfa->nodes_len].accept_mb = + ((token.type == OP_PERIOD && dfa->mb_cur_max > 1) + || token.type == COMPLEX_BRACKET); +#endif + dfa->nexts[dfa->nodes_len] = -1; + re_node_set_init_empty (dfa->edests + dfa->nodes_len); + re_node_set_init_empty (dfa->eclosures + dfa->nodes_len); + return dfa->nodes_len++; +} + +static re_hashval_t +calc_state_hash (const re_node_set *nodes, unsigned int context) +{ + re_hashval_t hash = nodes->nelem + context; + Idx i; + for (i = 0 ; i < nodes->nelem ; i++) + hash += nodes->elems[i]; + return hash; +} + +/* Search for the state whose node_set is equivalent to NODES. + Return the pointer to the state, if we found it in the DFA. + Otherwise create the new one and return it. In case of an error + return NULL and set the error code in ERR. + Note: - We assume NULL as the invalid state, then it is possible that + return value is NULL and ERR is REG_NOERROR. + - We never return non-NULL value in case of any errors, it is for + optimization. */ + +static re_dfastate_t * +__attribute_warn_unused_result__ +re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, + const re_node_set *nodes) +{ + re_hashval_t hash; + re_dfastate_t *new_state; + struct re_state_table_entry *spot; + Idx i; +#if defined GCC_LINT || defined lint + /* Suppress bogus uninitialized-variable warnings. */ + *err = REG_NOERROR; +#endif + if (__glibc_unlikely (nodes->nelem == 0)) + { + *err = REG_NOERROR; + return NULL; + } + hash = calc_state_hash (nodes, 0); + spot = dfa->state_table + (hash & dfa->state_hash_mask); + + for (i = 0 ; i < spot->num ; i++) + { + re_dfastate_t *state = spot->array[i]; + if (hash != state->hash) + continue; + if (re_node_set_compare (&state->nodes, nodes)) + return state; + } + + /* There are no appropriate state in the dfa, create the new one. */ + new_state = create_ci_newstate (dfa, nodes, hash); + if (__glibc_unlikely (new_state == NULL)) + *err = REG_ESPACE; + + return new_state; +} + +/* Search for the state whose node_set is equivalent to NODES and + whose context is equivalent to CONTEXT. + Return the pointer to the state, if we found it in the DFA. + Otherwise create the new one and return it. In case of an error + return NULL and set the error code in ERR. + Note: - We assume NULL as the invalid state, then it is possible that + return value is NULL and ERR is REG_NOERROR. + - We never return non-NULL value in case of any errors, it is for + optimization. */ + +static re_dfastate_t * +__attribute_warn_unused_result__ +re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, + const re_node_set *nodes, unsigned int context) +{ + re_hashval_t hash; + re_dfastate_t *new_state; + struct re_state_table_entry *spot; + Idx i; +#if defined GCC_LINT || defined lint + /* Suppress bogus uninitialized-variable warnings. */ + *err = REG_NOERROR; +#endif + if (nodes->nelem == 0) + { + *err = REG_NOERROR; + return NULL; + } + hash = calc_state_hash (nodes, context); + spot = dfa->state_table + (hash & dfa->state_hash_mask); + + for (i = 0 ; i < spot->num ; i++) + { + re_dfastate_t *state = spot->array[i]; + if (state->hash == hash + && state->context == context + && re_node_set_compare (state->entrance_nodes, nodes)) + return state; + } + /* There are no appropriate state in 'dfa', create the new one. */ + new_state = create_cd_newstate (dfa, nodes, context, hash); + if (__glibc_unlikely (new_state == NULL)) + *err = REG_ESPACE; + + return new_state; +} + +/* Finish initialization of the new state NEWSTATE, and using its hash value + HASH put in the appropriate bucket of DFA's state table. Return value + indicates the error code if failed. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +register_state (const re_dfa_t *dfa, re_dfastate_t *newstate, + re_hashval_t hash) +{ + struct re_state_table_entry *spot; + reg_errcode_t err; + Idx i; + + newstate->hash = hash; + err = re_node_set_alloc (&newstate->non_eps_nodes, newstate->nodes.nelem); + if (__glibc_unlikely (err != REG_NOERROR)) + return REG_ESPACE; + for (i = 0; i < newstate->nodes.nelem; i++) + { + Idx elem = newstate->nodes.elems[i]; + if (!IS_EPSILON_NODE (dfa->nodes[elem].type)) + if (! re_node_set_insert_last (&newstate->non_eps_nodes, elem)) + return REG_ESPACE; + } + + spot = dfa->state_table + (hash & dfa->state_hash_mask); + if (__glibc_unlikely (spot->alloc <= spot->num)) + { + Idx new_alloc = 2 * spot->num + 2; + re_dfastate_t **new_array = re_realloc (spot->array, re_dfastate_t *, + new_alloc); + if (__glibc_unlikely (new_array == NULL)) + return REG_ESPACE; + spot->array = new_array; + spot->alloc = new_alloc; + } + spot->array[spot->num++] = newstate; + return REG_NOERROR; +} + +static void +free_state (re_dfastate_t *state) +{ + re_node_set_free (&state->non_eps_nodes); + re_node_set_free (&state->inveclosure); + if (state->entrance_nodes != &state->nodes) + { + re_node_set_free (state->entrance_nodes); + re_free (state->entrance_nodes); + } + re_node_set_free (&state->nodes); + re_free (state->word_trtable); + re_free (state->trtable); + re_free (state); +} + +/* Create the new state which is independent of contexts. + Return the new state if succeeded, otherwise return NULL. */ + +static re_dfastate_t * +__attribute_warn_unused_result__ +create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, + re_hashval_t hash) +{ + Idx i; + reg_errcode_t err; + re_dfastate_t *newstate; + + newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); + if (__glibc_unlikely (newstate == NULL)) + return NULL; + err = re_node_set_init_copy (&newstate->nodes, nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_free (newstate); + return NULL; + } + + newstate->entrance_nodes = &newstate->nodes; + for (i = 0 ; i < nodes->nelem ; i++) + { + re_token_t *node = dfa->nodes + nodes->elems[i]; + re_token_type_t type = node->type; + if (type == CHARACTER && !node->constraint) + continue; +#ifdef RE_ENABLE_I18N + newstate->accept_mb |= node->accept_mb; +#endif /* RE_ENABLE_I18N */ + + /* If the state has the halt node, the state is a halt state. */ + if (type == END_OF_RE) + newstate->halt = 1; + else if (type == OP_BACK_REF) + newstate->has_backref = 1; + else if (type == ANCHOR || node->constraint) + newstate->has_constraint = 1; + } + err = register_state (dfa, newstate, hash); + if (__glibc_unlikely (err != REG_NOERROR)) + { + free_state (newstate); + newstate = NULL; + } + return newstate; +} + +/* Create the new state which is depend on the context CONTEXT. + Return the new state if succeeded, otherwise return NULL. */ + +static re_dfastate_t * +__attribute_warn_unused_result__ +create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, + unsigned int context, re_hashval_t hash) +{ + Idx i, nctx_nodes = 0; + reg_errcode_t err; + re_dfastate_t *newstate; + + newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); + if (__glibc_unlikely (newstate == NULL)) + return NULL; + err = re_node_set_init_copy (&newstate->nodes, nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_free (newstate); + return NULL; + } + + newstate->context = context; + newstate->entrance_nodes = &newstate->nodes; + + for (i = 0 ; i < nodes->nelem ; i++) + { + re_token_t *node = dfa->nodes + nodes->elems[i]; + re_token_type_t type = node->type; + unsigned int constraint = node->constraint; + + if (type == CHARACTER && !constraint) + continue; +#ifdef RE_ENABLE_I18N + newstate->accept_mb |= node->accept_mb; +#endif /* RE_ENABLE_I18N */ + + /* If the state has the halt node, the state is a halt state. */ + if (type == END_OF_RE) + newstate->halt = 1; + else if (type == OP_BACK_REF) + newstate->has_backref = 1; + + if (constraint) + { + if (newstate->entrance_nodes == &newstate->nodes) + { + newstate->entrance_nodes = re_malloc (re_node_set, 1); + if (__glibc_unlikely (newstate->entrance_nodes == NULL)) + { + free_state (newstate); + return NULL; + } + if (re_node_set_init_copy (newstate->entrance_nodes, nodes) + != REG_NOERROR) + return NULL; + nctx_nodes = 0; + newstate->has_constraint = 1; + } + + if (NOT_SATISFY_PREV_CONSTRAINT (constraint,context)) + { + re_node_set_remove_at (&newstate->nodes, i - nctx_nodes); + ++nctx_nodes; + } + } + } + err = register_state (dfa, newstate, hash); + if (__glibc_unlikely (err != REG_NOERROR)) + { + free_state (newstate); + newstate = NULL; + } + return newstate; +} diff --git a/lib/regex_internal.h b/lib/regex_internal.h new file mode 100644 index 00000000000..a3aedda8915 --- /dev/null +++ b/lib/regex_internal.h @@ -0,0 +1,874 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#ifndef _REGEX_INTERNAL_H +#define _REGEX_INTERNAL_H 1 + +#include <assert.h> +#include <ctype.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include <langinfo.h> +#include <locale.h> +#include <wchar.h> +#include <wctype.h> +#include <stdbool.h> +#include <stdint.h> + +#include <intprops.h> + +#ifdef _LIBC +# include <libc-lock.h> +# define lock_define(name) __libc_lock_define (, name) +# define lock_init(lock) (__libc_lock_init (lock), 0) +# define lock_fini(lock) ((void) 0) +# define lock_lock(lock) __libc_lock_lock (lock) +# define lock_unlock(lock) __libc_lock_unlock (lock) +#elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO +# include "glthread/lock.h" + /* Use gl_lock_define if empty macro arguments are known to work. + Otherwise, fall back on less-portable substitutes. */ +# if ((defined __GNUC__ && !defined __STRICT_ANSI__) \ + || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__)) +# define lock_define(name) gl_lock_define (, name) +# elif USE_POSIX_THREADS +# define lock_define(name) pthread_mutex_t name; +# elif USE_PTH_THREADS +# define lock_define(name) pth_mutex_t name; +# elif USE_SOLARIS_THREADS +# define lock_define(name) mutex_t name; +# elif USE_WINDOWS_THREADS +# define lock_define(name) gl_lock_t name; +# else +# define lock_define(name) +# endif +# define lock_init(lock) glthread_lock_init (&(lock)) +# define lock_fini(lock) glthread_lock_destroy (&(lock)) +# define lock_lock(lock) glthread_lock_lock (&(lock)) +# define lock_unlock(lock) glthread_lock_unlock (&(lock)) +#elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO +# include <pthread.h> +# define lock_define(name) pthread_mutex_t name; +# define lock_init(lock) pthread_mutex_init (&(lock), 0) +# define lock_fini(lock) pthread_mutex_destroy (&(lock)) +# define lock_lock(lock) pthread_mutex_lock (&(lock)) +# define lock_unlock(lock) pthread_mutex_unlock (&(lock)) +#else +# define lock_define(name) +# define lock_init(lock) 0 +# define lock_fini(lock) ((void) 0) + /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ +# define lock_lock(lock) ((void) dfa) +# define lock_unlock(lock) ((void) 0) +#endif + +/* In case that the system doesn't have isblank(). */ +#if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) +# define isblank(ch) ((ch) == ' ' || (ch) == '\t') +#endif + +#ifdef _LIBC +# ifndef _RE_DEFINE_LOCALE_FUNCTIONS +# define _RE_DEFINE_LOCALE_FUNCTIONS 1 +# include <locale/localeinfo.h> +# include <locale/coll-lookup.h> +# endif +#endif + +/* This is for other GNU distributions with internationalized messages. */ +#if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC +# include <libintl.h> +# ifdef _LIBC +# undef gettext +# define gettext(msgid) \ + __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) +# endif +#else +# undef gettext +# define gettext(msgid) (msgid) +#endif + +#ifndef gettext_noop +/* This define is so xgettext can find the internationalizable + strings. */ +# define gettext_noop(String) String +#endif + +#if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC +# define RE_ENABLE_I18N +#endif + +/* Number of ASCII characters. */ +#define ASCII_CHARS 0x80 + +/* Number of single byte characters. */ +#define SBC_MAX (UCHAR_MAX + 1) + +#define COLL_ELEM_LEN_MAX 8 + +/* The character which represents newline. */ +#define NEWLINE_CHAR '\n' +#define WIDE_NEWLINE_CHAR L'\n' + +/* Rename to standard API for using out of glibc. */ +#ifndef _LIBC +# undef __wctype +# undef __iswalnum +# undef __iswctype +# undef __towlower +# undef __towupper +# define __wctype wctype +# define __iswalnum iswalnum +# define __iswctype iswctype +# define __towlower towlower +# define __towupper towupper +# define __btowc btowc +# define __mbrtowc mbrtowc +# define __wcrtomb wcrtomb +# define __regfree regfree +#endif /* not _LIBC */ + +#ifndef SSIZE_MAX +# define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2)) +#endif + +/* The type of indexes into strings. This is signed, not size_t, + since the API requires indexes to fit in regoff_t anyway, and using + signed integers makes the code a bit smaller and presumably faster. + The traditional GNU regex implementation uses int for indexes. + The POSIX-compatible implementation uses a possibly-wider type. + The name 'Idx' is three letters to minimize the hassle of + reindenting a lot of regex code that formerly used 'int'. */ +typedef regoff_t Idx; +#ifdef _REGEX_LARGE_OFFSETS +# define IDX_MAX SSIZE_MAX +#else +# define IDX_MAX INT_MAX +#endif + +/* A hash value, suitable for computing hash tables. */ +typedef __re_size_t re_hashval_t; + +/* An integer used to represent a set of bits. It must be unsigned, + and must be at least as wide as unsigned int. */ +typedef unsigned long int bitset_word_t; +/* All bits set in a bitset_word_t. */ +#define BITSET_WORD_MAX ULONG_MAX + +/* Number of bits in a bitset_word_t. For portability to hosts with + padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; + instead, deduce it directly from BITSET_WORD_MAX. Avoid + greater-than-32-bit integers and unconditional shifts by more than + 31 bits, as they're not portable. */ +#if BITSET_WORD_MAX == 0xffffffffUL +# define BITSET_WORD_BITS 32 +#elif BITSET_WORD_MAX >> 31 >> 4 == 1 +# define BITSET_WORD_BITS 36 +#elif BITSET_WORD_MAX >> 31 >> 16 == 1 +# define BITSET_WORD_BITS 48 +#elif BITSET_WORD_MAX >> 31 >> 28 == 1 +# define BITSET_WORD_BITS 60 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 +# define BITSET_WORD_BITS 64 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 +# define BITSET_WORD_BITS 72 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 +# define BITSET_WORD_BITS 128 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 +# define BITSET_WORD_BITS 256 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 +# define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ +# if BITSET_WORD_BITS <= SBC_MAX +# error "Invalid SBC_MAX" +# endif +#else +# error "Add case for new bitset_word_t size" +#endif + +/* Number of bitset_word_t values in a bitset_t. */ +#define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) + +typedef bitset_word_t bitset_t[BITSET_WORDS]; +typedef bitset_word_t *re_bitset_ptr_t; +typedef const bitset_word_t *re_const_bitset_ptr_t; + +#define PREV_WORD_CONSTRAINT 0x0001 +#define PREV_NOTWORD_CONSTRAINT 0x0002 +#define NEXT_WORD_CONSTRAINT 0x0004 +#define NEXT_NOTWORD_CONSTRAINT 0x0008 +#define PREV_NEWLINE_CONSTRAINT 0x0010 +#define NEXT_NEWLINE_CONSTRAINT 0x0020 +#define PREV_BEGBUF_CONSTRAINT 0x0040 +#define NEXT_ENDBUF_CONSTRAINT 0x0080 +#define WORD_DELIM_CONSTRAINT 0x0100 +#define NOT_WORD_DELIM_CONSTRAINT 0x0200 + +typedef enum +{ + INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, + WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, + WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, + INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, + LINE_FIRST = PREV_NEWLINE_CONSTRAINT, + LINE_LAST = NEXT_NEWLINE_CONSTRAINT, + BUF_FIRST = PREV_BEGBUF_CONSTRAINT, + BUF_LAST = NEXT_ENDBUF_CONSTRAINT, + WORD_DELIM = WORD_DELIM_CONSTRAINT, + NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT +} re_context_type; + +typedef struct +{ + Idx alloc; + Idx nelem; + Idx *elems; +} re_node_set; + +typedef enum +{ + NON_TYPE = 0, + + /* Node type, These are used by token, node, tree. */ + CHARACTER = 1, + END_OF_RE = 2, + SIMPLE_BRACKET = 3, + OP_BACK_REF = 4, + OP_PERIOD = 5, +#ifdef RE_ENABLE_I18N + COMPLEX_BRACKET = 6, + OP_UTF8_PERIOD = 7, +#endif /* RE_ENABLE_I18N */ + + /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used + when the debugger shows values of this enum type. */ +#define EPSILON_BIT 8 + OP_OPEN_SUBEXP = EPSILON_BIT | 0, + OP_CLOSE_SUBEXP = EPSILON_BIT | 1, + OP_ALT = EPSILON_BIT | 2, + OP_DUP_ASTERISK = EPSILON_BIT | 3, + ANCHOR = EPSILON_BIT | 4, + + /* Tree type, these are used only by tree. */ + CONCAT = 16, + SUBEXP = 17, + + /* Token type, these are used only by token. */ + OP_DUP_PLUS = 18, + OP_DUP_QUESTION, + OP_OPEN_BRACKET, + OP_CLOSE_BRACKET, + OP_CHARSET_RANGE, + OP_OPEN_DUP_NUM, + OP_CLOSE_DUP_NUM, + OP_NON_MATCH_LIST, + OP_OPEN_COLL_ELEM, + OP_CLOSE_COLL_ELEM, + OP_OPEN_EQUIV_CLASS, + OP_CLOSE_EQUIV_CLASS, + OP_OPEN_CHAR_CLASS, + OP_CLOSE_CHAR_CLASS, + OP_WORD, + OP_NOTWORD, + OP_SPACE, + OP_NOTSPACE, + BACK_SLASH + +} re_token_type_t; + +#ifdef RE_ENABLE_I18N +typedef struct +{ + /* Multibyte characters. */ + wchar_t *mbchars; + + /* Collating symbols. */ +# ifdef _LIBC + int32_t *coll_syms; +# endif + + /* Equivalence classes. */ +# ifdef _LIBC + int32_t *equiv_classes; +# endif + + /* Range expressions. */ +# ifdef _LIBC + uint32_t *range_starts; + uint32_t *range_ends; +# else /* not _LIBC */ + wchar_t *range_starts; + wchar_t *range_ends; +# endif /* not _LIBC */ + + /* Character classes. */ + wctype_t *char_classes; + + /* If this character set is the non-matching list. */ + unsigned int non_match : 1; + + /* # of multibyte characters. */ + Idx nmbchars; + + /* # of collating symbols. */ + Idx ncoll_syms; + + /* # of equivalence classes. */ + Idx nequiv_classes; + + /* # of range expressions. */ + Idx nranges; + + /* # of character classes. */ + Idx nchar_classes; +} re_charset_t; +#endif /* RE_ENABLE_I18N */ + +typedef struct +{ + union + { + unsigned char c; /* for CHARACTER */ + re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; /* for COMPLEX_BRACKET */ +#endif /* RE_ENABLE_I18N */ + Idx idx; /* for BACK_REF */ + re_context_type ctx_type; /* for ANCHOR */ + } opr; +#if __GNUC__ >= 2 && !defined __STRICT_ANSI__ + re_token_type_t type : 8; +#else + re_token_type_t type; +#endif + unsigned int constraint : 10; /* context constraint */ + unsigned int duplicated : 1; + unsigned int opt_subexp : 1; +#ifdef RE_ENABLE_I18N + unsigned int accept_mb : 1; + /* These 2 bits can be moved into the union if needed (e.g. if running out + of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ + unsigned int mb_partial : 1; +#endif + unsigned int word_char : 1; +} re_token_t; + +#define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) + +struct re_string_t +{ + /* Indicate the raw buffer which is the original string passed as an + argument of regexec(), re_search(), etc.. */ + const unsigned char *raw_mbs; + /* Store the multibyte string. In case of "case insensitive mode" like + REG_ICASE, upper cases of the string are stored, otherwise MBS points + the same address that RAW_MBS points. */ + unsigned char *mbs; +#ifdef RE_ENABLE_I18N + /* Store the wide character string which is corresponding to MBS. */ + wint_t *wcs; + Idx *offsets; + mbstate_t cur_state; +#endif + /* Index in RAW_MBS. Each character mbs[i] corresponds to + raw_mbs[raw_mbs_idx + i]. */ + Idx raw_mbs_idx; + /* The length of the valid characters in the buffers. */ + Idx valid_len; + /* The corresponding number of bytes in raw_mbs array. */ + Idx valid_raw_len; + /* The length of the buffers MBS and WCS. */ + Idx bufs_len; + /* The index in MBS, which is updated by re_string_fetch_byte. */ + Idx cur_idx; + /* length of RAW_MBS array. */ + Idx raw_len; + /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ + Idx len; + /* End of the buffer may be shorter than its length in the cases such + as re_match_2, re_search_2. Then, we use STOP for end of the buffer + instead of LEN. */ + Idx raw_stop; + /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ + Idx stop; + + /* The context of mbs[0]. We store the context independently, since + the context of mbs[0] may be different from raw_mbs[0], which is + the beginning of the input string. */ + unsigned int tip_context; + /* The translation passed as a part of an argument of re_compile_pattern. */ + RE_TRANSLATE_TYPE trans; + /* Copy of re_dfa_t's word_char. */ + re_const_bitset_ptr_t word_char; + /* true if REG_ICASE. */ + unsigned char icase; + unsigned char is_utf8; + unsigned char map_notascii; + unsigned char mbs_allocated; + unsigned char offsets_needed; + unsigned char newline_anchor; + unsigned char word_ops_used; + int mb_cur_max; +}; +typedef struct re_string_t re_string_t; + + +struct re_dfa_t; +typedef struct re_dfa_t re_dfa_t; + +#ifndef _LIBC +# define IS_IN(libc) false +#endif + +#define re_string_peek_byte(pstr, offset) \ + ((pstr)->mbs[(pstr)->cur_idx + offset]) +#define re_string_fetch_byte(pstr) \ + ((pstr)->mbs[(pstr)->cur_idx++]) +#define re_string_first_byte(pstr, idx) \ + ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) +#define re_string_is_single_byte_char(pstr, idx) \ + ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ + || (pstr)->wcs[(idx) + 1] != WEOF)) +#define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) +#define re_string_cur_idx(pstr) ((pstr)->cur_idx) +#define re_string_get_buffer(pstr) ((pstr)->mbs) +#define re_string_length(pstr) ((pstr)->len) +#define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) +#define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) +#define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) + +#if defined _LIBC || HAVE_ALLOCA +# include <alloca.h> +#endif + +#ifndef _LIBC +# if HAVE_ALLOCA +/* The OS usually guarantees only one guard page at the bottom of the stack, + and a page size can be as small as 4096 bytes. So we cannot safely + allocate anything larger than 4096 bytes. Also care for the possibility + of a few compiler-allocated temporary stack slots. */ +# define __libc_use_alloca(n) ((n) < 4032) +# else +/* alloca is implemented with malloc, so just use malloc. */ +# define __libc_use_alloca(n) 0 +# undef alloca +# define alloca(n) malloc (n) +# endif +#endif + +#ifdef _LIBC +# define MALLOC_0_IS_NONNULL 1 +#elif !defined MALLOC_0_IS_NONNULL +# define MALLOC_0_IS_NONNULL 0 +#endif + +#ifndef MAX +# define MAX(a,b) ((a) < (b) ? (b) : (a)) +#endif +#ifndef MIN +# define MIN(a,b) ((a) < (b) ? (a) : (b)) +#endif + +#define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) +#define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) +#define re_free(p) free (p) + +struct bin_tree_t +{ + struct bin_tree_t *parent; + struct bin_tree_t *left; + struct bin_tree_t *right; + struct bin_tree_t *first; + struct bin_tree_t *next; + + re_token_t token; + + /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. + Otherwise 'type' indicate the type of this node. */ + Idx node_idx; +}; +typedef struct bin_tree_t bin_tree_t; + +#define BIN_TREE_STORAGE_SIZE \ + ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) + +struct bin_tree_storage_t +{ + struct bin_tree_storage_t *next; + bin_tree_t data[BIN_TREE_STORAGE_SIZE]; +}; +typedef struct bin_tree_storage_t bin_tree_storage_t; + +#define CONTEXT_WORD 1 +#define CONTEXT_NEWLINE (CONTEXT_WORD << 1) +#define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) +#define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) + +#define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) +#define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) +#define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) +#define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) +#define IS_ORDINARY_CONTEXT(c) ((c) == 0) + +#define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') +#define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) +#define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') +#define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) + +#define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ + ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ + || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ + || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ + || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) + +#define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ + ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ + || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ + || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ + || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) + +struct re_dfastate_t +{ + re_hashval_t hash; + re_node_set nodes; + re_node_set non_eps_nodes; + re_node_set inveclosure; + re_node_set *entrance_nodes; + struct re_dfastate_t **trtable, **word_trtable; + unsigned int context : 4; + unsigned int halt : 1; + /* If this state can accept "multi byte". + Note that we refer to multibyte characters, and multi character + collating elements as "multi byte". */ + unsigned int accept_mb : 1; + /* If this state has backreference node(s). */ + unsigned int has_backref : 1; + unsigned int has_constraint : 1; +}; +typedef struct re_dfastate_t re_dfastate_t; + +struct re_state_table_entry +{ + Idx num; + Idx alloc; + re_dfastate_t **array; +}; + +/* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ + +typedef struct +{ + Idx next_idx; + Idx alloc; + re_dfastate_t **array; +} state_array_t; + +/* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ + +typedef struct +{ + Idx node; + Idx str_idx; /* The position NODE match at. */ + state_array_t path; +} re_sub_match_last_t; + +/* Store information about the node NODE whose type is OP_OPEN_SUBEXP. + And information about the node, whose type is OP_CLOSE_SUBEXP, + corresponding to NODE is stored in LASTS. */ + +typedef struct +{ + Idx str_idx; + Idx node; + state_array_t *path; + Idx alasts; /* Allocation size of LASTS. */ + Idx nlasts; /* The number of LASTS. */ + re_sub_match_last_t **lasts; +} re_sub_match_top_t; + +struct re_backref_cache_entry +{ + Idx node; + Idx str_idx; + Idx subexp_from; + Idx subexp_to; + char more; + char unused; + unsigned short int eps_reachable_subexps_map; +}; + +typedef struct +{ + /* The string object corresponding to the input string. */ + re_string_t input; +#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) + const re_dfa_t *const dfa; +#else + const re_dfa_t *dfa; +#endif + /* EFLAGS of the argument of regexec. */ + int eflags; + /* Where the matching ends. */ + Idx match_last; + Idx last_node; + /* The state log used by the matcher. */ + re_dfastate_t **state_log; + Idx state_log_top; + /* Back reference cache. */ + Idx nbkref_ents; + Idx abkref_ents; + struct re_backref_cache_entry *bkref_ents; + int max_mb_elem_len; + Idx nsub_tops; + Idx asub_tops; + re_sub_match_top_t **sub_tops; +} re_match_context_t; + +typedef struct +{ + re_dfastate_t **sifted_states; + re_dfastate_t **limited_states; + Idx last_node; + Idx last_str_idx; + re_node_set limits; +} re_sift_context_t; + +struct re_fail_stack_ent_t +{ + Idx idx; + Idx node; + regmatch_t *regs; + re_node_set eps_via_nodes; +}; + +struct re_fail_stack_t +{ + Idx num; + Idx alloc; + struct re_fail_stack_ent_t *stack; +}; + +struct re_dfa_t +{ + re_token_t *nodes; + size_t nodes_alloc; + size_t nodes_len; + Idx *nexts; + Idx *org_indices; + re_node_set *edests; + re_node_set *eclosures; + re_node_set *inveclosures; + struct re_state_table_entry *state_table; + re_dfastate_t *init_state; + re_dfastate_t *init_state_word; + re_dfastate_t *init_state_nl; + re_dfastate_t *init_state_begbuf; + bin_tree_t *str_tree; + bin_tree_storage_t *str_tree_storage; + re_bitset_ptr_t sb_char; + int str_tree_storage_idx; + + /* number of subexpressions 're_nsub' is in regex_t. */ + re_hashval_t state_hash_mask; + Idx init_node; + Idx nbackref; /* The number of backreference in this dfa. */ + + /* Bitmap expressing which backreference is used. */ + bitset_word_t used_bkref_map; + bitset_word_t completed_bkref_map; + + unsigned int has_plural_match : 1; + /* If this dfa has "multibyte node", which is a backreference or + a node which can accept multibyte character or multi character + collating element. */ + unsigned int has_mb_node : 1; + unsigned int is_utf8 : 1; + unsigned int map_notascii : 1; + unsigned int word_ops_used : 1; + int mb_cur_max; + bitset_t word_char; + reg_syntax_t syntax; + Idx *subexp_map; +#ifdef DEBUG + char* re_str; +#endif + lock_define (lock) +}; + +#define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) +#define re_node_set_remove(set,id) \ + (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) +#define re_node_set_empty(p) ((p)->nelem = 0) +#define re_node_set_free(set) re_free ((set)->elems) + + +typedef enum +{ + SB_CHAR, + MB_CHAR, + EQUIV_CLASS, + COLL_SYM, + CHAR_CLASS +} bracket_elem_type; + +typedef struct +{ + bracket_elem_type type; + union + { + unsigned char ch; + unsigned char *name; + wchar_t wch; + } opr; +} bracket_elem_t; + + +/* Functions for bitset_t operation. */ + +static inline void +bitset_set (bitset_t set, Idx i) +{ + set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; +} + +static inline void +bitset_clear (bitset_t set, Idx i) +{ + set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); +} + +static inline bool +bitset_contain (const bitset_t set, Idx i) +{ + return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; +} + +static inline void +bitset_empty (bitset_t set) +{ + memset (set, '\0', sizeof (bitset_t)); +} + +static inline void +bitset_set_all (bitset_t set) +{ + memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); + if (SBC_MAX % BITSET_WORD_BITS != 0) + set[BITSET_WORDS - 1] = + ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; +} + +static inline void +bitset_copy (bitset_t dest, const bitset_t src) +{ + memcpy (dest, src, sizeof (bitset_t)); +} + +static inline void +bitset_not (bitset_t set) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) + set[bitset_i] = ~set[bitset_i]; + if (SBC_MAX % BITSET_WORD_BITS != 0) + set[BITSET_WORDS - 1] = + ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) + & ~set[BITSET_WORDS - 1]); +} + +static inline void +bitset_merge (bitset_t dest, const bitset_t src) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) + dest[bitset_i] |= src[bitset_i]; +} + +static inline void +bitset_mask (bitset_t dest, const bitset_t src) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) + dest[bitset_i] &= src[bitset_i]; +} + +#ifdef RE_ENABLE_I18N +/* Functions for re_string. */ +static int +__attribute__ ((pure, unused)) +re_string_char_size_at (const re_string_t *pstr, Idx idx) +{ + int byte_idx; + if (pstr->mb_cur_max == 1) + return 1; + for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) + if (pstr->wcs[idx + byte_idx] != WEOF) + break; + return byte_idx; +} + +static wint_t +__attribute__ ((pure, unused)) +re_string_wchar_at (const re_string_t *pstr, Idx idx) +{ + if (pstr->mb_cur_max == 1) + return (wint_t) pstr->mbs[idx]; + return (wint_t) pstr->wcs[idx]; +} + +# ifdef _LIBC +# include <locale/weight.h> +# endif + +static int +__attribute__ ((pure, unused)) +re_string_elem_size_at (const re_string_t *pstr, Idx idx) +{ +# ifdef _LIBC + const unsigned char *p, *extra; + const int32_t *table, *indirect; + uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + + if (nrules != 0) + { + table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_INDIRECTMB); + p = pstr->mbs + idx; + findidx (table, indirect, extra, &p, pstr->len - idx); + return p - pstr->mbs - idx; + } + else +# endif /* _LIBC */ + return 1; +} +#endif /* RE_ENABLE_I18N */ + +#ifndef FALLTHROUGH +# if __GNUC__ < 7 +# define FALLTHROUGH ((void) 0) +# else +# define FALLTHROUGH __attribute__ ((__fallthrough__)) +# endif +#endif + +#endif /* _REGEX_INTERNAL_H */ diff --git a/lib/regexec.c b/lib/regexec.c new file mode 100644 index 00000000000..f464869fb03 --- /dev/null +++ b/lib/regexec.c @@ -0,0 +1,4336 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +static reg_errcode_t match_ctx_init (re_match_context_t *cache, int eflags, + Idx n); +static void match_ctx_clean (re_match_context_t *mctx); +static void match_ctx_free (re_match_context_t *cache); +static reg_errcode_t match_ctx_add_entry (re_match_context_t *cache, Idx node, + Idx str_idx, Idx from, Idx to); +static Idx search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx); +static reg_errcode_t match_ctx_add_subtop (re_match_context_t *mctx, Idx node, + Idx str_idx); +static re_sub_match_last_t * match_ctx_add_sublast (re_sub_match_top_t *subtop, + Idx node, Idx str_idx); +static void sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts, + re_dfastate_t **limited_sts, Idx last_node, + Idx last_str_idx); +static reg_errcode_t re_search_internal (const regex_t *preg, + const char *string, Idx length, + Idx start, Idx last_start, Idx stop, + size_t nmatch, regmatch_t pmatch[], + int eflags); +static regoff_t re_search_2_stub (struct re_pattern_buffer *bufp, + const char *string1, Idx length1, + const char *string2, Idx length2, + Idx start, regoff_t range, + struct re_registers *regs, + Idx stop, bool ret_len); +static regoff_t re_search_stub (struct re_pattern_buffer *bufp, + const char *string, Idx length, Idx start, + regoff_t range, Idx stop, + struct re_registers *regs, + bool ret_len); +static unsigned re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, + Idx nregs, int regs_allocated); +static reg_errcode_t prune_impossible_nodes (re_match_context_t *mctx); +static Idx check_matching (re_match_context_t *mctx, bool fl_longest_match, + Idx *p_match_first); +static Idx check_halt_state_context (const re_match_context_t *mctx, + const re_dfastate_t *state, Idx idx); +static void update_regs (const re_dfa_t *dfa, regmatch_t *pmatch, + regmatch_t *prev_idx_match, Idx cur_node, + Idx cur_idx, Idx nmatch); +static reg_errcode_t push_fail_stack (struct re_fail_stack_t *fs, + Idx str_idx, Idx dest_node, Idx nregs, + regmatch_t *regs, + re_node_set *eps_via_nodes); +static reg_errcode_t set_regs (const regex_t *preg, + const re_match_context_t *mctx, + size_t nmatch, regmatch_t *pmatch, + bool fl_backtrack); +static reg_errcode_t free_fail_stack_return (struct re_fail_stack_t *fs); + +#ifdef RE_ENABLE_I18N +static int sift_states_iter_mb (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx node_idx, Idx str_idx, Idx max_str_idx); +#endif /* RE_ENABLE_I18N */ +static reg_errcode_t sift_states_backward (const re_match_context_t *mctx, + re_sift_context_t *sctx); +static reg_errcode_t build_sifted_states (const re_match_context_t *mctx, + re_sift_context_t *sctx, Idx str_idx, + re_node_set *cur_dest); +static reg_errcode_t update_cur_sifted_state (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx str_idx, + re_node_set *dest_nodes); +static reg_errcode_t add_epsilon_src_nodes (const re_dfa_t *dfa, + re_node_set *dest_nodes, + const re_node_set *candidates); +static bool check_dst_limits (const re_match_context_t *mctx, + const re_node_set *limits, + Idx dst_node, Idx dst_idx, Idx src_node, + Idx src_idx); +static int check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, + int boundaries, Idx subexp_idx, + Idx from_node, Idx bkref_idx); +static int check_dst_limits_calc_pos (const re_match_context_t *mctx, + Idx limit, Idx subexp_idx, + Idx node, Idx str_idx, + Idx bkref_idx); +static reg_errcode_t check_subexp_limits (const re_dfa_t *dfa, + re_node_set *dest_nodes, + const re_node_set *candidates, + re_node_set *limits, + struct re_backref_cache_entry *bkref_ents, + Idx str_idx); +static reg_errcode_t sift_states_bkref (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx str_idx, const re_node_set *candidates); +static reg_errcode_t merge_state_array (const re_dfa_t *dfa, + re_dfastate_t **dst, + re_dfastate_t **src, Idx num); +static re_dfastate_t *find_recover_state (reg_errcode_t *err, + re_match_context_t *mctx); +static re_dfastate_t *transit_state (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *state); +static re_dfastate_t *merge_state_with_log (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *next_state); +static reg_errcode_t check_subexp_matching_top (re_match_context_t *mctx, + re_node_set *cur_nodes, + Idx str_idx); +#if 0 +static re_dfastate_t *transit_state_sb (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *pstate); +#endif +#ifdef RE_ENABLE_I18N +static reg_errcode_t transit_state_mb (re_match_context_t *mctx, + re_dfastate_t *pstate); +#endif /* RE_ENABLE_I18N */ +static reg_errcode_t transit_state_bkref (re_match_context_t *mctx, + const re_node_set *nodes); +static reg_errcode_t get_subexp (re_match_context_t *mctx, + Idx bkref_node, Idx bkref_str_idx); +static reg_errcode_t get_subexp_sub (re_match_context_t *mctx, + const re_sub_match_top_t *sub_top, + re_sub_match_last_t *sub_last, + Idx bkref_node, Idx bkref_str); +static Idx find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes, + Idx subexp_idx, int type); +static reg_errcode_t check_arrival (re_match_context_t *mctx, + state_array_t *path, Idx top_node, + Idx top_str, Idx last_node, Idx last_str, + int type); +static reg_errcode_t check_arrival_add_next_nodes (re_match_context_t *mctx, + Idx str_idx, + re_node_set *cur_nodes, + re_node_set *next_nodes); +static reg_errcode_t check_arrival_expand_ecl (const re_dfa_t *dfa, + re_node_set *cur_nodes, + Idx ex_subexp, int type); +static reg_errcode_t check_arrival_expand_ecl_sub (const re_dfa_t *dfa, + re_node_set *dst_nodes, + Idx target, Idx ex_subexp, + int type); +static reg_errcode_t expand_bkref_cache (re_match_context_t *mctx, + re_node_set *cur_nodes, Idx cur_str, + Idx subexp_num, int type); +static bool build_trtable (const re_dfa_t *dfa, re_dfastate_t *state); +#ifdef RE_ENABLE_I18N +static int check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx, + const re_string_t *input, Idx idx); +# ifdef _LIBC +static unsigned int find_collation_sequence_value (const unsigned char *mbs, + size_t name_len); +# endif /* _LIBC */ +#endif /* RE_ENABLE_I18N */ +static Idx group_nodes_into_DFAstates (const re_dfa_t *dfa, + const re_dfastate_t *state, + re_node_set *states_node, + bitset_t *states_ch); +static bool check_node_accept (const re_match_context_t *mctx, + const re_token_t *node, Idx idx); +static reg_errcode_t extend_buffers (re_match_context_t *mctx, int min_len); + +/* Entry point for POSIX code. */ + +/* regexec searches for a given pattern, specified by PREG, in the + string STRING. + + If NMATCH is zero or REG_NOSUB was set in the cflags argument to + 'regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at + least NMATCH elements, and we set them to the offsets of the + corresponding matched substrings. + + EFLAGS specifies "execution flags" which affect matching: if + REG_NOTBOL is set, then ^ does not match at the beginning of the + string; if REG_NOTEOL is set, then $ does not match at the end. + + We return 0 if we find a match and REG_NOMATCH if not. */ + +int +regexec (const regex_t *__restrict preg, const char *__restrict string, + size_t nmatch, regmatch_t pmatch[], int eflags) +{ + reg_errcode_t err; + Idx start, length; + re_dfa_t *dfa = preg->buffer; + + if (eflags & ~(REG_NOTBOL | REG_NOTEOL | REG_STARTEND)) + return REG_BADPAT; + + if (eflags & REG_STARTEND) + { + start = pmatch[0].rm_so; + length = pmatch[0].rm_eo; + } + else + { + start = 0; + length = strlen (string); + } + + lock_lock (dfa->lock); + if (preg->no_sub) + err = re_search_internal (preg, string, length, start, length, + length, 0, NULL, eflags); + else + err = re_search_internal (preg, string, length, start, length, + length, nmatch, pmatch, eflags); + lock_unlock (dfa->lock); + return err != REG_NOERROR; +} + +#ifdef _LIBC +libc_hidden_def (__regexec) + +# include <shlib-compat.h> +versioned_symbol (libc, __regexec, regexec, GLIBC_2_3_4); + +# if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3_4) +__typeof__ (__regexec) __compat_regexec; + +int +attribute_compat_text_section +__compat_regexec (const regex_t *__restrict preg, + const char *__restrict string, size_t nmatch, + regmatch_t pmatch[], int eflags) +{ + return regexec (preg, string, nmatch, pmatch, + eflags & (REG_NOTBOL | REG_NOTEOL)); +} +compat_symbol (libc, __compat_regexec, regexec, GLIBC_2_0); +# endif +#endif + +/* Entry points for GNU code. */ + +/* re_match, re_search, re_match_2, re_search_2 + + The former two functions operate on STRING with length LENGTH, + while the later two operate on concatenation of STRING1 and STRING2 + with lengths LENGTH1 and LENGTH2, respectively. + + re_match() matches the compiled pattern in BUFP against the string, + starting at index START. + + re_search() first tries matching at index START, then it tries to match + starting from index START + 1, and so on. The last start position tried + is START + RANGE. (Thus RANGE = 0 forces re_search to operate the same + way as re_match().) + + The parameter STOP of re_{match,search}_2 specifies that no match exceeding + the first STOP characters of the concatenation of the strings should be + concerned. + + If REGS is not NULL, and BUFP->no_sub is not set, the offsets of the match + and all groups is stored in REGS. (For the "_2" variants, the offsets are + computed relative to the concatenation, not relative to the individual + strings.) + + On success, re_match* functions return the length of the match, re_search* + return the position of the start of the match. Return value -1 means no + match was found and -2 indicates an internal error. */ + +regoff_t +re_match (struct re_pattern_buffer *bufp, const char *string, Idx length, + Idx start, struct re_registers *regs) +{ + return re_search_stub (bufp, string, length, start, 0, length, regs, true); +} +#ifdef _LIBC +weak_alias (__re_match, re_match) +#endif + +regoff_t +re_search (struct re_pattern_buffer *bufp, const char *string, Idx length, + Idx start, regoff_t range, struct re_registers *regs) +{ + return re_search_stub (bufp, string, length, start, range, length, regs, + false); +} +#ifdef _LIBC +weak_alias (__re_search, re_search) +#endif + +regoff_t +re_match_2 (struct re_pattern_buffer *bufp, const char *string1, Idx length1, + const char *string2, Idx length2, Idx start, + struct re_registers *regs, Idx stop) +{ + return re_search_2_stub (bufp, string1, length1, string2, length2, + start, 0, regs, stop, true); +} +#ifdef _LIBC +weak_alias (__re_match_2, re_match_2) +#endif + +regoff_t +re_search_2 (struct re_pattern_buffer *bufp, const char *string1, Idx length1, + const char *string2, Idx length2, Idx start, regoff_t range, + struct re_registers *regs, Idx stop) +{ + return re_search_2_stub (bufp, string1, length1, string2, length2, + start, range, regs, stop, false); +} +#ifdef _LIBC +weak_alias (__re_search_2, re_search_2) +#endif + +static regoff_t +re_search_2_stub (struct re_pattern_buffer *bufp, const char *string1, + Idx length1, const char *string2, Idx length2, Idx start, + regoff_t range, struct re_registers *regs, + Idx stop, bool ret_len) +{ + const char *str; + regoff_t rval; + Idx len; + char *s = NULL; + + if (__glibc_unlikely ((length1 < 0 || length2 < 0 || stop < 0 + || INT_ADD_WRAPV (length1, length2, &len)))) + return -2; + + /* Concatenate the strings. */ + if (length2 > 0) + if (length1 > 0) + { + s = re_malloc (char, len); + + if (__glibc_unlikely (s == NULL)) + return -2; +#ifdef _LIBC + memcpy (__mempcpy (s, string1, length1), string2, length2); +#else + memcpy (s, string1, length1); + memcpy (s + length1, string2, length2); +#endif + str = s; + } + else + str = string2; + else + str = string1; + + rval = re_search_stub (bufp, str, len, start, range, stop, regs, + ret_len); + re_free (s); + return rval; +} + +/* The parameters have the same meaning as those of re_search. + Additional parameters: + If RET_LEN is true the length of the match is returned (re_match style); + otherwise the position of the match is returned. */ + +static regoff_t +re_search_stub (struct re_pattern_buffer *bufp, const char *string, Idx length, + Idx start, regoff_t range, Idx stop, struct re_registers *regs, + bool ret_len) +{ + reg_errcode_t result; + regmatch_t *pmatch; + Idx nregs; + regoff_t rval; + int eflags = 0; + re_dfa_t *dfa = bufp->buffer; + Idx last_start = start + range; + + /* Check for out-of-range. */ + if (__glibc_unlikely (start < 0 || start > length)) + return -1; + if (__glibc_unlikely (length < last_start + || (0 <= range && last_start < start))) + last_start = length; + else if (__glibc_unlikely (last_start < 0 + || (range < 0 && start <= last_start))) + last_start = 0; + + lock_lock (dfa->lock); + + eflags |= (bufp->not_bol) ? REG_NOTBOL : 0; + eflags |= (bufp->not_eol) ? REG_NOTEOL : 0; + + /* Compile fastmap if we haven't yet. */ + if (start < last_start && bufp->fastmap != NULL && !bufp->fastmap_accurate) + re_compile_fastmap (bufp); + + if (__glibc_unlikely (bufp->no_sub)) + regs = NULL; + + /* We need at least 1 register. */ + if (regs == NULL) + nregs = 1; + else if (__glibc_unlikely (bufp->regs_allocated == REGS_FIXED + && regs->num_regs <= bufp->re_nsub)) + { + nregs = regs->num_regs; + if (__glibc_unlikely (nregs < 1)) + { + /* Nothing can be copied to regs. */ + regs = NULL; + nregs = 1; + } + } + else + nregs = bufp->re_nsub + 1; + pmatch = re_malloc (regmatch_t, nregs); + if (__glibc_unlikely (pmatch == NULL)) + { + rval = -2; + goto out; + } + + result = re_search_internal (bufp, string, length, start, last_start, stop, + nregs, pmatch, eflags); + + rval = 0; + + /* I hope we needn't fill their regs with -1's when no match was found. */ + if (result != REG_NOERROR) + rval = result == REG_NOMATCH ? -1 : -2; + else if (regs != NULL) + { + /* If caller wants register contents data back, copy them. */ + bufp->regs_allocated = re_copy_regs (regs, pmatch, nregs, + bufp->regs_allocated); + if (__glibc_unlikely (bufp->regs_allocated == REGS_UNALLOCATED)) + rval = -2; + } + + if (__glibc_likely (rval == 0)) + { + if (ret_len) + { + assert (pmatch[0].rm_so == start); + rval = pmatch[0].rm_eo - start; + } + else + rval = pmatch[0].rm_so; + } + re_free (pmatch); + out: + lock_unlock (dfa->lock); + return rval; +} + +static unsigned +re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, Idx nregs, + int regs_allocated) +{ + int rval = REGS_REALLOCATE; + Idx i; + Idx need_regs = nregs + 1; + /* We need one extra element beyond 'num_regs' for the '-1' marker GNU code + uses. */ + + /* Have the register data arrays been allocated? */ + if (regs_allocated == REGS_UNALLOCATED) + { /* No. So allocate them with malloc. */ + regs->start = re_malloc (regoff_t, need_regs); + if (__glibc_unlikely (regs->start == NULL)) + return REGS_UNALLOCATED; + regs->end = re_malloc (regoff_t, need_regs); + if (__glibc_unlikely (regs->end == NULL)) + { + re_free (regs->start); + return REGS_UNALLOCATED; + } + regs->num_regs = need_regs; + } + else if (regs_allocated == REGS_REALLOCATE) + { /* Yes. If we need more elements than were already + allocated, reallocate them. If we need fewer, just + leave it alone. */ + if (__glibc_unlikely (need_regs > regs->num_regs)) + { + regoff_t *new_start = re_realloc (regs->start, regoff_t, need_regs); + regoff_t *new_end; + if (__glibc_unlikely (new_start == NULL)) + return REGS_UNALLOCATED; + new_end = re_realloc (regs->end, regoff_t, need_regs); + if (__glibc_unlikely (new_end == NULL)) + { + re_free (new_start); + return REGS_UNALLOCATED; + } + regs->start = new_start; + regs->end = new_end; + regs->num_regs = need_regs; + } + } + else + { + assert (regs_allocated == REGS_FIXED); + /* This function may not be called with REGS_FIXED and nregs too big. */ + assert (regs->num_regs >= nregs); + rval = REGS_FIXED; + } + + /* Copy the regs. */ + for (i = 0; i < nregs; ++i) + { + regs->start[i] = pmatch[i].rm_so; + regs->end[i] = pmatch[i].rm_eo; + } + for ( ; i < regs->num_regs; ++i) + regs->start[i] = regs->end[i] = -1; + + return rval; +} + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use + this memory for recording register information. STARTS and ENDS + must be allocated using the malloc library routine, and must each + be at least NUM_REGS * sizeof (regoff_t) bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ + +void +re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs, + __re_size_t num_regs, regoff_t *starts, regoff_t *ends) +{ + if (num_regs) + { + bufp->regs_allocated = REGS_REALLOCATE; + regs->num_regs = num_regs; + regs->start = starts; + regs->end = ends; + } + else + { + bufp->regs_allocated = REGS_UNALLOCATED; + regs->num_regs = 0; + regs->start = regs->end = NULL; + } +} +#ifdef _LIBC +weak_alias (__re_set_registers, re_set_registers) +#endif + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them unless specifically requested. */ + +#if defined _REGEX_RE_COMP || defined _LIBC +int +# ifdef _LIBC +weak_function +# endif +re_exec (const char *s) +{ + return 0 == regexec (&re_comp_buf, s, 0, NULL, 0); +} +#endif /* _REGEX_RE_COMP */ + +/* Internal entry point. */ + +/* Searches for a compiled pattern PREG in the string STRING, whose + length is LENGTH. NMATCH, PMATCH, and EFLAGS have the same + meaning as with regexec. LAST_START is START + RANGE, where + START and RANGE have the same meaning as with re_search. + Return REG_NOERROR if we find a match, and REG_NOMATCH if not, + otherwise return the error code. + Note: We assume front end functions already check ranges. + (0 <= LAST_START && LAST_START <= LENGTH) */ + +static reg_errcode_t +__attribute_warn_unused_result__ +re_search_internal (const regex_t *preg, const char *string, Idx length, + Idx start, Idx last_start, Idx stop, size_t nmatch, + regmatch_t pmatch[], int eflags) +{ + reg_errcode_t err; + const re_dfa_t *dfa = preg->buffer; + Idx left_lim, right_lim; + int incr; + bool fl_longest_match; + int match_kind; + Idx match_first; + Idx match_last = -1; + Idx extra_nmatch; + bool sb; + int ch; +#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) + re_match_context_t mctx = { .dfa = dfa }; +#else + re_match_context_t mctx; +#endif + char *fastmap = ((preg->fastmap != NULL && preg->fastmap_accurate + && start != last_start && !preg->can_be_null) + ? preg->fastmap : NULL); + RE_TRANSLATE_TYPE t = preg->translate; + +#if !(defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)) + memset (&mctx, '\0', sizeof (re_match_context_t)); + mctx.dfa = dfa; +#endif + + extra_nmatch = (nmatch > preg->re_nsub) ? nmatch - (preg->re_nsub + 1) : 0; + nmatch -= extra_nmatch; + + /* Check if the DFA haven't been compiled. */ + if (__glibc_unlikely (preg->used == 0 || dfa->init_state == NULL + || dfa->init_state_word == NULL + || dfa->init_state_nl == NULL + || dfa->init_state_begbuf == NULL)) + return REG_NOMATCH; + +#ifdef DEBUG + /* We assume front-end functions already check them. */ + assert (0 <= last_start && last_start <= length); +#endif + + /* If initial states with non-begbuf contexts have no elements, + the regex must be anchored. If preg->newline_anchor is set, + we'll never use init_state_nl, so do not check it. */ + if (dfa->init_state->nodes.nelem == 0 + && dfa->init_state_word->nodes.nelem == 0 + && (dfa->init_state_nl->nodes.nelem == 0 + || !preg->newline_anchor)) + { + if (start != 0 && last_start != 0) + return REG_NOMATCH; + start = last_start = 0; + } + + /* We must check the longest matching, if nmatch > 0. */ + fl_longest_match = (nmatch != 0 || dfa->nbackref); + + err = re_string_allocate (&mctx.input, string, length, dfa->nodes_len + 1, + preg->translate, (preg->syntax & RE_ICASE) != 0, + dfa); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + mctx.input.stop = stop; + mctx.input.raw_stop = stop; + mctx.input.newline_anchor = preg->newline_anchor; + + err = match_ctx_init (&mctx, eflags, dfa->nbackref * 2); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + + /* We will log all the DFA states through which the dfa pass, + if nmatch > 1, or this dfa has "multibyte node", which is a + back-reference or a node which can accept multibyte character or + multi character collating element. */ + if (nmatch > 1 || dfa->has_mb_node) + { + /* Avoid overflow. */ + if (__glibc_unlikely ((MIN (IDX_MAX, SIZE_MAX / sizeof (re_dfastate_t *)) + <= mctx.input.bufs_len))) + { + err = REG_ESPACE; + goto free_return; + } + + mctx.state_log = re_malloc (re_dfastate_t *, mctx.input.bufs_len + 1); + if (__glibc_unlikely (mctx.state_log == NULL)) + { + err = REG_ESPACE; + goto free_return; + } + } + else + mctx.state_log = NULL; + + match_first = start; + mctx.input.tip_context = (eflags & REG_NOTBOL) ? CONTEXT_BEGBUF + : CONTEXT_NEWLINE | CONTEXT_BEGBUF; + + /* Check incrementally whether the input string matches. */ + incr = (last_start < start) ? -1 : 1; + left_lim = (last_start < start) ? last_start : start; + right_lim = (last_start < start) ? start : last_start; + sb = dfa->mb_cur_max == 1; + match_kind = + (fastmap + ? ((sb || !(preg->syntax & RE_ICASE || t) ? 4 : 0) + | (start <= last_start ? 2 : 0) + | (t != NULL ? 1 : 0)) + : 8); + + for (;; match_first += incr) + { + err = REG_NOMATCH; + if (match_first < left_lim || right_lim < match_first) + goto free_return; + + /* Advance as rapidly as possible through the string, until we + find a plausible place to start matching. This may be done + with varying efficiency, so there are various possibilities: + only the most common of them are specialized, in order to + save on code size. We use a switch statement for speed. */ + switch (match_kind) + { + case 8: + /* No fastmap. */ + break; + + case 7: + /* Fastmap with single-byte translation, match forward. */ + while (__glibc_likely (match_first < right_lim) + && !fastmap[t[(unsigned char) string[match_first]]]) + ++match_first; + goto forward_match_found_start_or_reached_end; + + case 6: + /* Fastmap without translation, match forward. */ + while (__glibc_likely (match_first < right_lim) + && !fastmap[(unsigned char) string[match_first]]) + ++match_first; + + forward_match_found_start_or_reached_end: + if (__glibc_unlikely (match_first == right_lim)) + { + ch = match_first >= length + ? 0 : (unsigned char) string[match_first]; + if (!fastmap[t ? t[ch] : ch]) + goto free_return; + } + break; + + case 4: + case 5: + /* Fastmap without multi-byte translation, match backwards. */ + while (match_first >= left_lim) + { + ch = match_first >= length + ? 0 : (unsigned char) string[match_first]; + if (fastmap[t ? t[ch] : ch]) + break; + --match_first; + } + if (match_first < left_lim) + goto free_return; + break; + + default: + /* In this case, we can't determine easily the current byte, + since it might be a component byte of a multibyte + character. Then we use the constructed buffer instead. */ + for (;;) + { + /* If MATCH_FIRST is out of the valid range, reconstruct the + buffers. */ + __re_size_t offset = match_first - mctx.input.raw_mbs_idx; + if (__glibc_unlikely (offset + >= (__re_size_t) mctx.input.valid_raw_len)) + { + err = re_string_reconstruct (&mctx.input, match_first, + eflags); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + + offset = match_first - mctx.input.raw_mbs_idx; + } + /* If MATCH_FIRST is out of the buffer, leave it as '\0'. + Note that MATCH_FIRST must not be smaller than 0. */ + ch = (match_first >= length + ? 0 : re_string_byte_at (&mctx.input, offset)); + if (fastmap[ch]) + break; + match_first += incr; + if (match_first < left_lim || match_first > right_lim) + { + err = REG_NOMATCH; + goto free_return; + } + } + break; + } + + /* Reconstruct the buffers so that the matcher can assume that + the matching starts from the beginning of the buffer. */ + err = re_string_reconstruct (&mctx.input, match_first, eflags); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + +#ifdef RE_ENABLE_I18N + /* Don't consider this char as a possible match start if it part, + yet isn't the head, of a multibyte character. */ + if (!sb && !re_string_first_byte (&mctx.input, 0)) + continue; +#endif + + /* It seems to be appropriate one, then use the matcher. */ + /* We assume that the matching starts from 0. */ + mctx.state_log_top = mctx.nbkref_ents = mctx.max_mb_elem_len = 0; + match_last = check_matching (&mctx, fl_longest_match, + start <= last_start ? &match_first : NULL); + if (match_last != -1) + { + if (__glibc_unlikely (match_last == -2)) + { + err = REG_ESPACE; + goto free_return; + } + else + { + mctx.match_last = match_last; + if ((!preg->no_sub && nmatch > 1) || dfa->nbackref) + { + re_dfastate_t *pstate = mctx.state_log[match_last]; + mctx.last_node = check_halt_state_context (&mctx, pstate, + match_last); + } + if ((!preg->no_sub && nmatch > 1 && dfa->has_plural_match) + || dfa->nbackref) + { + err = prune_impossible_nodes (&mctx); + if (err == REG_NOERROR) + break; + if (__glibc_unlikely (err != REG_NOMATCH)) + goto free_return; + match_last = -1; + } + else + break; /* We found a match. */ + } + } + + match_ctx_clean (&mctx); + } + +#ifdef DEBUG + assert (match_last != -1); + assert (err == REG_NOERROR); +#endif + + /* Set pmatch[] if we need. */ + if (nmatch > 0) + { + Idx reg_idx; + + /* Initialize registers. */ + for (reg_idx = 1; reg_idx < nmatch; ++reg_idx) + pmatch[reg_idx].rm_so = pmatch[reg_idx].rm_eo = -1; + + /* Set the points where matching start/end. */ + pmatch[0].rm_so = 0; + pmatch[0].rm_eo = mctx.match_last; + /* FIXME: This function should fail if mctx.match_last exceeds + the maximum possible regoff_t value. We need a new error + code REG_OVERFLOW. */ + + if (!preg->no_sub && nmatch > 1) + { + err = set_regs (preg, &mctx, nmatch, pmatch, + dfa->has_plural_match && dfa->nbackref > 0); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + + /* At last, add the offset to each register, since we slid + the buffers so that we could assume that the matching starts + from 0. */ + for (reg_idx = 0; reg_idx < nmatch; ++reg_idx) + if (pmatch[reg_idx].rm_so != -1) + { +#ifdef RE_ENABLE_I18N + if (__glibc_unlikely (mctx.input.offsets_needed != 0)) + { + pmatch[reg_idx].rm_so = + (pmatch[reg_idx].rm_so == mctx.input.valid_len + ? mctx.input.valid_raw_len + : mctx.input.offsets[pmatch[reg_idx].rm_so]); + pmatch[reg_idx].rm_eo = + (pmatch[reg_idx].rm_eo == mctx.input.valid_len + ? mctx.input.valid_raw_len + : mctx.input.offsets[pmatch[reg_idx].rm_eo]); + } +#else + assert (mctx.input.offsets_needed == 0); +#endif + pmatch[reg_idx].rm_so += match_first; + pmatch[reg_idx].rm_eo += match_first; + } + for (reg_idx = 0; reg_idx < extra_nmatch; ++reg_idx) + { + pmatch[nmatch + reg_idx].rm_so = -1; + pmatch[nmatch + reg_idx].rm_eo = -1; + } + + if (dfa->subexp_map) + for (reg_idx = 0; reg_idx + 1 < nmatch; reg_idx++) + if (dfa->subexp_map[reg_idx] != reg_idx) + { + pmatch[reg_idx + 1].rm_so + = pmatch[dfa->subexp_map[reg_idx] + 1].rm_so; + pmatch[reg_idx + 1].rm_eo + = pmatch[dfa->subexp_map[reg_idx] + 1].rm_eo; + } + } + + free_return: + re_free (mctx.state_log); + if (dfa->nbackref) + match_ctx_free (&mctx); + re_string_destruct (&mctx.input); + return err; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +prune_impossible_nodes (re_match_context_t *mctx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx halt_node, match_last; + reg_errcode_t ret; + re_dfastate_t **sifted_states; + re_dfastate_t **lim_states = NULL; + re_sift_context_t sctx; +#ifdef DEBUG + assert (mctx->state_log != NULL); +#endif + match_last = mctx->match_last; + halt_node = mctx->last_node; + + /* Avoid overflow. */ + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / sizeof (re_dfastate_t *)) + <= match_last)) + return REG_ESPACE; + + sifted_states = re_malloc (re_dfastate_t *, match_last + 1); + if (__glibc_unlikely (sifted_states == NULL)) + { + ret = REG_ESPACE; + goto free_return; + } + if (dfa->nbackref) + { + lim_states = re_malloc (re_dfastate_t *, match_last + 1); + if (__glibc_unlikely (lim_states == NULL)) + { + ret = REG_ESPACE; + goto free_return; + } + while (1) + { + memset (lim_states, '\0', + sizeof (re_dfastate_t *) * (match_last + 1)); + sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, + match_last); + ret = sift_states_backward (mctx, &sctx); + re_node_set_free (&sctx.limits); + if (__glibc_unlikely (ret != REG_NOERROR)) + goto free_return; + if (sifted_states[0] != NULL || lim_states[0] != NULL) + break; + do + { + --match_last; + if (match_last < 0) + { + ret = REG_NOMATCH; + goto free_return; + } + } while (mctx->state_log[match_last] == NULL + || !mctx->state_log[match_last]->halt); + halt_node = check_halt_state_context (mctx, + mctx->state_log[match_last], + match_last); + } + ret = merge_state_array (dfa, sifted_states, lim_states, + match_last + 1); + re_free (lim_states); + lim_states = NULL; + if (__glibc_unlikely (ret != REG_NOERROR)) + goto free_return; + } + else + { + sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, match_last); + ret = sift_states_backward (mctx, &sctx); + re_node_set_free (&sctx.limits); + if (__glibc_unlikely (ret != REG_NOERROR)) + goto free_return; + if (sifted_states[0] == NULL) + { + ret = REG_NOMATCH; + goto free_return; + } + } + re_free (mctx->state_log); + mctx->state_log = sifted_states; + sifted_states = NULL; + mctx->last_node = halt_node; + mctx->match_last = match_last; + ret = REG_NOERROR; + free_return: + re_free (sifted_states); + re_free (lim_states); + return ret; +} + +/* Acquire an initial state and return it. + We must select appropriate initial state depending on the context, + since initial states may have constraints like "\<", "^", etc.. */ + +static inline re_dfastate_t * +__attribute__ ((always_inline)) +acquire_init_state_context (reg_errcode_t *err, const re_match_context_t *mctx, + Idx idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + if (dfa->init_state->has_constraint) + { + unsigned int context; + context = re_string_context_at (&mctx->input, idx - 1, mctx->eflags); + if (IS_WORD_CONTEXT (context)) + return dfa->init_state_word; + else if (IS_ORDINARY_CONTEXT (context)) + return dfa->init_state; + else if (IS_BEGBUF_CONTEXT (context) && IS_NEWLINE_CONTEXT (context)) + return dfa->init_state_begbuf; + else if (IS_NEWLINE_CONTEXT (context)) + return dfa->init_state_nl; + else if (IS_BEGBUF_CONTEXT (context)) + { + /* It is relatively rare case, then calculate on demand. */ + return re_acquire_state_context (err, dfa, + dfa->init_state->entrance_nodes, + context); + } + else + /* Must not happen? */ + return dfa->init_state; + } + else + return dfa->init_state; +} + +/* Check whether the regular expression match input string INPUT or not, + and return the index where the matching end. Return -1 if + there is no match, and return -2 in case of an error. + FL_LONGEST_MATCH means we want the POSIX longest matching. + If P_MATCH_FIRST is not NULL, and the match fails, it is set to the + next place where we may want to try matching. + Note that the matcher assumes that the matching starts from the current + index of the buffer. */ + +static Idx +__attribute_warn_unused_result__ +check_matching (re_match_context_t *mctx, bool fl_longest_match, + Idx *p_match_first) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx match = 0; + Idx match_last = -1; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + re_dfastate_t *cur_state; + bool at_init_state = p_match_first != NULL; + Idx next_start_idx = cur_str_idx; + + err = REG_NOERROR; + cur_state = acquire_init_state_context (&err, mctx, cur_str_idx); + /* An initial state must not be NULL (invalid). */ + if (__glibc_unlikely (cur_state == NULL)) + { + assert (err == REG_ESPACE); + return -2; + } + + if (mctx->state_log != NULL) + { + mctx->state_log[cur_str_idx] = cur_state; + + /* Check OP_OPEN_SUBEXP in the initial state in case that we use them + later. E.g. Processing back references. */ + if (__glibc_unlikely (dfa->nbackref)) + { + at_init_state = false; + err = check_subexp_matching_top (mctx, &cur_state->nodes, 0); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + if (cur_state->has_backref) + { + err = transit_state_bkref (mctx, &cur_state->nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + } + + /* If the RE accepts NULL string. */ + if (__glibc_unlikely (cur_state->halt)) + { + if (!cur_state->has_constraint + || check_halt_state_context (mctx, cur_state, cur_str_idx)) + { + if (!fl_longest_match) + return cur_str_idx; + else + { + match_last = cur_str_idx; + match = 1; + } + } + } + + while (!re_string_eoi (&mctx->input)) + { + re_dfastate_t *old_state = cur_state; + Idx next_char_idx = re_string_cur_idx (&mctx->input) + 1; + + if ((__glibc_unlikely (next_char_idx >= mctx->input.bufs_len) + && mctx->input.bufs_len < mctx->input.len) + || (__glibc_unlikely (next_char_idx >= mctx->input.valid_len) + && mctx->input.valid_len < mctx->input.len)) + { + err = extend_buffers (mctx, next_char_idx + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + { + assert (err == REG_ESPACE); + return -2; + } + } + + cur_state = transit_state (&err, mctx, cur_state); + if (mctx->state_log != NULL) + cur_state = merge_state_with_log (&err, mctx, cur_state); + + if (cur_state == NULL) + { + /* Reached the invalid state or an error. Try to recover a valid + state using the state log, if available and if we have not + already found a valid (even if not the longest) match. */ + if (__glibc_unlikely (err != REG_NOERROR)) + return -2; + + if (mctx->state_log == NULL + || (match && !fl_longest_match) + || (cur_state = find_recover_state (&err, mctx)) == NULL) + break; + } + + if (__glibc_unlikely (at_init_state)) + { + if (old_state == cur_state) + next_start_idx = next_char_idx; + else + at_init_state = false; + } + + if (cur_state->halt) + { + /* Reached a halt state. + Check the halt state can satisfy the current context. */ + if (!cur_state->has_constraint + || check_halt_state_context (mctx, cur_state, + re_string_cur_idx (&mctx->input))) + { + /* We found an appropriate halt state. */ + match_last = re_string_cur_idx (&mctx->input); + match = 1; + + /* We found a match, do not modify match_first below. */ + p_match_first = NULL; + if (!fl_longest_match) + break; + } + } + } + + if (p_match_first) + *p_match_first += next_start_idx; + + return match_last; +} + +/* Check NODE match the current context. */ + +static bool +check_halt_node_context (const re_dfa_t *dfa, Idx node, unsigned int context) +{ + re_token_type_t type = dfa->nodes[node].type; + unsigned int constraint = dfa->nodes[node].constraint; + if (type != END_OF_RE) + return false; + if (!constraint) + return true; + if (NOT_SATISFY_NEXT_CONSTRAINT (constraint, context)) + return false; + return true; +} + +/* Check the halt state STATE match the current context. + Return 0 if not match, if the node, STATE has, is a halt node and + match the context, return the node. */ + +static Idx +check_halt_state_context (const re_match_context_t *mctx, + const re_dfastate_t *state, Idx idx) +{ + Idx i; + unsigned int context; +#ifdef DEBUG + assert (state->halt); +#endif + context = re_string_context_at (&mctx->input, idx, mctx->eflags); + for (i = 0; i < state->nodes.nelem; ++i) + if (check_halt_node_context (mctx->dfa, state->nodes.elems[i], context)) + return state->nodes.elems[i]; + return 0; +} + +/* Compute the next node to which "NFA" transit from NODE("NFA" is a NFA + corresponding to the DFA). + Return the destination node, and update EPS_VIA_NODES; + return -1 in case of errors. */ + +static Idx +proceed_next_node (const re_match_context_t *mctx, Idx nregs, regmatch_t *regs, + Idx *pidx, Idx node, re_node_set *eps_via_nodes, + struct re_fail_stack_t *fs) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx i; + bool ok; + if (IS_EPSILON_NODE (dfa->nodes[node].type)) + { + re_node_set *cur_nodes = &mctx->state_log[*pidx]->nodes; + re_node_set *edests = &dfa->edests[node]; + Idx dest_node; + ok = re_node_set_insert (eps_via_nodes, node); + if (__glibc_unlikely (! ok)) + return -2; + /* Pick up a valid destination, or return -1 if none + is found. */ + for (dest_node = -1, i = 0; i < edests->nelem; ++i) + { + Idx candidate = edests->elems[i]; + if (!re_node_set_contains (cur_nodes, candidate)) + continue; + if (dest_node == -1) + dest_node = candidate; + + else + { + /* In order to avoid infinite loop like "(a*)*", return the second + epsilon-transition if the first was already considered. */ + if (re_node_set_contains (eps_via_nodes, dest_node)) + return candidate; + + /* Otherwise, push the second epsilon-transition on the fail stack. */ + else if (fs != NULL + && push_fail_stack (fs, *pidx, candidate, nregs, regs, + eps_via_nodes)) + return -2; + + /* We know we are going to exit. */ + break; + } + } + return dest_node; + } + else + { + Idx naccepted = 0; + re_token_type_t type = dfa->nodes[node].type; + +#ifdef RE_ENABLE_I18N + if (dfa->nodes[node].accept_mb) + naccepted = check_node_accept_bytes (dfa, node, &mctx->input, *pidx); + else +#endif /* RE_ENABLE_I18N */ + if (type == OP_BACK_REF) + { + Idx subexp_idx = dfa->nodes[node].opr.idx + 1; + naccepted = regs[subexp_idx].rm_eo - regs[subexp_idx].rm_so; + if (fs != NULL) + { + if (regs[subexp_idx].rm_so == -1 || regs[subexp_idx].rm_eo == -1) + return -1; + else if (naccepted) + { + char *buf = (char *) re_string_get_buffer (&mctx->input); + if (mctx->input.valid_len - *pidx < naccepted + || (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx, + naccepted) + != 0)) + return -1; + } + } + + if (naccepted == 0) + { + Idx dest_node; + ok = re_node_set_insert (eps_via_nodes, node); + if (__glibc_unlikely (! ok)) + return -2; + dest_node = dfa->edests[node].elems[0]; + if (re_node_set_contains (&mctx->state_log[*pidx]->nodes, + dest_node)) + return dest_node; + } + } + + if (naccepted != 0 + || check_node_accept (mctx, dfa->nodes + node, *pidx)) + { + Idx dest_node = dfa->nexts[node]; + *pidx = (naccepted == 0) ? *pidx + 1 : *pidx + naccepted; + if (fs && (*pidx > mctx->match_last || mctx->state_log[*pidx] == NULL + || !re_node_set_contains (&mctx->state_log[*pidx]->nodes, + dest_node))) + return -1; + re_node_set_empty (eps_via_nodes); + return dest_node; + } + } + return -1; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +push_fail_stack (struct re_fail_stack_t *fs, Idx str_idx, Idx dest_node, + Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes) +{ + reg_errcode_t err; + Idx num = fs->num++; + if (fs->num == fs->alloc) + { + struct re_fail_stack_ent_t *new_array; + new_array = re_realloc (fs->stack, struct re_fail_stack_ent_t, + fs->alloc * 2); + if (new_array == NULL) + return REG_ESPACE; + fs->alloc *= 2; + fs->stack = new_array; + } + fs->stack[num].idx = str_idx; + fs->stack[num].node = dest_node; + fs->stack[num].regs = re_malloc (regmatch_t, nregs); + if (fs->stack[num].regs == NULL) + return REG_ESPACE; + memcpy (fs->stack[num].regs, regs, sizeof (regmatch_t) * nregs); + err = re_node_set_init_copy (&fs->stack[num].eps_via_nodes, eps_via_nodes); + return err; +} + +static Idx +pop_fail_stack (struct re_fail_stack_t *fs, Idx *pidx, Idx nregs, + regmatch_t *regs, re_node_set *eps_via_nodes) +{ + Idx num = --fs->num; + assert (num >= 0); + *pidx = fs->stack[num].idx; + memcpy (regs, fs->stack[num].regs, sizeof (regmatch_t) * nregs); + re_node_set_free (eps_via_nodes); + re_free (fs->stack[num].regs); + *eps_via_nodes = fs->stack[num].eps_via_nodes; + return fs->stack[num].node; +} + +/* Set the positions where the subexpressions are starts/ends to registers + PMATCH. + Note: We assume that pmatch[0] is already set, and + pmatch[i].rm_so == pmatch[i].rm_eo == -1 for 0 < i < nmatch. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +set_regs (const regex_t *preg, const re_match_context_t *mctx, size_t nmatch, + regmatch_t *pmatch, bool fl_backtrack) +{ + const re_dfa_t *dfa = preg->buffer; + Idx idx, cur_node; + re_node_set eps_via_nodes; + struct re_fail_stack_t *fs; + struct re_fail_stack_t fs_body = { 0, 2, NULL }; + regmatch_t *prev_idx_match; + bool prev_idx_match_malloced = false; + +#ifdef DEBUG + assert (nmatch > 1); + assert (mctx->state_log != NULL); +#endif + if (fl_backtrack) + { + fs = &fs_body; + fs->stack = re_malloc (struct re_fail_stack_ent_t, fs->alloc); + if (fs->stack == NULL) + return REG_ESPACE; + } + else + fs = NULL; + + cur_node = dfa->init_node; + re_node_set_init_empty (&eps_via_nodes); + + if (__libc_use_alloca (nmatch * sizeof (regmatch_t))) + prev_idx_match = (regmatch_t *) alloca (nmatch * sizeof (regmatch_t)); + else + { + prev_idx_match = re_malloc (regmatch_t, nmatch); + if (prev_idx_match == NULL) + { + free_fail_stack_return (fs); + return REG_ESPACE; + } + prev_idx_match_malloced = true; + } + memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch); + + for (idx = pmatch[0].rm_so; idx <= pmatch[0].rm_eo ;) + { + update_regs (dfa, pmatch, prev_idx_match, cur_node, idx, nmatch); + + if (idx == pmatch[0].rm_eo && cur_node == mctx->last_node) + { + Idx reg_idx; + if (fs) + { + for (reg_idx = 0; reg_idx < nmatch; ++reg_idx) + if (pmatch[reg_idx].rm_so > -1 && pmatch[reg_idx].rm_eo == -1) + break; + if (reg_idx == nmatch) + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return free_fail_stack_return (fs); + } + cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch, + &eps_via_nodes); + } + else + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return REG_NOERROR; + } + } + + /* Proceed to next node. */ + cur_node = proceed_next_node (mctx, nmatch, pmatch, &idx, cur_node, + &eps_via_nodes, fs); + + if (__glibc_unlikely (cur_node < 0)) + { + if (__glibc_unlikely (cur_node == -2)) + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + free_fail_stack_return (fs); + return REG_ESPACE; + } + if (fs) + cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch, + &eps_via_nodes); + else + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return REG_NOMATCH; + } + } + } + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return free_fail_stack_return (fs); +} + +static reg_errcode_t +free_fail_stack_return (struct re_fail_stack_t *fs) +{ + if (fs) + { + Idx fs_idx; + for (fs_idx = 0; fs_idx < fs->num; ++fs_idx) + { + re_node_set_free (&fs->stack[fs_idx].eps_via_nodes); + re_free (fs->stack[fs_idx].regs); + } + re_free (fs->stack); + } + return REG_NOERROR; +} + +static void +update_regs (const re_dfa_t *dfa, regmatch_t *pmatch, + regmatch_t *prev_idx_match, Idx cur_node, Idx cur_idx, Idx nmatch) +{ + int type = dfa->nodes[cur_node].type; + if (type == OP_OPEN_SUBEXP) + { + Idx reg_num = dfa->nodes[cur_node].opr.idx + 1; + + /* We are at the first node of this sub expression. */ + if (reg_num < nmatch) + { + pmatch[reg_num].rm_so = cur_idx; + pmatch[reg_num].rm_eo = -1; + } + } + else if (type == OP_CLOSE_SUBEXP) + { + Idx reg_num = dfa->nodes[cur_node].opr.idx + 1; + if (reg_num < nmatch) + { + /* We are at the last node of this sub expression. */ + if (pmatch[reg_num].rm_so < cur_idx) + { + pmatch[reg_num].rm_eo = cur_idx; + /* This is a non-empty match or we are not inside an optional + subexpression. Accept this right away. */ + memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch); + } + else + { + if (dfa->nodes[cur_node].opt_subexp + && prev_idx_match[reg_num].rm_so != -1) + /* We transited through an empty match for an optional + subexpression, like (a?)*, and this is not the subexp's + first match. Copy back the old content of the registers + so that matches of an inner subexpression are undone as + well, like in ((a?))*. */ + memcpy (pmatch, prev_idx_match, sizeof (regmatch_t) * nmatch); + else + /* We completed a subexpression, but it may be part of + an optional one, so do not update PREV_IDX_MATCH. */ + pmatch[reg_num].rm_eo = cur_idx; + } + } + } +} + +/* This function checks the STATE_LOG from the SCTX->last_str_idx to 0 + and sift the nodes in each states according to the following rules. + Updated state_log will be wrote to STATE_LOG. + + Rules: We throw away the Node 'a' in the STATE_LOG[STR_IDX] if... + 1. When STR_IDX == MATCH_LAST(the last index in the state_log): + If 'a' isn't the LAST_NODE and 'a' can't epsilon transit to + the LAST_NODE, we throw away the node 'a'. + 2. When 0 <= STR_IDX < MATCH_LAST and 'a' accepts + string 's' and transit to 'b': + i. If 'b' isn't in the STATE_LOG[STR_IDX+strlen('s')], we throw + away the node 'a'. + ii. If 'b' is in the STATE_LOG[STR_IDX+strlen('s')] but 'b' is + thrown away, we throw away the node 'a'. + 3. When 0 <= STR_IDX < MATCH_LAST and 'a' epsilon transit to 'b': + i. If 'b' isn't in the STATE_LOG[STR_IDX], we throw away the + node 'a'. + ii. If 'b' is in the STATE_LOG[STR_IDX] but 'b' is thrown away, + we throw away the node 'a'. */ + +#define STATE_NODE_CONTAINS(state,node) \ + ((state) != NULL && re_node_set_contains (&(state)->nodes, node)) + +static reg_errcode_t +sift_states_backward (const re_match_context_t *mctx, re_sift_context_t *sctx) +{ + reg_errcode_t err; + int null_cnt = 0; + Idx str_idx = sctx->last_str_idx; + re_node_set cur_dest; + +#ifdef DEBUG + assert (mctx->state_log != NULL && mctx->state_log[str_idx] != NULL); +#endif + + /* Build sifted state_log[str_idx]. It has the nodes which can epsilon + transit to the last_node and the last_node itself. */ + err = re_node_set_init_1 (&cur_dest, sctx->last_node); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + + /* Then check each states in the state_log. */ + while (str_idx > 0) + { + /* Update counters. */ + null_cnt = (sctx->sifted_states[str_idx] == NULL) ? null_cnt + 1 : 0; + if (null_cnt > mctx->max_mb_elem_len) + { + memset (sctx->sifted_states, '\0', + sizeof (re_dfastate_t *) * str_idx); + re_node_set_free (&cur_dest); + return REG_NOERROR; + } + re_node_set_empty (&cur_dest); + --str_idx; + + if (mctx->state_log[str_idx]) + { + err = build_sifted_states (mctx, sctx, str_idx, &cur_dest); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + + /* Add all the nodes which satisfy the following conditions: + - It can epsilon transit to a node in CUR_DEST. + - It is in CUR_SRC. + And update state_log. */ + err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + err = REG_NOERROR; + free_return: + re_node_set_free (&cur_dest); + return err; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +build_sifted_states (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx str_idx, re_node_set *cur_dest) +{ + const re_dfa_t *const dfa = mctx->dfa; + const re_node_set *cur_src = &mctx->state_log[str_idx]->non_eps_nodes; + Idx i; + + /* Then build the next sifted state. + We build the next sifted state on 'cur_dest', and update + 'sifted_states[str_idx]' with 'cur_dest'. + Note: + 'cur_dest' is the sifted state from 'state_log[str_idx + 1]'. + 'cur_src' points the node_set of the old 'state_log[str_idx]' + (with the epsilon nodes pre-filtered out). */ + for (i = 0; i < cur_src->nelem; i++) + { + Idx prev_node = cur_src->elems[i]; + int naccepted = 0; + bool ok; + +#ifdef DEBUG + re_token_type_t type = dfa->nodes[prev_node].type; + assert (!IS_EPSILON_NODE (type)); +#endif +#ifdef RE_ENABLE_I18N + /* If the node may accept "multi byte". */ + if (dfa->nodes[prev_node].accept_mb) + naccepted = sift_states_iter_mb (mctx, sctx, prev_node, + str_idx, sctx->last_str_idx); +#endif /* RE_ENABLE_I18N */ + + /* We don't check backreferences here. + See update_cur_sifted_state(). */ + if (!naccepted + && check_node_accept (mctx, dfa->nodes + prev_node, str_idx) + && STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + 1], + dfa->nexts[prev_node])) + naccepted = 1; + + if (naccepted == 0) + continue; + + if (sctx->limits.nelem) + { + Idx to_idx = str_idx + naccepted; + if (check_dst_limits (mctx, &sctx->limits, + dfa->nexts[prev_node], to_idx, + prev_node, str_idx)) + continue; + } + ok = re_node_set_insert (cur_dest, prev_node); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + + return REG_NOERROR; +} + +/* Helper functions. */ + +static reg_errcode_t +clean_state_log_if_needed (re_match_context_t *mctx, Idx next_state_log_idx) +{ + Idx top = mctx->state_log_top; + + if ((next_state_log_idx >= mctx->input.bufs_len + && mctx->input.bufs_len < mctx->input.len) + || (next_state_log_idx >= mctx->input.valid_len + && mctx->input.valid_len < mctx->input.len)) + { + reg_errcode_t err; + err = extend_buffers (mctx, next_state_log_idx + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + + if (top < next_state_log_idx) + { + memset (mctx->state_log + top + 1, '\0', + sizeof (re_dfastate_t *) * (next_state_log_idx - top)); + mctx->state_log_top = next_state_log_idx; + } + return REG_NOERROR; +} + +static reg_errcode_t +merge_state_array (const re_dfa_t *dfa, re_dfastate_t **dst, + re_dfastate_t **src, Idx num) +{ + Idx st_idx; + reg_errcode_t err; + for (st_idx = 0; st_idx < num; ++st_idx) + { + if (dst[st_idx] == NULL) + dst[st_idx] = src[st_idx]; + else if (src[st_idx] != NULL) + { + re_node_set merged_set; + err = re_node_set_init_union (&merged_set, &dst[st_idx]->nodes, + &src[st_idx]->nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + dst[st_idx] = re_acquire_state (&err, dfa, &merged_set); + re_node_set_free (&merged_set); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + return REG_NOERROR; +} + +static reg_errcode_t +update_cur_sifted_state (const re_match_context_t *mctx, + re_sift_context_t *sctx, Idx str_idx, + re_node_set *dest_nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err = REG_NOERROR; + const re_node_set *candidates; + candidates = ((mctx->state_log[str_idx] == NULL) ? NULL + : &mctx->state_log[str_idx]->nodes); + + if (dest_nodes->nelem == 0) + sctx->sifted_states[str_idx] = NULL; + else + { + if (candidates) + { + /* At first, add the nodes which can epsilon transit to a node in + DEST_NODE. */ + err = add_epsilon_src_nodes (dfa, dest_nodes, candidates); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + /* Then, check the limitations in the current sift_context. */ + if (sctx->limits.nelem) + { + err = check_subexp_limits (dfa, dest_nodes, candidates, &sctx->limits, + mctx->bkref_ents, str_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + + sctx->sifted_states[str_idx] = re_acquire_state (&err, dfa, dest_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + + if (candidates && mctx->state_log[str_idx]->has_backref) + { + err = sift_states_bkref (mctx, sctx, str_idx, candidates); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + return REG_NOERROR; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +add_epsilon_src_nodes (const re_dfa_t *dfa, re_node_set *dest_nodes, + const re_node_set *candidates) +{ + reg_errcode_t err = REG_NOERROR; + Idx i; + + re_dfastate_t *state = re_acquire_state (&err, dfa, dest_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + if (!state->inveclosure.alloc) + { + err = re_node_set_alloc (&state->inveclosure, dest_nodes->nelem); + if (__glibc_unlikely (err != REG_NOERROR)) + return REG_ESPACE; + for (i = 0; i < dest_nodes->nelem; i++) + { + err = re_node_set_merge (&state->inveclosure, + dfa->inveclosures + dest_nodes->elems[i]); + if (__glibc_unlikely (err != REG_NOERROR)) + return REG_ESPACE; + } + } + return re_node_set_add_intersect (dest_nodes, candidates, + &state->inveclosure); +} + +static reg_errcode_t +sub_epsilon_src_nodes (const re_dfa_t *dfa, Idx node, re_node_set *dest_nodes, + const re_node_set *candidates) +{ + Idx ecl_idx; + reg_errcode_t err; + re_node_set *inv_eclosure = dfa->inveclosures + node; + re_node_set except_nodes; + re_node_set_init_empty (&except_nodes); + for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx) + { + Idx cur_node = inv_eclosure->elems[ecl_idx]; + if (cur_node == node) + continue; + if (IS_EPSILON_NODE (dfa->nodes[cur_node].type)) + { + Idx edst1 = dfa->edests[cur_node].elems[0]; + Idx edst2 = ((dfa->edests[cur_node].nelem > 1) + ? dfa->edests[cur_node].elems[1] : -1); + if ((!re_node_set_contains (inv_eclosure, edst1) + && re_node_set_contains (dest_nodes, edst1)) + || (edst2 > 0 + && !re_node_set_contains (inv_eclosure, edst2) + && re_node_set_contains (dest_nodes, edst2))) + { + err = re_node_set_add_intersect (&except_nodes, candidates, + dfa->inveclosures + cur_node); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&except_nodes); + return err; + } + } + } + } + for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx) + { + Idx cur_node = inv_eclosure->elems[ecl_idx]; + if (!re_node_set_contains (&except_nodes, cur_node)) + { + Idx idx = re_node_set_contains (dest_nodes, cur_node) - 1; + re_node_set_remove_at (dest_nodes, idx); + } + } + re_node_set_free (&except_nodes); + return REG_NOERROR; +} + +static bool +check_dst_limits (const re_match_context_t *mctx, const re_node_set *limits, + Idx dst_node, Idx dst_idx, Idx src_node, Idx src_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx lim_idx, src_pos, dst_pos; + + Idx dst_bkref_idx = search_cur_bkref_entry (mctx, dst_idx); + Idx src_bkref_idx = search_cur_bkref_entry (mctx, src_idx); + for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx) + { + Idx subexp_idx; + struct re_backref_cache_entry *ent; + ent = mctx->bkref_ents + limits->elems[lim_idx]; + subexp_idx = dfa->nodes[ent->node].opr.idx; + + dst_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx], + subexp_idx, dst_node, dst_idx, + dst_bkref_idx); + src_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx], + subexp_idx, src_node, src_idx, + src_bkref_idx); + + /* In case of: + <src> <dst> ( <subexp> ) + ( <subexp> ) <src> <dst> + ( <subexp1> <src> <subexp2> <dst> <subexp3> ) */ + if (src_pos == dst_pos) + continue; /* This is unrelated limitation. */ + else + return true; + } + return false; +} + +static int +check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, int boundaries, + Idx subexp_idx, Idx from_node, Idx bkref_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + const re_node_set *eclosures = dfa->eclosures + from_node; + Idx node_idx; + + /* Else, we are on the boundary: examine the nodes on the epsilon + closure. */ + for (node_idx = 0; node_idx < eclosures->nelem; ++node_idx) + { + Idx node = eclosures->elems[node_idx]; + switch (dfa->nodes[node].type) + { + case OP_BACK_REF: + if (bkref_idx != -1) + { + struct re_backref_cache_entry *ent = mctx->bkref_ents + bkref_idx; + do + { + Idx dst; + int cpos; + + if (ent->node != node) + continue; + + if (subexp_idx < BITSET_WORD_BITS + && !(ent->eps_reachable_subexps_map + & ((bitset_word_t) 1 << subexp_idx))) + continue; + + /* Recurse trying to reach the OP_OPEN_SUBEXP and + OP_CLOSE_SUBEXP cases below. But, if the + destination node is the same node as the source + node, don't recurse because it would cause an + infinite loop: a regex that exhibits this behavior + is ()\1*\1* */ + dst = dfa->edests[node].elems[0]; + if (dst == from_node) + { + if (boundaries & 1) + return -1; + else /* if (boundaries & 2) */ + return 0; + } + + cpos = + check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx, + dst, bkref_idx); + if (cpos == -1 /* && (boundaries & 1) */) + return -1; + if (cpos == 0 && (boundaries & 2)) + return 0; + + if (subexp_idx < BITSET_WORD_BITS) + ent->eps_reachable_subexps_map + &= ~((bitset_word_t) 1 << subexp_idx); + } + while (ent++->more); + } + break; + + case OP_OPEN_SUBEXP: + if ((boundaries & 1) && subexp_idx == dfa->nodes[node].opr.idx) + return -1; + break; + + case OP_CLOSE_SUBEXP: + if ((boundaries & 2) && subexp_idx == dfa->nodes[node].opr.idx) + return 0; + break; + + default: + break; + } + } + + return (boundaries & 2) ? 1 : 0; +} + +static int +check_dst_limits_calc_pos (const re_match_context_t *mctx, Idx limit, + Idx subexp_idx, Idx from_node, Idx str_idx, + Idx bkref_idx) +{ + struct re_backref_cache_entry *lim = mctx->bkref_ents + limit; + int boundaries; + + /* If we are outside the range of the subexpression, return -1 or 1. */ + if (str_idx < lim->subexp_from) + return -1; + + if (lim->subexp_to < str_idx) + return 1; + + /* If we are within the subexpression, return 0. */ + boundaries = (str_idx == lim->subexp_from); + boundaries |= (str_idx == lim->subexp_to) << 1; + if (boundaries == 0) + return 0; + + /* Else, examine epsilon closure. */ + return check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx, + from_node, bkref_idx); +} + +/* Check the limitations of sub expressions LIMITS, and remove the nodes + which are against limitations from DEST_NODES. */ + +static reg_errcode_t +check_subexp_limits (const re_dfa_t *dfa, re_node_set *dest_nodes, + const re_node_set *candidates, re_node_set *limits, + struct re_backref_cache_entry *bkref_ents, Idx str_idx) +{ + reg_errcode_t err; + Idx node_idx, lim_idx; + + for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx) + { + Idx subexp_idx; + struct re_backref_cache_entry *ent; + ent = bkref_ents + limits->elems[lim_idx]; + + if (str_idx <= ent->subexp_from || ent->str_idx < str_idx) + continue; /* This is unrelated limitation. */ + + subexp_idx = dfa->nodes[ent->node].opr.idx; + if (ent->subexp_to == str_idx) + { + Idx ops_node = -1; + Idx cls_node = -1; + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + re_token_type_t type = dfa->nodes[node].type; + if (type == OP_OPEN_SUBEXP + && subexp_idx == dfa->nodes[node].opr.idx) + ops_node = node; + else if (type == OP_CLOSE_SUBEXP + && subexp_idx == dfa->nodes[node].opr.idx) + cls_node = node; + } + + /* Check the limitation of the open subexpression. */ + /* Note that (ent->subexp_to = str_idx != ent->subexp_from). */ + if (ops_node >= 0) + { + err = sub_epsilon_src_nodes (dfa, ops_node, dest_nodes, + candidates); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + + /* Check the limitation of the close subexpression. */ + if (cls_node >= 0) + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + if (!re_node_set_contains (dfa->inveclosures + node, + cls_node) + && !re_node_set_contains (dfa->eclosures + node, + cls_node)) + { + /* It is against this limitation. + Remove it form the current sifted state. */ + err = sub_epsilon_src_nodes (dfa, node, dest_nodes, + candidates); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + --node_idx; + } + } + } + else /* (ent->subexp_to != str_idx) */ + { + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + re_token_type_t type = dfa->nodes[node].type; + if (type == OP_CLOSE_SUBEXP || type == OP_OPEN_SUBEXP) + { + if (subexp_idx != dfa->nodes[node].opr.idx) + continue; + /* It is against this limitation. + Remove it form the current sifted state. */ + err = sub_epsilon_src_nodes (dfa, node, dest_nodes, + candidates); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + } + } + return REG_NOERROR; +} + +static reg_errcode_t +__attribute_warn_unused_result__ +sift_states_bkref (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx str_idx, const re_node_set *candidates) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx node_idx, node; + re_sift_context_t local_sctx; + Idx first_idx = search_cur_bkref_entry (mctx, str_idx); + + if (first_idx == -1) + return REG_NOERROR; + + local_sctx.sifted_states = NULL; /* Mark that it hasn't been initialized. */ + + for (node_idx = 0; node_idx < candidates->nelem; ++node_idx) + { + Idx enabled_idx; + re_token_type_t type; + struct re_backref_cache_entry *entry; + node = candidates->elems[node_idx]; + type = dfa->nodes[node].type; + /* Avoid infinite loop for the REs like "()\1+". */ + if (node == sctx->last_node && str_idx == sctx->last_str_idx) + continue; + if (type != OP_BACK_REF) + continue; + + entry = mctx->bkref_ents + first_idx; + enabled_idx = first_idx; + do + { + Idx subexp_len; + Idx to_idx; + Idx dst_node; + bool ok; + re_dfastate_t *cur_state; + + if (entry->node != node) + continue; + subexp_len = entry->subexp_to - entry->subexp_from; + to_idx = str_idx + subexp_len; + dst_node = (subexp_len ? dfa->nexts[node] + : dfa->edests[node].elems[0]); + + if (to_idx > sctx->last_str_idx + || sctx->sifted_states[to_idx] == NULL + || !STATE_NODE_CONTAINS (sctx->sifted_states[to_idx], dst_node) + || check_dst_limits (mctx, &sctx->limits, node, + str_idx, dst_node, to_idx)) + continue; + + if (local_sctx.sifted_states == NULL) + { + local_sctx = *sctx; + err = re_node_set_init_copy (&local_sctx.limits, &sctx->limits); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + local_sctx.last_node = node; + local_sctx.last_str_idx = str_idx; + ok = re_node_set_insert (&local_sctx.limits, enabled_idx); + if (__glibc_unlikely (! ok)) + { + err = REG_ESPACE; + goto free_return; + } + cur_state = local_sctx.sifted_states[str_idx]; + err = sift_states_backward (mctx, &local_sctx); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + if (sctx->limited_states != NULL) + { + err = merge_state_array (dfa, sctx->limited_states, + local_sctx.sifted_states, + str_idx + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + local_sctx.sifted_states[str_idx] = cur_state; + re_node_set_remove (&local_sctx.limits, enabled_idx); + + /* mctx->bkref_ents may have changed, reload the pointer. */ + entry = mctx->bkref_ents + enabled_idx; + } + while (enabled_idx++, entry++->more); + } + err = REG_NOERROR; + free_return: + if (local_sctx.sifted_states != NULL) + { + re_node_set_free (&local_sctx.limits); + } + + return err; +} + + +#ifdef RE_ENABLE_I18N +static int +sift_states_iter_mb (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx node_idx, Idx str_idx, Idx max_str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + int naccepted; + /* Check the node can accept "multi byte". */ + naccepted = check_node_accept_bytes (dfa, node_idx, &mctx->input, str_idx); + if (naccepted > 0 && str_idx + naccepted <= max_str_idx + && !STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + naccepted], + dfa->nexts[node_idx])) + /* The node can't accept the "multi byte", or the + destination was already thrown away, then the node + couldn't accept the current input "multi byte". */ + naccepted = 0; + /* Otherwise, it is sure that the node could accept + 'naccepted' bytes input. */ + return naccepted; +} +#endif /* RE_ENABLE_I18N */ + + +/* Functions for state transition. */ + +/* Return the next state to which the current state STATE will transit by + accepting the current input byte, and update STATE_LOG if necessary. + If STATE can accept a multibyte char/collating element/back reference + update the destination of STATE_LOG. */ + +static re_dfastate_t * +__attribute_warn_unused_result__ +transit_state (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *state) +{ + re_dfastate_t **trtable; + unsigned char ch; + +#ifdef RE_ENABLE_I18N + /* If the current state can accept multibyte. */ + if (__glibc_unlikely (state->accept_mb)) + { + *err = transit_state_mb (mctx, state); + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + } +#endif /* RE_ENABLE_I18N */ + + /* Then decide the next state with the single byte. */ +#if 0 + if (0) + /* don't use transition table */ + return transit_state_sb (err, mctx, state); +#endif + + /* Use transition table */ + ch = re_string_fetch_byte (&mctx->input); + for (;;) + { + trtable = state->trtable; + if (__glibc_likely (trtable != NULL)) + return trtable[ch]; + + trtable = state->word_trtable; + if (__glibc_likely (trtable != NULL)) + { + unsigned int context; + context + = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input) - 1, + mctx->eflags); + if (IS_WORD_CONTEXT (context)) + return trtable[ch + SBC_MAX]; + else + return trtable[ch]; + } + + if (!build_trtable (mctx->dfa, state)) + { + *err = REG_ESPACE; + return NULL; + } + + /* Retry, we now have a transition table. */ + } +} + +/* Update the state_log if we need */ +static re_dfastate_t * +merge_state_with_log (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *next_state) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx cur_idx = re_string_cur_idx (&mctx->input); + + if (cur_idx > mctx->state_log_top) + { + mctx->state_log[cur_idx] = next_state; + mctx->state_log_top = cur_idx; + } + else if (mctx->state_log[cur_idx] == 0) + { + mctx->state_log[cur_idx] = next_state; + } + else + { + re_dfastate_t *pstate; + unsigned int context; + re_node_set next_nodes, *log_nodes, *table_nodes = NULL; + /* If (state_log[cur_idx] != 0), it implies that cur_idx is + the destination of a multibyte char/collating element/ + back reference. Then the next state is the union set of + these destinations and the results of the transition table. */ + pstate = mctx->state_log[cur_idx]; + log_nodes = pstate->entrance_nodes; + if (next_state != NULL) + { + table_nodes = next_state->entrance_nodes; + *err = re_node_set_init_union (&next_nodes, table_nodes, + log_nodes); + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + } + else + next_nodes = *log_nodes; + /* Note: We already add the nodes of the initial state, + then we don't need to add them here. */ + + context = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input) - 1, + mctx->eflags); + next_state = mctx->state_log[cur_idx] + = re_acquire_state_context (err, dfa, &next_nodes, context); + /* We don't need to check errors here, since the return value of + this function is next_state and ERR is already set. */ + + if (table_nodes != NULL) + re_node_set_free (&next_nodes); + } + + if (__glibc_unlikely (dfa->nbackref) && next_state != NULL) + { + /* Check OP_OPEN_SUBEXP in the current state in case that we use them + later. We must check them here, since the back references in the + next state might use them. */ + *err = check_subexp_matching_top (mctx, &next_state->nodes, + cur_idx); + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + + /* If the next state has back references. */ + if (next_state->has_backref) + { + *err = transit_state_bkref (mctx, &next_state->nodes); + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + next_state = mctx->state_log[cur_idx]; + } + } + + return next_state; +} + +/* Skip bytes in the input that correspond to part of a + multi-byte match, then look in the log for a state + from which to restart matching. */ +static re_dfastate_t * +find_recover_state (reg_errcode_t *err, re_match_context_t *mctx) +{ + re_dfastate_t *cur_state; + do + { + Idx max = mctx->state_log_top; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + + do + { + if (++cur_str_idx > max) + return NULL; + re_string_skip_bytes (&mctx->input, 1); + } + while (mctx->state_log[cur_str_idx] == NULL); + + cur_state = merge_state_with_log (err, mctx, NULL); + } + while (*err == REG_NOERROR && cur_state == NULL); + return cur_state; +} + +/* Helper functions for transit_state. */ + +/* From the node set CUR_NODES, pick up the nodes whose types are + OP_OPEN_SUBEXP and which have corresponding back references in the regular + expression. And register them to use them later for evaluating the + corresponding back references. */ + +static reg_errcode_t +check_subexp_matching_top (re_match_context_t *mctx, re_node_set *cur_nodes, + Idx str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx node_idx; + reg_errcode_t err; + + /* TODO: This isn't efficient. + Because there might be more than one nodes whose types are + OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all + nodes. + E.g. RE: (a){2} */ + for (node_idx = 0; node_idx < cur_nodes->nelem; ++node_idx) + { + Idx node = cur_nodes->elems[node_idx]; + if (dfa->nodes[node].type == OP_OPEN_SUBEXP + && dfa->nodes[node].opr.idx < BITSET_WORD_BITS + && (dfa->used_bkref_map + & ((bitset_word_t) 1 << dfa->nodes[node].opr.idx))) + { + err = match_ctx_add_subtop (mctx, node, str_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + return REG_NOERROR; +} + +#if 0 +/* Return the next state to which the current state STATE will transit by + accepting the current input byte. */ + +static re_dfastate_t * +transit_state_sb (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *state) +{ + const re_dfa_t *const dfa = mctx->dfa; + re_node_set next_nodes; + re_dfastate_t *next_state; + Idx node_cnt, cur_str_idx = re_string_cur_idx (&mctx->input); + unsigned int context; + + *err = re_node_set_alloc (&next_nodes, state->nodes.nelem + 1); + if (__glibc_unlikely (*err != REG_NOERROR)) + return NULL; + for (node_cnt = 0; node_cnt < state->nodes.nelem; ++node_cnt) + { + Idx cur_node = state->nodes.elems[node_cnt]; + if (check_node_accept (mctx, dfa->nodes + cur_node, cur_str_idx)) + { + *err = re_node_set_merge (&next_nodes, + dfa->eclosures + dfa->nexts[cur_node]); + if (__glibc_unlikely (*err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return NULL; + } + } + } + context = re_string_context_at (&mctx->input, cur_str_idx, mctx->eflags); + next_state = re_acquire_state_context (err, dfa, &next_nodes, context); + /* We don't need to check errors here, since the return value of + this function is next_state and ERR is already set. */ + + re_node_set_free (&next_nodes); + re_string_skip_bytes (&mctx->input, 1); + return next_state; +} +#endif + +#ifdef RE_ENABLE_I18N +static reg_errcode_t +transit_state_mb (re_match_context_t *mctx, re_dfastate_t *pstate) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx i; + + for (i = 0; i < pstate->nodes.nelem; ++i) + { + re_node_set dest_nodes, *new_nodes; + Idx cur_node_idx = pstate->nodes.elems[i]; + int naccepted; + Idx dest_idx; + unsigned int context; + re_dfastate_t *dest_state; + + if (!dfa->nodes[cur_node_idx].accept_mb) + continue; + + if (dfa->nodes[cur_node_idx].constraint) + { + context = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input), + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (dfa->nodes[cur_node_idx].constraint, + context)) + continue; + } + + /* How many bytes the node can accept? */ + naccepted = check_node_accept_bytes (dfa, cur_node_idx, &mctx->input, + re_string_cur_idx (&mctx->input)); + if (naccepted == 0) + continue; + + /* The node can accepts 'naccepted' bytes. */ + dest_idx = re_string_cur_idx (&mctx->input) + naccepted; + mctx->max_mb_elem_len = ((mctx->max_mb_elem_len < naccepted) ? naccepted + : mctx->max_mb_elem_len); + err = clean_state_log_if_needed (mctx, dest_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; +#ifdef DEBUG + assert (dfa->nexts[cur_node_idx] != -1); +#endif + new_nodes = dfa->eclosures + dfa->nexts[cur_node_idx]; + + dest_state = mctx->state_log[dest_idx]; + if (dest_state == NULL) + dest_nodes = *new_nodes; + else + { + err = re_node_set_init_union (&dest_nodes, + dest_state->entrance_nodes, new_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + context = re_string_context_at (&mctx->input, dest_idx - 1, + mctx->eflags); + mctx->state_log[dest_idx] + = re_acquire_state_context (&err, dfa, &dest_nodes, context); + if (dest_state != NULL) + re_node_set_free (&dest_nodes); + if (__glibc_unlikely (mctx->state_log[dest_idx] == NULL + && err != REG_NOERROR)) + return err; + } + return REG_NOERROR; +} +#endif /* RE_ENABLE_I18N */ + +static reg_errcode_t +transit_state_bkref (re_match_context_t *mctx, const re_node_set *nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx i; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + + for (i = 0; i < nodes->nelem; ++i) + { + Idx dest_str_idx, prev_nelem, bkc_idx; + Idx node_idx = nodes->elems[i]; + unsigned int context; + const re_token_t *node = dfa->nodes + node_idx; + re_node_set *new_dest_nodes; + + /* Check whether 'node' is a backreference or not. */ + if (node->type != OP_BACK_REF) + continue; + + if (node->constraint) + { + context = re_string_context_at (&mctx->input, cur_str_idx, + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context)) + continue; + } + + /* 'node' is a backreference. + Check the substring which the substring matched. */ + bkc_idx = mctx->nbkref_ents; + err = get_subexp (mctx, node_idx, cur_str_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + + /* And add the epsilon closures (which is 'new_dest_nodes') of + the backreference to appropriate state_log. */ +#ifdef DEBUG + assert (dfa->nexts[node_idx] != -1); +#endif + for (; bkc_idx < mctx->nbkref_ents; ++bkc_idx) + { + Idx subexp_len; + re_dfastate_t *dest_state; + struct re_backref_cache_entry *bkref_ent; + bkref_ent = mctx->bkref_ents + bkc_idx; + if (bkref_ent->node != node_idx || bkref_ent->str_idx != cur_str_idx) + continue; + subexp_len = bkref_ent->subexp_to - bkref_ent->subexp_from; + new_dest_nodes = (subexp_len == 0 + ? dfa->eclosures + dfa->edests[node_idx].elems[0] + : dfa->eclosures + dfa->nexts[node_idx]); + dest_str_idx = (cur_str_idx + bkref_ent->subexp_to + - bkref_ent->subexp_from); + context = re_string_context_at (&mctx->input, dest_str_idx - 1, + mctx->eflags); + dest_state = mctx->state_log[dest_str_idx]; + prev_nelem = ((mctx->state_log[cur_str_idx] == NULL) ? 0 + : mctx->state_log[cur_str_idx]->nodes.nelem); + /* Add 'new_dest_node' to state_log. */ + if (dest_state == NULL) + { + mctx->state_log[dest_str_idx] + = re_acquire_state_context (&err, dfa, new_dest_nodes, + context); + if (__glibc_unlikely (mctx->state_log[dest_str_idx] == NULL + && err != REG_NOERROR)) + goto free_return; + } + else + { + re_node_set dest_nodes; + err = re_node_set_init_union (&dest_nodes, + dest_state->entrance_nodes, + new_dest_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&dest_nodes); + goto free_return; + } + mctx->state_log[dest_str_idx] + = re_acquire_state_context (&err, dfa, &dest_nodes, context); + re_node_set_free (&dest_nodes); + if (__glibc_unlikely (mctx->state_log[dest_str_idx] == NULL + && err != REG_NOERROR)) + goto free_return; + } + /* We need to check recursively if the backreference can epsilon + transit. */ + if (subexp_len == 0 + && mctx->state_log[cur_str_idx]->nodes.nelem > prev_nelem) + { + err = check_subexp_matching_top (mctx, new_dest_nodes, + cur_str_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + err = transit_state_bkref (mctx, new_dest_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + goto free_return; + } + } + } + err = REG_NOERROR; + free_return: + return err; +} + +/* Enumerate all the candidates which the backreference BKREF_NODE can match + at BKREF_STR_IDX, and register them by match_ctx_add_entry(). + Note that we might collect inappropriate candidates here. + However, the cost of checking them strictly here is too high, then we + delay these checking for prune_impossible_nodes(). */ + +static reg_errcode_t +__attribute_warn_unused_result__ +get_subexp (re_match_context_t *mctx, Idx bkref_node, Idx bkref_str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx subexp_num, sub_top_idx; + const char *buf = (const char *) re_string_get_buffer (&mctx->input); + /* Return if we have already checked BKREF_NODE at BKREF_STR_IDX. */ + Idx cache_idx = search_cur_bkref_entry (mctx, bkref_str_idx); + if (cache_idx != -1) + { + const struct re_backref_cache_entry *entry + = mctx->bkref_ents + cache_idx; + do + if (entry->node == bkref_node) + return REG_NOERROR; /* We already checked it. */ + while (entry++->more); + } + + subexp_num = dfa->nodes[bkref_node].opr.idx; + + /* For each sub expression */ + for (sub_top_idx = 0; sub_top_idx < mctx->nsub_tops; ++sub_top_idx) + { + reg_errcode_t err; + re_sub_match_top_t *sub_top = mctx->sub_tops[sub_top_idx]; + re_sub_match_last_t *sub_last; + Idx sub_last_idx, sl_str, bkref_str_off; + + if (dfa->nodes[sub_top->node].opr.idx != subexp_num) + continue; /* It isn't related. */ + + sl_str = sub_top->str_idx; + bkref_str_off = bkref_str_idx; + /* At first, check the last node of sub expressions we already + evaluated. */ + for (sub_last_idx = 0; sub_last_idx < sub_top->nlasts; ++sub_last_idx) + { + regoff_t sl_str_diff; + sub_last = sub_top->lasts[sub_last_idx]; + sl_str_diff = sub_last->str_idx - sl_str; + /* The matched string by the sub expression match with the substring + at the back reference? */ + if (sl_str_diff > 0) + { + if (__glibc_unlikely (bkref_str_off + sl_str_diff + > mctx->input.valid_len)) + { + /* Not enough chars for a successful match. */ + if (bkref_str_off + sl_str_diff > mctx->input.len) + break; + + err = clean_state_log_if_needed (mctx, + bkref_str_off + + sl_str_diff); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + buf = (const char *) re_string_get_buffer (&mctx->input); + } + if (memcmp (buf + bkref_str_off, buf + sl_str, sl_str_diff) != 0) + /* We don't need to search this sub expression any more. */ + break; + } + bkref_str_off += sl_str_diff; + sl_str += sl_str_diff; + err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node, + bkref_str_idx); + + /* Reload buf, since the preceding call might have reallocated + the buffer. */ + buf = (const char *) re_string_get_buffer (&mctx->input); + + if (err == REG_NOMATCH) + continue; + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + + if (sub_last_idx < sub_top->nlasts) + continue; + if (sub_last_idx > 0) + ++sl_str; + /* Then, search for the other last nodes of the sub expression. */ + for (; sl_str <= bkref_str_idx; ++sl_str) + { + Idx cls_node; + regoff_t sl_str_off; + const re_node_set *nodes; + sl_str_off = sl_str - sub_top->str_idx; + /* The matched string by the sub expression match with the substring + at the back reference? */ + if (sl_str_off > 0) + { + if (__glibc_unlikely (bkref_str_off >= mctx->input.valid_len)) + { + /* If we are at the end of the input, we cannot match. */ + if (bkref_str_off >= mctx->input.len) + break; + + err = extend_buffers (mctx, bkref_str_off + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + + buf = (const char *) re_string_get_buffer (&mctx->input); + } + if (buf [bkref_str_off++] != buf[sl_str - 1]) + break; /* We don't need to search this sub expression + any more. */ + } + if (mctx->state_log[sl_str] == NULL) + continue; + /* Does this state have a ')' of the sub expression? */ + nodes = &mctx->state_log[sl_str]->nodes; + cls_node = find_subexp_node (dfa, nodes, subexp_num, + OP_CLOSE_SUBEXP); + if (cls_node == -1) + continue; /* No. */ + if (sub_top->path == NULL) + { + sub_top->path = calloc (sizeof (state_array_t), + sl_str - sub_top->str_idx + 1); + if (sub_top->path == NULL) + return REG_ESPACE; + } + /* Can the OP_OPEN_SUBEXP node arrive the OP_CLOSE_SUBEXP node + in the current context? */ + err = check_arrival (mctx, sub_top->path, sub_top->node, + sub_top->str_idx, cls_node, sl_str, + OP_CLOSE_SUBEXP); + if (err == REG_NOMATCH) + continue; + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + sub_last = match_ctx_add_sublast (sub_top, cls_node, sl_str); + if (__glibc_unlikely (sub_last == NULL)) + return REG_ESPACE; + err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node, + bkref_str_idx); + buf = (const char *) re_string_get_buffer (&mctx->input); + if (err == REG_NOMATCH) + continue; + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + } + return REG_NOERROR; +} + +/* Helper functions for get_subexp(). */ + +/* Check SUB_LAST can arrive to the back reference BKREF_NODE at BKREF_STR. + If it can arrive, register the sub expression expressed with SUB_TOP + and SUB_LAST. */ + +static reg_errcode_t +get_subexp_sub (re_match_context_t *mctx, const re_sub_match_top_t *sub_top, + re_sub_match_last_t *sub_last, Idx bkref_node, Idx bkref_str) +{ + reg_errcode_t err; + Idx to_idx; + /* Can the subexpression arrive the back reference? */ + err = check_arrival (mctx, &sub_last->path, sub_last->node, + sub_last->str_idx, bkref_node, bkref_str, + OP_OPEN_SUBEXP); + if (err != REG_NOERROR) + return err; + err = match_ctx_add_entry (mctx, bkref_node, bkref_str, sub_top->str_idx, + sub_last->str_idx); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + to_idx = bkref_str + sub_last->str_idx - sub_top->str_idx; + return clean_state_log_if_needed (mctx, to_idx); +} + +/* Find the first node which is '(' or ')' and whose index is SUBEXP_IDX. + Search '(' if FL_OPEN, or search ')' otherwise. + TODO: This function isn't efficient... + Because there might be more than one nodes whose types are + OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all + nodes. + E.g. RE: (a){2} */ + +static Idx +find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes, + Idx subexp_idx, int type) +{ + Idx cls_idx; + for (cls_idx = 0; cls_idx < nodes->nelem; ++cls_idx) + { + Idx cls_node = nodes->elems[cls_idx]; + const re_token_t *node = dfa->nodes + cls_node; + if (node->type == type + && node->opr.idx == subexp_idx) + return cls_node; + } + return -1; +} + +/* Check whether the node TOP_NODE at TOP_STR can arrive to the node + LAST_NODE at LAST_STR. We record the path onto PATH since it will be + heavily reused. + Return REG_NOERROR if it can arrive, or REG_NOMATCH otherwise. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +check_arrival (re_match_context_t *mctx, state_array_t *path, Idx top_node, + Idx top_str, Idx last_node, Idx last_str, int type) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err = REG_NOERROR; + Idx subexp_num, backup_cur_idx, str_idx, null_cnt; + re_dfastate_t *cur_state = NULL; + re_node_set *cur_nodes, next_nodes; + re_dfastate_t **backup_state_log; + unsigned int context; + + subexp_num = dfa->nodes[top_node].opr.idx; + /* Extend the buffer if we need. */ + if (__glibc_unlikely (path->alloc < last_str + mctx->max_mb_elem_len + 1)) + { + re_dfastate_t **new_array; + Idx old_alloc = path->alloc; + Idx incr_alloc = last_str + mctx->max_mb_elem_len + 1; + Idx new_alloc; + if (__glibc_unlikely (IDX_MAX - old_alloc < incr_alloc)) + return REG_ESPACE; + new_alloc = old_alloc + incr_alloc; + if (__glibc_unlikely (SIZE_MAX / sizeof (re_dfastate_t *) < new_alloc)) + return REG_ESPACE; + new_array = re_realloc (path->array, re_dfastate_t *, new_alloc); + if (__glibc_unlikely (new_array == NULL)) + return REG_ESPACE; + path->array = new_array; + path->alloc = new_alloc; + memset (new_array + old_alloc, '\0', + sizeof (re_dfastate_t *) * (path->alloc - old_alloc)); + } + + str_idx = path->next_idx ? path->next_idx : top_str; + + /* Temporary modify MCTX. */ + backup_state_log = mctx->state_log; + backup_cur_idx = mctx->input.cur_idx; + mctx->state_log = path->array; + mctx->input.cur_idx = str_idx; + + /* Setup initial node set. */ + context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags); + if (str_idx == top_str) + { + err = re_node_set_init_1 (&next_nodes, top_node); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + } + else + { + cur_state = mctx->state_log[str_idx]; + if (cur_state && cur_state->has_backref) + { + err = re_node_set_init_copy (&next_nodes, &cur_state->nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + else + re_node_set_init_empty (&next_nodes); + } + if (str_idx == top_str || (cur_state && cur_state->has_backref)) + { + if (next_nodes.nelem) + { + err = expand_bkref_cache (mctx, &next_nodes, str_idx, + subexp_num, type); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + } + cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context); + if (__glibc_unlikely (cur_state == NULL && err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + mctx->state_log[str_idx] = cur_state; + } + + for (null_cnt = 0; str_idx < last_str && null_cnt <= mctx->max_mb_elem_len;) + { + re_node_set_empty (&next_nodes); + if (mctx->state_log[str_idx + 1]) + { + err = re_node_set_merge (&next_nodes, + &mctx->state_log[str_idx + 1]->nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + } + if (cur_state) + { + err = check_arrival_add_next_nodes (mctx, str_idx, + &cur_state->non_eps_nodes, + &next_nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + } + ++str_idx; + if (next_nodes.nelem) + { + err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + err = expand_bkref_cache (mctx, &next_nodes, str_idx, + subexp_num, type); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + } + context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags); + cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context); + if (__glibc_unlikely (cur_state == NULL && err != REG_NOERROR)) + { + re_node_set_free (&next_nodes); + return err; + } + mctx->state_log[str_idx] = cur_state; + null_cnt = cur_state == NULL ? null_cnt + 1 : 0; + } + re_node_set_free (&next_nodes); + cur_nodes = (mctx->state_log[last_str] == NULL ? NULL + : &mctx->state_log[last_str]->nodes); + path->next_idx = str_idx; + + /* Fix MCTX. */ + mctx->state_log = backup_state_log; + mctx->input.cur_idx = backup_cur_idx; + + /* Then check the current node set has the node LAST_NODE. */ + if (cur_nodes != NULL && re_node_set_contains (cur_nodes, last_node)) + return REG_NOERROR; + + return REG_NOMATCH; +} + +/* Helper functions for check_arrival. */ + +/* Calculate the destination nodes of CUR_NODES at STR_IDX, and append them + to NEXT_NODES. + TODO: This function is similar to the functions transit_state*(), + however this function has many additional works. + Can't we unify them? */ + +static reg_errcode_t +__attribute_warn_unused_result__ +check_arrival_add_next_nodes (re_match_context_t *mctx, Idx str_idx, + re_node_set *cur_nodes, re_node_set *next_nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + bool ok; + Idx cur_idx; +#ifdef RE_ENABLE_I18N + reg_errcode_t err = REG_NOERROR; +#endif + re_node_set union_set; + re_node_set_init_empty (&union_set); + for (cur_idx = 0; cur_idx < cur_nodes->nelem; ++cur_idx) + { + int naccepted = 0; + Idx cur_node = cur_nodes->elems[cur_idx]; +#ifdef DEBUG + re_token_type_t type = dfa->nodes[cur_node].type; + assert (!IS_EPSILON_NODE (type)); +#endif +#ifdef RE_ENABLE_I18N + /* If the node may accept "multi byte". */ + if (dfa->nodes[cur_node].accept_mb) + { + naccepted = check_node_accept_bytes (dfa, cur_node, &mctx->input, + str_idx); + if (naccepted > 1) + { + re_dfastate_t *dest_state; + Idx next_node = dfa->nexts[cur_node]; + Idx next_idx = str_idx + naccepted; + dest_state = mctx->state_log[next_idx]; + re_node_set_empty (&union_set); + if (dest_state) + { + err = re_node_set_merge (&union_set, &dest_state->nodes); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&union_set); + return err; + } + } + ok = re_node_set_insert (&union_set, next_node); + if (__glibc_unlikely (! ok)) + { + re_node_set_free (&union_set); + return REG_ESPACE; + } + mctx->state_log[next_idx] = re_acquire_state (&err, dfa, + &union_set); + if (__glibc_unlikely (mctx->state_log[next_idx] == NULL + && err != REG_NOERROR)) + { + re_node_set_free (&union_set); + return err; + } + } + } +#endif /* RE_ENABLE_I18N */ + if (naccepted + || check_node_accept (mctx, dfa->nodes + cur_node, str_idx)) + { + ok = re_node_set_insert (next_nodes, dfa->nexts[cur_node]); + if (__glibc_unlikely (! ok)) + { + re_node_set_free (&union_set); + return REG_ESPACE; + } + } + } + re_node_set_free (&union_set); + return REG_NOERROR; +} + +/* For all the nodes in CUR_NODES, add the epsilon closures of them to + CUR_NODES, however exclude the nodes which are: + - inside the sub expression whose number is EX_SUBEXP, if FL_OPEN. + - out of the sub expression whose number is EX_SUBEXP, if !FL_OPEN. +*/ + +static reg_errcode_t +check_arrival_expand_ecl (const re_dfa_t *dfa, re_node_set *cur_nodes, + Idx ex_subexp, int type) +{ + reg_errcode_t err; + Idx idx, outside_node; + re_node_set new_nodes; +#ifdef DEBUG + assert (cur_nodes->nelem); +#endif + err = re_node_set_alloc (&new_nodes, cur_nodes->nelem); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + /* Create a new node set NEW_NODES with the nodes which are epsilon + closures of the node in CUR_NODES. */ + + for (idx = 0; idx < cur_nodes->nelem; ++idx) + { + Idx cur_node = cur_nodes->elems[idx]; + const re_node_set *eclosure = dfa->eclosures + cur_node; + outside_node = find_subexp_node (dfa, eclosure, ex_subexp, type); + if (outside_node == -1) + { + /* There are no problematic nodes, just merge them. */ + err = re_node_set_merge (&new_nodes, eclosure); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&new_nodes); + return err; + } + } + else + { + /* There are problematic nodes, re-calculate incrementally. */ + err = check_arrival_expand_ecl_sub (dfa, &new_nodes, cur_node, + ex_subexp, type); + if (__glibc_unlikely (err != REG_NOERROR)) + { + re_node_set_free (&new_nodes); + return err; + } + } + } + re_node_set_free (cur_nodes); + *cur_nodes = new_nodes; + return REG_NOERROR; +} + +/* Helper function for check_arrival_expand_ecl. + Check incrementally the epsilon closure of TARGET, and if it isn't + problematic append it to DST_NODES. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +check_arrival_expand_ecl_sub (const re_dfa_t *dfa, re_node_set *dst_nodes, + Idx target, Idx ex_subexp, int type) +{ + Idx cur_node; + for (cur_node = target; !re_node_set_contains (dst_nodes, cur_node);) + { + bool ok; + + if (dfa->nodes[cur_node].type == type + && dfa->nodes[cur_node].opr.idx == ex_subexp) + { + if (type == OP_CLOSE_SUBEXP) + { + ok = re_node_set_insert (dst_nodes, cur_node); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + } + break; + } + ok = re_node_set_insert (dst_nodes, cur_node); + if (__glibc_unlikely (! ok)) + return REG_ESPACE; + if (dfa->edests[cur_node].nelem == 0) + break; + if (dfa->edests[cur_node].nelem == 2) + { + reg_errcode_t err; + err = check_arrival_expand_ecl_sub (dfa, dst_nodes, + dfa->edests[cur_node].elems[1], + ex_subexp, type); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + cur_node = dfa->edests[cur_node].elems[0]; + } + return REG_NOERROR; +} + + +/* For all the back references in the current state, calculate the + destination of the back references by the appropriate entry + in MCTX->BKREF_ENTS. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +expand_bkref_cache (re_match_context_t *mctx, re_node_set *cur_nodes, + Idx cur_str, Idx subexp_num, int type) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx cache_idx_start = search_cur_bkref_entry (mctx, cur_str); + struct re_backref_cache_entry *ent; + + if (cache_idx_start == -1) + return REG_NOERROR; + + restart: + ent = mctx->bkref_ents + cache_idx_start; + do + { + Idx to_idx, next_node; + + /* Is this entry ENT is appropriate? */ + if (!re_node_set_contains (cur_nodes, ent->node)) + continue; /* No. */ + + to_idx = cur_str + ent->subexp_to - ent->subexp_from; + /* Calculate the destination of the back reference, and append it + to MCTX->STATE_LOG. */ + if (to_idx == cur_str) + { + /* The backreference did epsilon transit, we must re-check all the + node in the current state. */ + re_node_set new_dests; + reg_errcode_t err2, err3; + next_node = dfa->edests[ent->node].elems[0]; + if (re_node_set_contains (cur_nodes, next_node)) + continue; + err = re_node_set_init_1 (&new_dests, next_node); + err2 = check_arrival_expand_ecl (dfa, &new_dests, subexp_num, type); + err3 = re_node_set_merge (cur_nodes, &new_dests); + re_node_set_free (&new_dests); + if (__glibc_unlikely (err != REG_NOERROR || err2 != REG_NOERROR + || err3 != REG_NOERROR)) + { + err = (err != REG_NOERROR ? err + : (err2 != REG_NOERROR ? err2 : err3)); + return err; + } + /* TODO: It is still inefficient... */ + goto restart; + } + else + { + re_node_set union_set; + next_node = dfa->nexts[ent->node]; + if (mctx->state_log[to_idx]) + { + bool ok; + if (re_node_set_contains (&mctx->state_log[to_idx]->nodes, + next_node)) + continue; + err = re_node_set_init_copy (&union_set, + &mctx->state_log[to_idx]->nodes); + ok = re_node_set_insert (&union_set, next_node); + if (__glibc_unlikely (err != REG_NOERROR || ! ok)) + { + re_node_set_free (&union_set); + err = err != REG_NOERROR ? err : REG_ESPACE; + return err; + } + } + else + { + err = re_node_set_init_1 (&union_set, next_node); + if (__glibc_unlikely (err != REG_NOERROR)) + return err; + } + mctx->state_log[to_idx] = re_acquire_state (&err, dfa, &union_set); + re_node_set_free (&union_set); + if (__glibc_unlikely (mctx->state_log[to_idx] == NULL + && err != REG_NOERROR)) + return err; + } + } + while (ent++->more); + return REG_NOERROR; +} + +/* Build transition table for the state. + Return true if successful. */ + +static bool +build_trtable (const re_dfa_t *dfa, re_dfastate_t *state) +{ + reg_errcode_t err; + Idx i, j; + int ch; + bool need_word_trtable = false; + bitset_word_t elem, mask; + bool dests_node_malloced = false; + bool dest_states_malloced = false; + Idx ndests; /* Number of the destination states from 'state'. */ + re_dfastate_t **trtable; + re_dfastate_t **dest_states = NULL, **dest_states_word, **dest_states_nl; + re_node_set follows, *dests_node; + bitset_t *dests_ch; + bitset_t acceptable; + + struct dests_alloc + { + re_node_set dests_node[SBC_MAX]; + bitset_t dests_ch[SBC_MAX]; + } *dests_alloc; + + /* We build DFA states which corresponds to the destination nodes + from 'state'. 'dests_node[i]' represents the nodes which i-th + destination state contains, and 'dests_ch[i]' represents the + characters which i-th destination state accepts. */ + if (__libc_use_alloca (sizeof (struct dests_alloc))) + dests_alloc = (struct dests_alloc *) alloca (sizeof (struct dests_alloc)); + else + { + dests_alloc = re_malloc (struct dests_alloc, 1); + if (__glibc_unlikely (dests_alloc == NULL)) + return false; + dests_node_malloced = true; + } + dests_node = dests_alloc->dests_node; + dests_ch = dests_alloc->dests_ch; + + /* Initialize transition table. */ + state->word_trtable = state->trtable = NULL; + + /* At first, group all nodes belonging to 'state' into several + destinations. */ + ndests = group_nodes_into_DFAstates (dfa, state, dests_node, dests_ch); + if (__glibc_unlikely (ndests <= 0)) + { + if (dests_node_malloced) + re_free (dests_alloc); + /* Return false in case of an error, true otherwise. */ + if (ndests == 0) + { + state->trtable = (re_dfastate_t **) + calloc (sizeof (re_dfastate_t *), SBC_MAX); + if (__glibc_unlikely (state->trtable == NULL)) + return false; + return true; + } + return false; + } + + err = re_node_set_alloc (&follows, ndests + 1); + if (__glibc_unlikely (err != REG_NOERROR)) + goto out_free; + + /* Avoid arithmetic overflow in size calculation. */ + size_t ndests_max + = ((SIZE_MAX - (sizeof (re_node_set) + sizeof (bitset_t)) * SBC_MAX) + / (3 * sizeof (re_dfastate_t *))); + if (__glibc_unlikely (ndests_max < ndests)) + goto out_free; + + if (__libc_use_alloca ((sizeof (re_node_set) + sizeof (bitset_t)) * SBC_MAX + + ndests * 3 * sizeof (re_dfastate_t *))) + dest_states = (re_dfastate_t **) + alloca (ndests * 3 * sizeof (re_dfastate_t *)); + else + { + dest_states = re_malloc (re_dfastate_t *, ndests * 3); + if (__glibc_unlikely (dest_states == NULL)) + { +out_free: + if (dest_states_malloced) + re_free (dest_states); + re_node_set_free (&follows); + for (i = 0; i < ndests; ++i) + re_node_set_free (dests_node + i); + if (dests_node_malloced) + re_free (dests_alloc); + return false; + } + dest_states_malloced = true; + } + dest_states_word = dest_states + ndests; + dest_states_nl = dest_states_word + ndests; + bitset_empty (acceptable); + + /* Then build the states for all destinations. */ + for (i = 0; i < ndests; ++i) + { + Idx next_node; + re_node_set_empty (&follows); + /* Merge the follows of this destination states. */ + for (j = 0; j < dests_node[i].nelem; ++j) + { + next_node = dfa->nexts[dests_node[i].elems[j]]; + if (next_node != -1) + { + err = re_node_set_merge (&follows, dfa->eclosures + next_node); + if (__glibc_unlikely (err != REG_NOERROR)) + goto out_free; + } + } + dest_states[i] = re_acquire_state_context (&err, dfa, &follows, 0); + if (__glibc_unlikely (dest_states[i] == NULL && err != REG_NOERROR)) + goto out_free; + /* If the new state has context constraint, + build appropriate states for these contexts. */ + if (dest_states[i]->has_constraint) + { + dest_states_word[i] = re_acquire_state_context (&err, dfa, &follows, + CONTEXT_WORD); + if (__glibc_unlikely (dest_states_word[i] == NULL + && err != REG_NOERROR)) + goto out_free; + + if (dest_states[i] != dest_states_word[i] && dfa->mb_cur_max > 1) + need_word_trtable = true; + + dest_states_nl[i] = re_acquire_state_context (&err, dfa, &follows, + CONTEXT_NEWLINE); + if (__glibc_unlikely (dest_states_nl[i] == NULL && err != REG_NOERROR)) + goto out_free; + } + else + { + dest_states_word[i] = dest_states[i]; + dest_states_nl[i] = dest_states[i]; + } + bitset_merge (acceptable, dests_ch[i]); + } + + if (!__glibc_unlikely (need_word_trtable)) + { + /* We don't care about whether the following character is a word + character, or we are in a single-byte character set so we can + discern by looking at the character code: allocate a + 256-entry transition table. */ + trtable = state->trtable = + (re_dfastate_t **) calloc (sizeof (re_dfastate_t *), SBC_MAX); + if (__glibc_unlikely (trtable == NULL)) + goto out_free; + + /* For all characters ch...: */ + for (i = 0; i < BITSET_WORDS; ++i) + for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1; + elem; + mask <<= 1, elem >>= 1, ++ch) + if (__glibc_unlikely (elem & 1)) + { + /* There must be exactly one destination which accepts + character ch. See group_nodes_into_DFAstates. */ + for (j = 0; (dests_ch[j][i] & mask) == 0; ++j) + ; + + /* j-th destination accepts the word character ch. */ + if (dfa->word_char[i] & mask) + trtable[ch] = dest_states_word[j]; + else + trtable[ch] = dest_states[j]; + } + } + else + { + /* We care about whether the following character is a word + character, and we are in a multi-byte character set: discern + by looking at the character code: build two 256-entry + transition tables, one starting at trtable[0] and one + starting at trtable[SBC_MAX]. */ + trtable = state->word_trtable = + (re_dfastate_t **) calloc (sizeof (re_dfastate_t *), 2 * SBC_MAX); + if (__glibc_unlikely (trtable == NULL)) + goto out_free; + + /* For all characters ch...: */ + for (i = 0; i < BITSET_WORDS; ++i) + for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1; + elem; + mask <<= 1, elem >>= 1, ++ch) + if (__glibc_unlikely (elem & 1)) + { + /* There must be exactly one destination which accepts + character ch. See group_nodes_into_DFAstates. */ + for (j = 0; (dests_ch[j][i] & mask) == 0; ++j) + ; + + /* j-th destination accepts the word character ch. */ + trtable[ch] = dest_states[j]; + trtable[ch + SBC_MAX] = dest_states_word[j]; + } + } + + /* new line */ + if (bitset_contain (acceptable, NEWLINE_CHAR)) + { + /* The current state accepts newline character. */ + for (j = 0; j < ndests; ++j) + if (bitset_contain (dests_ch[j], NEWLINE_CHAR)) + { + /* k-th destination accepts newline character. */ + trtable[NEWLINE_CHAR] = dest_states_nl[j]; + if (need_word_trtable) + trtable[NEWLINE_CHAR + SBC_MAX] = dest_states_nl[j]; + /* There must be only one destination which accepts + newline. See group_nodes_into_DFAstates. */ + break; + } + } + + if (dest_states_malloced) + re_free (dest_states); + + re_node_set_free (&follows); + for (i = 0; i < ndests; ++i) + re_node_set_free (dests_node + i); + + if (dests_node_malloced) + re_free (dests_alloc); + + return true; +} + +/* Group all nodes belonging to STATE into several destinations. + Then for all destinations, set the nodes belonging to the destination + to DESTS_NODE[i] and set the characters accepted by the destination + to DEST_CH[i]. This function return the number of destinations. */ + +static Idx +group_nodes_into_DFAstates (const re_dfa_t *dfa, const re_dfastate_t *state, + re_node_set *dests_node, bitset_t *dests_ch) +{ + reg_errcode_t err; + bool ok; + Idx i, j, k; + Idx ndests; /* Number of the destinations from 'state'. */ + bitset_t accepts; /* Characters a node can accept. */ + const re_node_set *cur_nodes = &state->nodes; + bitset_empty (accepts); + ndests = 0; + + /* For all the nodes belonging to 'state', */ + for (i = 0; i < cur_nodes->nelem; ++i) + { + re_token_t *node = &dfa->nodes[cur_nodes->elems[i]]; + re_token_type_t type = node->type; + unsigned int constraint = node->constraint; + + /* Enumerate all single byte character this node can accept. */ + if (type == CHARACTER) + bitset_set (accepts, node->opr.c); + else if (type == SIMPLE_BRACKET) + { + bitset_merge (accepts, node->opr.sbcset); + } + else if (type == OP_PERIOD) + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + bitset_merge (accepts, dfa->sb_char); + else +#endif + bitset_set_all (accepts); + if (!(dfa->syntax & RE_DOT_NEWLINE)) + bitset_clear (accepts, '\n'); + if (dfa->syntax & RE_DOT_NOT_NULL) + bitset_clear (accepts, '\0'); + } +#ifdef RE_ENABLE_I18N + else if (type == OP_UTF8_PERIOD) + { + if (ASCII_CHARS % BITSET_WORD_BITS == 0) + memset (accepts, -1, ASCII_CHARS / CHAR_BIT); + else + bitset_merge (accepts, utf8_sb_map); + if (!(dfa->syntax & RE_DOT_NEWLINE)) + bitset_clear (accepts, '\n'); + if (dfa->syntax & RE_DOT_NOT_NULL) + bitset_clear (accepts, '\0'); + } +#endif + else + continue; + + /* Check the 'accepts' and sift the characters which are not + match it the context. */ + if (constraint) + { + if (constraint & NEXT_NEWLINE_CONSTRAINT) + { + bool accepts_newline = bitset_contain (accepts, NEWLINE_CHAR); + bitset_empty (accepts); + if (accepts_newline) + bitset_set (accepts, NEWLINE_CHAR); + else + continue; + } + if (constraint & NEXT_ENDBUF_CONSTRAINT) + { + bitset_empty (accepts); + continue; + } + + if (constraint & NEXT_WORD_CONSTRAINT) + { + bitset_word_t any_set = 0; + if (type == CHARACTER && !node->word_char) + { + bitset_empty (accepts); + continue; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= (dfa->word_char[j] | ~dfa->sb_char[j])); + else +#endif + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= dfa->word_char[j]); + if (!any_set) + continue; + } + if (constraint & NEXT_NOTWORD_CONSTRAINT) + { + bitset_word_t any_set = 0; + if (type == CHARACTER && node->word_char) + { + bitset_empty (accepts); + continue; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= ~(dfa->word_char[j] & dfa->sb_char[j])); + else +#endif + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= ~dfa->word_char[j]); + if (!any_set) + continue; + } + } + + /* Then divide 'accepts' into DFA states, or create a new + state. Above, we make sure that accepts is not empty. */ + for (j = 0; j < ndests; ++j) + { + bitset_t intersec; /* Intersection sets, see below. */ + bitset_t remains; + /* Flags, see below. */ + bitset_word_t has_intersec, not_subset, not_consumed; + + /* Optimization, skip if this state doesn't accept the character. */ + if (type == CHARACTER && !bitset_contain (dests_ch[j], node->opr.c)) + continue; + + /* Enumerate the intersection set of this state and 'accepts'. */ + has_intersec = 0; + for (k = 0; k < BITSET_WORDS; ++k) + has_intersec |= intersec[k] = accepts[k] & dests_ch[j][k]; + /* And skip if the intersection set is empty. */ + if (!has_intersec) + continue; + + /* Then check if this state is a subset of 'accepts'. */ + not_subset = not_consumed = 0; + for (k = 0; k < BITSET_WORDS; ++k) + { + not_subset |= remains[k] = ~accepts[k] & dests_ch[j][k]; + not_consumed |= accepts[k] = accepts[k] & ~dests_ch[j][k]; + } + + /* If this state isn't a subset of 'accepts', create a + new group state, which has the 'remains'. */ + if (not_subset) + { + bitset_copy (dests_ch[ndests], remains); + bitset_copy (dests_ch[j], intersec); + err = re_node_set_init_copy (dests_node + ndests, &dests_node[j]); + if (__glibc_unlikely (err != REG_NOERROR)) + goto error_return; + ++ndests; + } + + /* Put the position in the current group. */ + ok = re_node_set_insert (&dests_node[j], cur_nodes->elems[i]); + if (__glibc_unlikely (! ok)) + goto error_return; + + /* If all characters are consumed, go to next node. */ + if (!not_consumed) + break; + } + /* Some characters remain, create a new group. */ + if (j == ndests) + { + bitset_copy (dests_ch[ndests], accepts); + err = re_node_set_init_1 (dests_node + ndests, cur_nodes->elems[i]); + if (__glibc_unlikely (err != REG_NOERROR)) + goto error_return; + ++ndests; + bitset_empty (accepts); + } + } + return ndests; + error_return: + for (j = 0; j < ndests; ++j) + re_node_set_free (dests_node + j); + return -1; +} + +#ifdef RE_ENABLE_I18N +/* Check how many bytes the node 'dfa->nodes[node_idx]' accepts. + Return the number of the bytes the node accepts. + STR_IDX is the current index of the input string. + + This function handles the nodes which can accept one character, or + one collating element like '.', '[a-z]', opposite to the other nodes + can only accept one byte. */ + +# ifdef _LIBC +# include <locale/weight.h> +# endif + +static int +check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx, + const re_string_t *input, Idx str_idx) +{ + const re_token_t *node = dfa->nodes + node_idx; + int char_len, elem_len; + Idx i; + + if (__glibc_unlikely (node->type == OP_UTF8_PERIOD)) + { + unsigned char c = re_string_byte_at (input, str_idx), d; + if (__glibc_likely (c < 0xc2)) + return 0; + + if (str_idx + 2 > input->len) + return 0; + + d = re_string_byte_at (input, str_idx + 1); + if (c < 0xe0) + return (d < 0x80 || d > 0xbf) ? 0 : 2; + else if (c < 0xf0) + { + char_len = 3; + if (c == 0xe0 && d < 0xa0) + return 0; + } + else if (c < 0xf8) + { + char_len = 4; + if (c == 0xf0 && d < 0x90) + return 0; + } + else if (c < 0xfc) + { + char_len = 5; + if (c == 0xf8 && d < 0x88) + return 0; + } + else if (c < 0xfe) + { + char_len = 6; + if (c == 0xfc && d < 0x84) + return 0; + } + else + return 0; + + if (str_idx + char_len > input->len) + return 0; + + for (i = 1; i < char_len; ++i) + { + d = re_string_byte_at (input, str_idx + i); + if (d < 0x80 || d > 0xbf) + return 0; + } + return char_len; + } + + char_len = re_string_char_size_at (input, str_idx); + if (node->type == OP_PERIOD) + { + if (char_len <= 1) + return 0; + /* FIXME: I don't think this if is needed, as both '\n' + and '\0' are char_len == 1. */ + /* '.' accepts any one character except the following two cases. */ + if ((!(dfa->syntax & RE_DOT_NEWLINE) + && re_string_byte_at (input, str_idx) == '\n') + || ((dfa->syntax & RE_DOT_NOT_NULL) + && re_string_byte_at (input, str_idx) == '\0')) + return 0; + return char_len; + } + + elem_len = re_string_elem_size_at (input, str_idx); + if ((elem_len <= 1 && char_len <= 1) || char_len == 0) + return 0; + + if (node->type == COMPLEX_BRACKET) + { + const re_charset_t *cset = node->opr.mbcset; +# ifdef _LIBC + const unsigned char *pin + = ((const unsigned char *) re_string_get_buffer (input) + str_idx); + Idx j; + uint32_t nrules; +# endif /* _LIBC */ + int match_len = 0; + wchar_t wc = ((cset->nranges || cset->nchar_classes || cset->nmbchars) + ? re_string_wchar_at (input, str_idx) : 0); + + /* match with multibyte character? */ + for (i = 0; i < cset->nmbchars; ++i) + if (wc == cset->mbchars[i]) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + /* match with character_class? */ + for (i = 0; i < cset->nchar_classes; ++i) + { + wctype_t wt = cset->char_classes[i]; + if (__iswctype (wc, wt)) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + } + +# ifdef _LIBC + nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules != 0) + { + unsigned int in_collseq = 0; + const int32_t *table, *indirect; + const unsigned char *weights, *extra; + const char *collseqwc; + + /* match with collating_symbol? */ + if (cset->ncoll_syms) + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); + for (i = 0; i < cset->ncoll_syms; ++i) + { + const unsigned char *coll_sym = extra + cset->coll_syms[i]; + /* Compare the length of input collating element and + the length of current collating element. */ + if (*coll_sym != elem_len) + continue; + /* Compare each bytes. */ + for (j = 0; j < *coll_sym; j++) + if (pin[j] != coll_sym[1 + j]) + break; + if (j == *coll_sym) + { + /* Match if every bytes is equal. */ + match_len = j; + goto check_node_accept_bytes_match; + } + } + + if (cset->nranges) + { + if (elem_len <= char_len) + { + collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC); + in_collseq = __collseq_table_lookup (collseqwc, wc); + } + else + in_collseq = find_collation_sequence_value (pin, elem_len); + } + /* match with range expression? */ + /* FIXME: Implement rational ranges here, too. */ + for (i = 0; i < cset->nranges; ++i) + if (cset->range_starts[i] <= in_collseq + && in_collseq <= cset->range_ends[i]) + { + match_len = elem_len; + goto check_node_accept_bytes_match; + } + + /* match with equivalence_class? */ + if (cset->nequiv_classes) + { + const unsigned char *cp = pin; + table = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + weights = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB); + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB); + int32_t idx = findidx (table, indirect, extra, &cp, elem_len); + int32_t rule = idx >> 24; + idx &= 0xffffff; + if (idx > 0) + { + size_t weight_len = weights[idx]; + for (i = 0; i < cset->nequiv_classes; ++i) + { + int32_t equiv_class_idx = cset->equiv_classes[i]; + int32_t equiv_class_rule = equiv_class_idx >> 24; + equiv_class_idx &= 0xffffff; + if (weights[equiv_class_idx] == weight_len + && equiv_class_rule == rule + && memcmp (weights + idx + 1, + weights + equiv_class_idx + 1, + weight_len) == 0) + { + match_len = elem_len; + goto check_node_accept_bytes_match; + } + } + } + } + } + else +# endif /* _LIBC */ + { + /* match with range expression? */ + for (i = 0; i < cset->nranges; ++i) + { + if (cset->range_starts[i] <= wc && wc <= cset->range_ends[i]) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + } + } + check_node_accept_bytes_match: + if (!cset->non_match) + return match_len; + else + { + if (match_len > 0) + return 0; + else + return (elem_len > char_len) ? elem_len : char_len; + } + } + return 0; +} + +# ifdef _LIBC +static unsigned int +find_collation_sequence_value (const unsigned char *mbs, size_t mbs_len) +{ + uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules == 0) + { + if (mbs_len == 1) + { + /* No valid character. Match it as a single byte character. */ + const unsigned char *collseq = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB); + return collseq[mbs[0]]; + } + return UINT_MAX; + } + else + { + int32_t idx; + const unsigned char *extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); + int32_t extrasize = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB + 1) - extra; + + for (idx = 0; idx < extrasize;) + { + int mbs_cnt; + bool found = false; + int32_t elem_mbs_len; + /* Skip the name of collating element name. */ + idx = idx + extra[idx] + 1; + elem_mbs_len = extra[idx++]; + if (mbs_len == elem_mbs_len) + { + for (mbs_cnt = 0; mbs_cnt < elem_mbs_len; ++mbs_cnt) + if (extra[idx + mbs_cnt] != mbs[mbs_cnt]) + break; + if (mbs_cnt == elem_mbs_len) + /* Found the entry. */ + found = true; + } + /* Skip the byte sequence of the collating element. */ + idx += elem_mbs_len; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + /* Skip the collation sequence value. */ + idx += sizeof (uint32_t); + /* Skip the wide char sequence of the collating element. */ + idx = idx + sizeof (uint32_t) * (*(int32_t *) (extra + idx) + 1); + /* If we found the entry, return the sequence value. */ + if (found) + return *(uint32_t *) (extra + idx); + /* Skip the collation sequence value. */ + idx += sizeof (uint32_t); + } + return UINT_MAX; + } +} +# endif /* _LIBC */ +#endif /* RE_ENABLE_I18N */ + +/* Check whether the node accepts the byte which is IDX-th + byte of the INPUT. */ + +static bool +check_node_accept (const re_match_context_t *mctx, const re_token_t *node, + Idx idx) +{ + unsigned char ch; + ch = re_string_byte_at (&mctx->input, idx); + switch (node->type) + { + case CHARACTER: + if (node->opr.c != ch) + return false; + break; + + case SIMPLE_BRACKET: + if (!bitset_contain (node->opr.sbcset, ch)) + return false; + break; + +#ifdef RE_ENABLE_I18N + case OP_UTF8_PERIOD: + if (ch >= ASCII_CHARS) + return false; + FALLTHROUGH; +#endif + case OP_PERIOD: + if ((ch == '\n' && !(mctx->dfa->syntax & RE_DOT_NEWLINE)) + || (ch == '\0' && (mctx->dfa->syntax & RE_DOT_NOT_NULL))) + return false; + break; + + default: + return false; + } + + if (node->constraint) + { + /* The node has constraints. Check whether the current context + satisfies the constraints. */ + unsigned int context = re_string_context_at (&mctx->input, idx, + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context)) + return false; + } + + return true; +} + +/* Extend the buffers, if the buffers have run out. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +extend_buffers (re_match_context_t *mctx, int min_len) +{ + reg_errcode_t ret; + re_string_t *pstr = &mctx->input; + + /* Avoid overflow. */ + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / sizeof (re_dfastate_t *)) / 2 + <= pstr->bufs_len)) + return REG_ESPACE; + + /* Double the lengths of the buffers, but allocate at least MIN_LEN. */ + ret = re_string_realloc_buffers (pstr, + MAX (min_len, + MIN (pstr->len, pstr->bufs_len * 2))); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + + if (mctx->state_log != NULL) + { + /* And double the length of state_log. */ + /* XXX We have no indication of the size of this buffer. If this + allocation fail we have no indication that the state_log array + does not have the right size. */ + re_dfastate_t **new_array = re_realloc (mctx->state_log, re_dfastate_t *, + pstr->bufs_len + 1); + if (__glibc_unlikely (new_array == NULL)) + return REG_ESPACE; + mctx->state_log = new_array; + } + + /* Then reconstruct the buffers. */ + if (pstr->icase) + { +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + ret = build_wcs_upper_buffer (pstr); + if (__glibc_unlikely (ret != REG_NOERROR)) + return ret; + } + else +#endif /* RE_ENABLE_I18N */ + build_upper_buffer (pstr); + } + else + { +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + build_wcs_buffer (pstr); + else +#endif /* RE_ENABLE_I18N */ + { + if (pstr->trans != NULL) + re_string_translate_buffer (pstr); + } + } + return REG_NOERROR; +} + + +/* Functions for matching context. */ + +/* Initialize MCTX. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +match_ctx_init (re_match_context_t *mctx, int eflags, Idx n) +{ + mctx->eflags = eflags; + mctx->match_last = -1; + if (n > 0) + { + /* Avoid overflow. */ + size_t max_object_size = + MAX (sizeof (struct re_backref_cache_entry), + sizeof (re_sub_match_top_t *)); + if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) < n)) + return REG_ESPACE; + + mctx->bkref_ents = re_malloc (struct re_backref_cache_entry, n); + mctx->sub_tops = re_malloc (re_sub_match_top_t *, n); + if (__glibc_unlikely (mctx->bkref_ents == NULL || mctx->sub_tops == NULL)) + return REG_ESPACE; + } + /* Already zero-ed by the caller. + else + mctx->bkref_ents = NULL; + mctx->nbkref_ents = 0; + mctx->nsub_tops = 0; */ + mctx->abkref_ents = n; + mctx->max_mb_elem_len = 1; + mctx->asub_tops = n; + return REG_NOERROR; +} + +/* Clean the entries which depend on the current input in MCTX. + This function must be invoked when the matcher changes the start index + of the input, or changes the input string. */ + +static void +match_ctx_clean (re_match_context_t *mctx) +{ + Idx st_idx; + for (st_idx = 0; st_idx < mctx->nsub_tops; ++st_idx) + { + Idx sl_idx; + re_sub_match_top_t *top = mctx->sub_tops[st_idx]; + for (sl_idx = 0; sl_idx < top->nlasts; ++sl_idx) + { + re_sub_match_last_t *last = top->lasts[sl_idx]; + re_free (last->path.array); + re_free (last); + } + re_free (top->lasts); + if (top->path) + { + re_free (top->path->array); + re_free (top->path); + } + re_free (top); + } + + mctx->nsub_tops = 0; + mctx->nbkref_ents = 0; +} + +/* Free all the memory associated with MCTX. */ + +static void +match_ctx_free (re_match_context_t *mctx) +{ + /* First, free all the memory associated with MCTX->SUB_TOPS. */ + match_ctx_clean (mctx); + re_free (mctx->sub_tops); + re_free (mctx->bkref_ents); +} + +/* Add a new backreference entry to MCTX. + Note that we assume that caller never call this function with duplicate + entry, and call with STR_IDX which isn't smaller than any existing entry. +*/ + +static reg_errcode_t +__attribute_warn_unused_result__ +match_ctx_add_entry (re_match_context_t *mctx, Idx node, Idx str_idx, Idx from, + Idx to) +{ + if (mctx->nbkref_ents >= mctx->abkref_ents) + { + struct re_backref_cache_entry* new_entry; + new_entry = re_realloc (mctx->bkref_ents, struct re_backref_cache_entry, + mctx->abkref_ents * 2); + if (__glibc_unlikely (new_entry == NULL)) + { + re_free (mctx->bkref_ents); + return REG_ESPACE; + } + mctx->bkref_ents = new_entry; + memset (mctx->bkref_ents + mctx->nbkref_ents, '\0', + sizeof (struct re_backref_cache_entry) * mctx->abkref_ents); + mctx->abkref_ents *= 2; + } + if (mctx->nbkref_ents > 0 + && mctx->bkref_ents[mctx->nbkref_ents - 1].str_idx == str_idx) + mctx->bkref_ents[mctx->nbkref_ents - 1].more = 1; + + mctx->bkref_ents[mctx->nbkref_ents].node = node; + mctx->bkref_ents[mctx->nbkref_ents].str_idx = str_idx; + mctx->bkref_ents[mctx->nbkref_ents].subexp_from = from; + mctx->bkref_ents[mctx->nbkref_ents].subexp_to = to; + + /* This is a cache that saves negative results of check_dst_limits_calc_pos. + If bit N is clear, means that this entry won't epsilon-transition to + an OP_OPEN_SUBEXP or OP_CLOSE_SUBEXP for the N+1-th subexpression. If + it is set, check_dst_limits_calc_pos_1 will recurse and try to find one + such node. + + A backreference does not epsilon-transition unless it is empty, so set + to all zeros if FROM != TO. */ + mctx->bkref_ents[mctx->nbkref_ents].eps_reachable_subexps_map + = (from == to ? -1 : 0); + + mctx->bkref_ents[mctx->nbkref_ents++].more = 0; + if (mctx->max_mb_elem_len < to - from) + mctx->max_mb_elem_len = to - from; + return REG_NOERROR; +} + +/* Return the first entry with the same str_idx, or -1 if none is + found. Note that MCTX->BKREF_ENTS is already sorted by MCTX->STR_IDX. */ + +static Idx +search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx) +{ + Idx left, right, mid, last; + last = right = mctx->nbkref_ents; + for (left = 0; left < right;) + { + mid = (left + right) / 2; + if (mctx->bkref_ents[mid].str_idx < str_idx) + left = mid + 1; + else + right = mid; + } + if (left < last && mctx->bkref_ents[left].str_idx == str_idx) + return left; + else + return -1; +} + +/* Register the node NODE, whose type is OP_OPEN_SUBEXP, and which matches + at STR_IDX. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +match_ctx_add_subtop (re_match_context_t *mctx, Idx node, Idx str_idx) +{ +#ifdef DEBUG + assert (mctx->sub_tops != NULL); + assert (mctx->asub_tops > 0); +#endif + if (__glibc_unlikely (mctx->nsub_tops == mctx->asub_tops)) + { + Idx new_asub_tops = mctx->asub_tops * 2; + re_sub_match_top_t **new_array = re_realloc (mctx->sub_tops, + re_sub_match_top_t *, + new_asub_tops); + if (__glibc_unlikely (new_array == NULL)) + return REG_ESPACE; + mctx->sub_tops = new_array; + mctx->asub_tops = new_asub_tops; + } + mctx->sub_tops[mctx->nsub_tops] = calloc (1, sizeof (re_sub_match_top_t)); + if (__glibc_unlikely (mctx->sub_tops[mctx->nsub_tops] == NULL)) + return REG_ESPACE; + mctx->sub_tops[mctx->nsub_tops]->node = node; + mctx->sub_tops[mctx->nsub_tops++]->str_idx = str_idx; + return REG_NOERROR; +} + +/* Register the node NODE, whose type is OP_CLOSE_SUBEXP, and which matches + at STR_IDX, whose corresponding OP_OPEN_SUBEXP is SUB_TOP. */ + +static re_sub_match_last_t * +match_ctx_add_sublast (re_sub_match_top_t *subtop, Idx node, Idx str_idx) +{ + re_sub_match_last_t *new_entry; + if (__glibc_unlikely (subtop->nlasts == subtop->alasts)) + { + Idx new_alasts = 2 * subtop->alasts + 1; + re_sub_match_last_t **new_array = re_realloc (subtop->lasts, + re_sub_match_last_t *, + new_alasts); + if (__glibc_unlikely (new_array == NULL)) + return NULL; + subtop->lasts = new_array; + subtop->alasts = new_alasts; + } + new_entry = calloc (1, sizeof (re_sub_match_last_t)); + if (__glibc_likely (new_entry != NULL)) + { + subtop->lasts[subtop->nlasts] = new_entry; + new_entry->node = node; + new_entry->str_idx = str_idx; + ++subtop->nlasts; + } + return new_entry; +} + +static void +sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts, + re_dfastate_t **limited_sts, Idx last_node, Idx last_str_idx) +{ + sctx->sifted_states = sifted_sts; + sctx->limited_states = limited_sts; + sctx->last_node = last_node; + sctx->last_str_idx = last_str_idx; + re_node_set_init_empty (&sctx->limits); +} diff --git a/lib/set-permissions.c b/lib/set-permissions.c index e99d51fcd20..38cd30a5a1c 100644 --- a/lib/set-permissions.c +++ b/lib/set-permissions.c @@ -24,7 +24,7 @@ #include "acl-internal.h" #if USE_ACL -# if ! defined HAVE_ACL_FROM_MODE && defined HAVE_ACL_FROM_TEXT /* FreeBSD, IRIX, Tru64 */ +# if ! defined HAVE_ACL_FROM_MODE && defined HAVE_ACL_FROM_TEXT /* FreeBSD, IRIX, Tru64, Cygwin >= 2.5 */ # if HAVE_ACL_GET_FILE && !HAVE_ACL_TYPE_EXTENDED static acl_t @@ -32,7 +32,7 @@ acl_from_mode (mode_t mode) { # if HAVE_ACL_FREE_TEXT /* Tru64 */ char acl_text[] = "u::---,g::---,o::---,"; -# else /* FreeBSD, IRIX */ +# else /* FreeBSD, IRIX, Cygwin >= 2.5 */ char acl_text[] = "u::---,g::---,o::---"; # endif @@ -51,7 +51,7 @@ acl_from_mode (mode_t mode) # endif # endif -# if HAVE_FACL && defined GETACL /* Solaris, Cygwin, not HP-UX */ +# if HAVE_FACL && defined GETACL /* Solaris, Cygwin < 2.5, not HP-UX */ static int set_acls_from_mode (const char *name, int desc, mode_t mode, bool *must_chmod) { @@ -229,14 +229,14 @@ set_acls_from_mode (const char *name, int desc, mode_t mode, bool *must_chmod) if (ret < 0 && errno != EINVAL && errno != ENOTSUP) { if (errno == ENOSYS) - { - *must_chmod = true; - return 0; - } + { + *must_chmod = true; + return 0; + } return -1; } if (ret == 0) - return 0; + return 0; } # endif @@ -256,18 +256,18 @@ set_acls_from_mode (const char *name, int desc, mode_t mode, bool *must_chmod) if (desc != -1) ret = facl (desc, SETACL, - sizeof (entries) / sizeof (aclent_t), entries); + sizeof (entries) / sizeof (aclent_t), entries); else ret = acl (name, SETACL, - sizeof (entries) / sizeof (aclent_t), entries); + sizeof (entries) / sizeof (aclent_t), entries); if (ret < 0) { - if (errno == ENOSYS || errno == EOPNOTSUPP) - { - *must_chmod = true; - return 0; - } - return -1; + if (errno == ENOSYS || errno == EOPNOTSUPP) + { + *must_chmod = true; + return 0; + } + return -1; } return 0; } @@ -483,15 +483,15 @@ context_acl_from_mode (struct permission_context *ctx) static int set_acls (struct permission_context *ctx, const char *name, int desc, - int from_mode, bool *must_chmod, bool *acls_set) + int from_mode, bool *must_chmod, bool *acls_set) { int ret = 0; # if HAVE_ACL_GET_FILE /* POSIX 1003.1e (draft 17 -- abandoned) specific version. */ - /* Linux, FreeBSD, Mac OS X, IRIX, Tru64 */ + /* Linux, FreeBSD, Mac OS X, IRIX, Tru64, Cygwin >= 2.5 */ # if !HAVE_ACL_TYPE_EXTENDED - /* Linux, FreeBSD, IRIX, Tru64 */ + /* Linux, FreeBSD, IRIX, Tru64, Cygwin >= 2.5 */ # ifndef HAVE_ACL_FROM_TEXT # error Must have acl_from_text (see POSIX 1003.1e draft 17). @@ -503,53 +503,53 @@ set_acls (struct permission_context *ctx, const char *name, int desc, if (! ctx->acls_not_supported) { if (ret == 0 && from_mode) - { - if (ctx->acl) - acl_free (ctx->acl); - ctx->acl = acl_from_mode (ctx->mode); - if (ctx->acl == NULL) - ret = -1; - } + { + if (ctx->acl) + acl_free (ctx->acl); + ctx->acl = acl_from_mode (ctx->mode); + if (ctx->acl == NULL) + ret = -1; + } if (ret == 0 && ctx->acl) - { - if (HAVE_ACL_SET_FD && desc != -1) - ret = acl_set_fd (desc, ctx->acl); - else - ret = acl_set_file (name, ACL_TYPE_ACCESS, ctx->acl); - if (ret != 0) - { - if (! acl_errno_valid (errno)) - { - ctx->acls_not_supported = true; - if (from_mode || acl_access_nontrivial (ctx->acl) == 0) - ret = 0; - } - } - else - { - *acls_set = true; - if (S_ISDIR(ctx->mode)) - { - if (! from_mode && ctx->default_acl && - acl_default_nontrivial (ctx->default_acl)) - ret = acl_set_file (name, ACL_TYPE_DEFAULT, - ctx->default_acl); - else - ret = acl_delete_def_file (name); - } - } - } + { + if (HAVE_ACL_SET_FD && desc != -1) + ret = acl_set_fd (desc, ctx->acl); + else + ret = acl_set_file (name, ACL_TYPE_ACCESS, ctx->acl); + if (ret != 0) + { + if (! acl_errno_valid (errno)) + { + ctx->acls_not_supported = true; + if (from_mode || acl_access_nontrivial (ctx->acl) == 0) + ret = 0; + } + } + else + { + *acls_set = true; + if (S_ISDIR(ctx->mode)) + { + if (! from_mode && ctx->default_acl && + acl_default_nontrivial (ctx->default_acl)) + ret = acl_set_file (name, ACL_TYPE_DEFAULT, + ctx->default_acl); + else + ret = acl_delete_def_file (name); + } + } + } } -# if HAVE_ACL_TYPE_NFS4 /* FreeBSD */ +# if HAVE_ACL_TYPE_NFS4 /* FreeBSD */ /* File systems either support POSIX ACLs (for example, ufs) or NFS4 ACLs (for example, zfs). */ /* TODO: Implement setting ACLs once get_permissions() reads them. */ -# endif +# endif # else /* HAVE_ACL_TYPE_EXTENDED */ /* Mac OS X */ @@ -573,38 +573,38 @@ set_acls (struct permission_context *ctx, const char *name, int desc, /* Remove ACLs if the file has ACLs. */ if (HAVE_ACL_GET_FD && desc != -1) - acl = acl_get_fd (desc); + acl = acl_get_fd (desc); else - acl = acl_get_file (name, ACL_TYPE_EXTENDED); + acl = acl_get_file (name, ACL_TYPE_EXTENDED); if (acl) - { - acl_free (acl); - - acl = acl_init (0); - if (acl) - { - if (HAVE_ACL_SET_FD && desc != -1) - ret = acl_set_fd (desc, acl); - else - ret = acl_set_file (name, ACL_TYPE_EXTENDED, acl); - acl_free (acl); - } - else - ret = -1; - } + { + acl_free (acl); + + acl = acl_init (0); + if (acl) + { + if (HAVE_ACL_SET_FD && desc != -1) + ret = acl_set_fd (desc, acl); + else + ret = acl_set_file (name, ACL_TYPE_EXTENDED, acl); + acl_free (acl); + } + else + ret = -1; + } } else { if (HAVE_ACL_SET_FD && desc != -1) - ret = acl_set_fd (desc, ctx->acl); + ret = acl_set_fd (desc, ctx->acl); else - ret = acl_set_file (name, ACL_TYPE_EXTENDED, ctx->acl); + ret = acl_set_file (name, ACL_TYPE_EXTENDED, ctx->acl); if (ret != 0) - { - if (! acl_errno_valid (errno) - && ! acl_extended_nontrivial (ctx->acl)) - ret = 0; - } + { + if (! acl_errno_valid (errno) + && ! acl_extended_nontrivial (ctx->acl)) + ret = 0; + } } *acls_set = true; @@ -626,34 +626,34 @@ set_acls (struct permission_context *ctx, const char *name, int desc, if (ret == 0 && ctx->count) { if (desc != -1) - ret = facl (desc, SETACL, ctx->count, ctx->entries); + ret = facl (desc, SETACL, ctx->count, ctx->entries); else - ret = acl (name, SETACL, ctx->count, ctx->entries); + ret = acl (name, SETACL, ctx->count, ctx->entries); if (ret < 0) - { - if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == EINVAL) - && acl_nontrivial (ctx->count, ctx->entries) == 0) - ret = 0; - } + { + if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == EINVAL) + && acl_nontrivial (ctx->count, ctx->entries) == 0) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # ifdef ACE_GETACL if (ret == 0 && ctx->ace_count) { if (desc != -1) - ret = facl (desc, ACE_SETACL, ctx->ace_count, ctx->ace_entries); + ret = facl (desc, ACE_SETACL, ctx->ace_count, ctx->ace_entries); else - ret = acl (name, ACE_SETACL, ctx->ace_count, ctx->ace_entries); + ret = acl (name, ACE_SETACL, ctx->ace_count, ctx->ace_entries); if (ret < 0) - { - if ((errno == ENOSYS || errno == EINVAL || errno == ENOTSUP) - && acl_ace_nontrivial (ctx->ace_count, ctx->ace_entries) == 0) - ret = 0; - } + { + if ((errno == ENOSYS || errno == EINVAL || errno == ENOTSUP) + && acl_ace_nontrivial (ctx->ace_count, ctx->ace_entries) == 0) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # endif @@ -665,17 +665,17 @@ set_acls (struct permission_context *ctx, const char *name, int desc, if (ret == 0 && ctx->count > 0) { if (desc != -1) - ret = fsetacl (desc, ctx->count, ctx->entries); + ret = fsetacl (desc, ctx->count, ctx->entries); else - ret = setacl (name, ctx->count, ctx->entries); + ret = setacl (name, ctx->count, ctx->entries); if (ret < 0) - { - if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == ENOTSUP) - && (from_mode || !acl_nontrivial (ctx->count, ctx->entries))) - ret = 0; - } + { + if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == ENOTSUP) + && (from_mode || !acl_nontrivial (ctx->count, ctx->entries))) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # if HAVE_ACLV_H @@ -686,13 +686,13 @@ set_acls (struct permission_context *ctx, const char *name, int desc, { ret = acl ((char *) name, ACL_SET, ctx->aclv_count, ctx->aclv_entries); if (ret < 0) - { - if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == EINVAL) - && (from_mode || !aclv_nontrivial (ctx->aclv_count, ctx->aclv_entries))) - ret = 0; - } + { + if ((errno == ENOSYS || errno == EOPNOTSUPP || errno == EINVAL) + && (from_mode || !aclv_nontrivial (ctx->aclv_count, ctx->aclv_entries))) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # endif @@ -711,16 +711,16 @@ set_acls (struct permission_context *ctx, const char *name, int desc, if (ret == 0 && ctx->have_u) { if (desc != -1) - ret = fchacl (desc, &ctx->u.a, ctx->u.a.acl_len); + ret = fchacl (desc, &ctx->u.a, ctx->u.a.acl_len); else - ret = chacl ((char *) name, &ctx->u.a, ctx->u.a.acl_len); + ret = chacl ((char *) name, &ctx->u.a, ctx->u.a.acl_len); if (ret < 0) - { - if (errno == ENOSYS && from_mode) - ret = 0; - } + { + if (errno == ENOSYS && from_mode) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # elif HAVE_ACLSORT /* NonStop Kernel */ @@ -732,12 +732,12 @@ set_acls (struct permission_context *ctx, const char *name, int desc, { ret = acl ((char *) name, ACL_SET, ctx->count, ctx->entries); if (ret != 0) - { - if (!acl_nontrivial (ctx->count, ctx->entries)) - ret = 0; - } + { + if (!acl_nontrivial (ctx->count, ctx->entries)) + ret = 0; + } else - *acls_set = true; + *acls_set = true; } # else /* No ACLs */ @@ -805,7 +805,7 @@ set_permissions (struct permission_context *ctx, const char *name, int desc) { ret = chmod_or_fchmod (name, desc, ctx->mode); if (ret != 0) - return -1; + return -1; } #if USE_ACL @@ -815,18 +815,18 @@ set_permissions (struct permission_context *ctx, const char *name, int desc) int saved_errno = ret ? errno : 0; /* If we can't set an acl which we expect to be able to set, try setting - the permissions to ctx->mode. Due to possible inherited permissions, - we cannot simply chmod. */ + the permissions to ctx->mode. Due to possible inherited permissions, + we cannot simply chmod. */ ret = set_acls (ctx, name, desc, true, &must_chmod, &acls_set); if (! acls_set) - must_chmod = true; + must_chmod = true; if (saved_errno) - { - errno = saved_errno; - ret = -1; - } + { + errno = saved_errno; + ret = -1; + } } #endif @@ -837,10 +837,10 @@ set_permissions (struct permission_context *ctx, const char *name, int desc) ret = chmod_or_fchmod (name, desc, ctx->mode); if (saved_errno) - { - errno = saved_errno; - ret = -1; - } + { + errno = saved_errno; + ret = -1; + } } return ret; diff --git a/lib/sha1.c b/lib/sha1.c index a64a488bcdc..d94c4373a55 100644 --- a/lib/sha1.c +++ b/lib/sha1.c @@ -1,8 +1,7 @@ /* sha1.c - Functions to compute SHA1 message digest of files or memory blocks according to the NIST specification FIPS-180-1. - Copyright (C) 2000-2001, 2003-2006, 2008-2019 Free Software - Foundation, Inc. + Copyright (C) 2000-2001, 2003-2006, 2008-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the @@ -38,11 +37,11 @@ # include "unlocked-io.h" #endif +#include <byteswap.h> #ifdef WORDS_BIGENDIAN # define SWAP(n) (n) #else -# define SWAP(n) \ - (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24)) +# define SWAP(n) bswap_32 (n) #endif #define BLOCKSIZE 32768 @@ -123,21 +122,29 @@ sha1_finish_ctx (struct sha1_ctx *ctx, void *resbuf) } #endif +#ifdef GL_COMPILE_CRYPTO_STREAM + +#include "af_alg.h" + /* Compute SHA1 message digest for bytes read from STREAM. The - resulting message digest number will be written into the 16 bytes + resulting message digest number will be written into the 20 bytes beginning at RESBLOCK. */ int sha1_stream (FILE *stream, void *resblock) { - struct sha1_ctx ctx; - size_t sum; + switch (afalg_stream (stream, "sha1", resblock, SHA1_DIGEST_SIZE)) + { + case 0: return 0; + case -EIO: return 1; + } char *buffer = malloc (BLOCKSIZE + 72); if (!buffer) return 1; - /* Initialize the computation context. */ + struct sha1_ctx ctx; sha1_init_ctx (&ctx); + size_t sum; /* Iterate over full file contents. */ while (1) @@ -151,6 +158,14 @@ sha1_stream (FILE *stream, void *resblock) /* Read block. Take care for partial reads. */ while (1) { + /* Either process a partial fread() from this loop, + or the fread() in afalg_stream may have gotten EOF. + We need to avoid a subsequent fread() as EOF may + not be sticky. For details of such systems, see: + https://sourceware.org/bugzilla/show_bug.cgi?id=1190 */ + if (feof (stream)) + goto process_partial_block; + n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); sum += n; @@ -170,12 +185,6 @@ sha1_stream (FILE *stream, void *resblock) } goto process_partial_block; } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; } /* Process buffer with BLOCKSIZE bytes. Note that @@ -195,6 +204,7 @@ sha1_stream (FILE *stream, void *resblock) free (buffer); return 0; } +#endif #if ! HAVE_OPENSSL_SHA1 /* Compute SHA1 message digest for LEN bytes beginning at BUFFER. The diff --git a/lib/sha1.h b/lib/sha1.h index cfc6842758b..617f7b0169a 100644 --- a/lib/sha1.h +++ b/lib/sha1.h @@ -87,8 +87,11 @@ extern void *sha1_read_ctx (const struct sha1_ctx *ctx, void *resbuf); extern void *sha1_buffer (const char *buffer, size_t len, void *resblock); # endif -/* Compute SHA1 message digest for bytes read from STREAM. The - resulting message digest number will be written into the 20 bytes +/* Compute SHA1 message digest for bytes read from STREAM. + STREAM is an open file stream. Regular files are handled more efficiently. + The contents of STREAM from its current position to its end will be read. + The case that the last operation on STREAM was an 'ungetc' is not supported. + The resulting message digest number will be written into the 20 bytes beginning at RESBLOCK. */ extern int sha1_stream (FILE *stream, void *resblock); diff --git a/lib/sha256.c b/lib/sha256.c index 1b944b59eb7..721e944a79f 100644 --- a/lib/sha256.c +++ b/lib/sha256.c @@ -36,11 +36,11 @@ # include "unlocked-io.h" #endif +#include <byteswap.h> #ifdef WORDS_BIGENDIAN # define SWAP(n) (n) #else -# define SWAP(n) \ - (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24)) +# define SWAP(n) bswap_32 (n) #endif #define BLOCKSIZE 32768 @@ -91,17 +91,17 @@ sha224_init_ctx (struct sha256_ctx *ctx) ctx->buflen = 0; } -/* Copy the value from v into the memory location pointed to by *cp, - If your architecture allows unaligned access this is equivalent to - * (uint32_t *) cp = v */ +/* Copy the value from v into the memory location pointed to by *CP, + If your architecture allows unaligned access, this is equivalent to + * (__typeof__ (v) *) cp = v */ static void set_uint32 (char *cp, uint32_t v) { memcpy (cp, &v, sizeof v); } -/* Put result from CTX in first 32 bytes following RESBUF. The result - must be in little endian byte order. */ +/* Put result from CTX in first 32 bytes following RESBUF. + The result must be in little endian byte order. */ void * sha256_read_ctx (const struct sha256_ctx *ctx, void *resbuf) { @@ -169,21 +169,32 @@ sha224_finish_ctx (struct sha256_ctx *ctx, void *resbuf) } #endif -/* Compute SHA256 message digest for bytes read from STREAM. The - resulting message digest number will be written into the 32 bytes - beginning at RESBLOCK. */ -int -sha256_stream (FILE *stream, void *resblock) +#ifdef GL_COMPILE_CRYPTO_STREAM + +#include "af_alg.h" + +/* Compute message digest for bytes read from STREAM using algorithm ALG. + Write the message digest into RESBLOCK, which contains HASHLEN bytes. + The initial and finishing operations are INIT_CTX and FINISH_CTX. + Return zero if and only if successful. */ +static int +shaxxx_stream (FILE *stream, char const *alg, void *resblock, + ssize_t hashlen, void (*init_ctx) (struct sha256_ctx *), + void *(*finish_ctx) (struct sha256_ctx *, void *)) { - struct sha256_ctx ctx; - size_t sum; + switch (afalg_stream (stream, alg, resblock, hashlen)) + { + case 0: return 0; + case -EIO: return 1; + } char *buffer = malloc (BLOCKSIZE + 72); if (!buffer) return 1; - /* Initialize the computation context. */ - sha256_init_ctx (&ctx); + struct sha256_ctx ctx; + init_ctx (&ctx); + size_t sum; /* Iterate over full file contents. */ while (1) @@ -197,6 +208,14 @@ sha256_stream (FILE *stream, void *resblock) /* Read block. Take care for partial reads. */ while (1) { + /* Either process a partial fread() from this loop, + or the fread() in afalg_stream may have gotten EOF. + We need to avoid a subsequent fread() as EOF may + not be sticky. For details of such systems, see: + https://sourceware.org/bugzilla/show_bug.cgi?id=1190 */ + if (feof (stream)) + goto process_partial_block; + n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); sum += n; @@ -216,12 +235,6 @@ sha256_stream (FILE *stream, void *resblock) } goto process_partial_block; } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; } /* Process buffer with BLOCKSIZE bytes. Note that @@ -237,84 +250,28 @@ sha256_stream (FILE *stream, void *resblock) sha256_process_bytes (buffer, sum, &ctx); /* Construct result in desired memory. */ - sha256_finish_ctx (&ctx, resblock); + finish_ctx (&ctx, resblock); free (buffer); return 0; } -/* FIXME: Avoid code duplication */ int -sha224_stream (FILE *stream, void *resblock) +sha256_stream (FILE *stream, void *resblock) { - struct sha256_ctx ctx; - size_t sum; - - char *buffer = malloc (BLOCKSIZE + 72); - if (!buffer) - return 1; - - /* Initialize the computation context. */ - sha224_init_ctx (&ctx); - - /* Iterate over full file contents. */ - while (1) - { - /* We read the file in blocks of BLOCKSIZE bytes. One call of the - computation function processes the whole buffer so that with the - next round of the loop another block can be read. */ - size_t n; - sum = 0; - - /* Read block. Take care for partial reads. */ - while (1) - { - n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); - - sum += n; - - if (sum == BLOCKSIZE) - break; - - if (n == 0) - { - /* Check for the error flag IFF N == 0, so that we don't - exit the loop after a partial read due to e.g., EAGAIN - or EWOULDBLOCK. */ - if (ferror (stream)) - { - free (buffer); - return 1; - } - goto process_partial_block; - } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; - } - - /* Process buffer with BLOCKSIZE bytes. Note that - BLOCKSIZE % 64 == 0 - */ - sha256_process_block (buffer, BLOCKSIZE, &ctx); - } - - process_partial_block:; - - /* Process any remaining bytes. */ - if (sum > 0) - sha256_process_bytes (buffer, sum, &ctx); + return shaxxx_stream (stream, "sha256", resblock, SHA256_DIGEST_SIZE, + sha256_init_ctx, sha256_finish_ctx); +} - /* Construct result in desired memory. */ - sha224_finish_ctx (&ctx, resblock); - free (buffer); - return 0; +int +sha224_stream (FILE *stream, void *resblock) +{ + return shaxxx_stream (stream, "sha224", resblock, SHA224_DIGEST_SIZE, + sha224_init_ctx, sha224_finish_ctx); } +#endif #if ! HAVE_OPENSSL_SHA256 -/* Compute SHA512 message digest for LEN bytes beginning at BUFFER. The +/* Compute SHA256 message digest for LEN bytes beginning at BUFFER. The result is always in little endian byte order, so that a byte-wise output yields to the wanted ASCII representation of the message digest. */ diff --git a/lib/sha256.h b/lib/sha256.h index a76bdffa191..b1ccb2aeb6b 100644 --- a/lib/sha256.h +++ b/lib/sha256.h @@ -89,8 +89,11 @@ extern void *sha256_buffer (const char *buffer, size_t len, void *resblock); extern void *sha224_buffer (const char *buffer, size_t len, void *resblock); # endif -/* Compute SHA256 (SHA224) message digest for bytes read from STREAM. The - resulting message digest number will be written into the 32 (28) bytes +/* Compute SHA256 (SHA224) message digest for bytes read from STREAM. + STREAM is an open file stream. Regular files are handled more efficiently. + The contents of STREAM from its current position to its end will be read. + The case that the last operation on STREAM was an 'ungetc' is not supported. + The resulting message digest number will be written into the 32 (28) bytes beginning at RESBLOCK. */ extern int sha256_stream (FILE *stream, void *resblock); extern int sha224_stream (FILE *stream, void *resblock); diff --git a/lib/sha512.c b/lib/sha512.c index 2e8c485efb9..e7f5bd5a159 100644 --- a/lib/sha512.c +++ b/lib/sha512.c @@ -36,18 +36,11 @@ # include "unlocked-io.h" #endif +#include <byteswap.h> #ifdef WORDS_BIGENDIAN # define SWAP(n) (n) #else -# define SWAP(n) \ - u64or (u64or (u64or (u64shl (n, 56), \ - u64shl (u64and (n, u64lo (0x0000ff00)), 40)), \ - u64or (u64shl (u64and (n, u64lo (0x00ff0000)), 24), \ - u64shl (u64and (n, u64lo (0xff000000)), 8))), \ - u64or (u64or (u64and (u64shr (n, 8), u64lo (0xff000000)), \ - u64and (u64shr (n, 24), u64lo (0x00ff0000))), \ - u64or (u64and (u64shr (n, 40), u64lo (0x0000ff00)), \ - u64shr (n, 56)))) +# define SWAP(n) bswap_64 (n) #endif #define BLOCKSIZE 32768 @@ -177,21 +170,32 @@ sha384_finish_ctx (struct sha512_ctx *ctx, void *resbuf) } #endif -/* Compute SHA512 message digest for bytes read from STREAM. The - resulting message digest number will be written into the 64 bytes - beginning at RESBLOCK. */ -int -sha512_stream (FILE *stream, void *resblock) +#ifdef GL_COMPILE_CRYPTO_STREAM + +#include "af_alg.h" + +/* Compute message digest for bytes read from STREAM using algorithm ALG. + Write the message digest into RESBLOCK, which contains HASHLEN bytes. + The initial and finishing operations are INIT_CTX and FINISH_CTX. + Return zero if and only if successful. */ +static int +shaxxx_stream (FILE *stream, char const *alg, void *resblock, + ssize_t hashlen, void (*init_ctx) (struct sha512_ctx *), + void *(*finish_ctx) (struct sha512_ctx *, void *)) { - struct sha512_ctx ctx; - size_t sum; + switch (afalg_stream (stream, alg, resblock, hashlen)) + { + case 0: return 0; + case -EIO: return 1; + } char *buffer = malloc (BLOCKSIZE + 72); if (!buffer) return 1; - /* Initialize the computation context. */ - sha512_init_ctx (&ctx); + struct sha512_ctx ctx; + init_ctx (&ctx); + size_t sum; /* Iterate over full file contents. */ while (1) @@ -205,6 +209,14 @@ sha512_stream (FILE *stream, void *resblock) /* Read block. Take care for partial reads. */ while (1) { + /* Either process a partial fread() from this loop, + or the fread() in afalg_stream may have gotten EOF. + We need to avoid a subsequent fread() as EOF may + not be sticky. For details of such systems, see: + https://sourceware.org/bugzilla/show_bug.cgi?id=1190 */ + if (feof (stream)) + goto process_partial_block; + n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); sum += n; @@ -224,12 +236,6 @@ sha512_stream (FILE *stream, void *resblock) } goto process_partial_block; } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; } /* Process buffer with BLOCKSIZE bytes. Note that @@ -245,81 +251,25 @@ sha512_stream (FILE *stream, void *resblock) sha512_process_bytes (buffer, sum, &ctx); /* Construct result in desired memory. */ - sha512_finish_ctx (&ctx, resblock); + finish_ctx (&ctx, resblock); free (buffer); return 0; } -/* FIXME: Avoid code duplication */ int -sha384_stream (FILE *stream, void *resblock) +sha512_stream (FILE *stream, void *resblock) { - struct sha512_ctx ctx; - size_t sum; - - char *buffer = malloc (BLOCKSIZE + 72); - if (!buffer) - return 1; - - /* Initialize the computation context. */ - sha384_init_ctx (&ctx); - - /* Iterate over full file contents. */ - while (1) - { - /* We read the file in blocks of BLOCKSIZE bytes. One call of the - computation function processes the whole buffer so that with the - next round of the loop another block can be read. */ - size_t n; - sum = 0; - - /* Read block. Take care for partial reads. */ - while (1) - { - n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); - - sum += n; - - if (sum == BLOCKSIZE) - break; - - if (n == 0) - { - /* Check for the error flag IFF N == 0, so that we don't - exit the loop after a partial read due to e.g., EAGAIN - or EWOULDBLOCK. */ - if (ferror (stream)) - { - free (buffer); - return 1; - } - goto process_partial_block; - } - - /* We've read at least one byte, so ignore errors. But always - check for EOF, since feof may be true even though N > 0. - Otherwise, we could end up calling fread after EOF. */ - if (feof (stream)) - goto process_partial_block; - } - - /* Process buffer with BLOCKSIZE bytes. Note that - BLOCKSIZE % 128 == 0 - */ - sha512_process_block (buffer, BLOCKSIZE, &ctx); - } - - process_partial_block:; - - /* Process any remaining bytes. */ - if (sum > 0) - sha512_process_bytes (buffer, sum, &ctx); + return shaxxx_stream (stream, "sha512", resblock, SHA512_DIGEST_SIZE, + sha512_init_ctx, sha512_finish_ctx); +} - /* Construct result in desired memory. */ - sha384_finish_ctx (&ctx, resblock); - free (buffer); - return 0; +int +sha384_stream (FILE *stream, void *resblock) +{ + return shaxxx_stream (stream, "sha384", resblock, SHA384_DIGEST_SIZE, + sha384_init_ctx, sha384_finish_ctx); } +#endif #if ! HAVE_OPENSSL_SHA512 /* Compute SHA512 message digest for LEN bytes beginning at BUFFER. The diff --git a/lib/sha512.h b/lib/sha512.h index 3c059d1084e..7e8cc2852aa 100644 --- a/lib/sha512.h +++ b/lib/sha512.h @@ -92,8 +92,11 @@ extern void *sha512_buffer (const char *buffer, size_t len, void *resblock); extern void *sha384_buffer (const char *buffer, size_t len, void *resblock); # endif -/* Compute SHA512 (SHA384) message digest for bytes read from STREAM. The - resulting message digest number will be written into the 64 (48) bytes +/* Compute SHA512 (SHA384) message digest for bytes read from STREAM. + STREAM is an open file stream. Regular files are handled more efficiently. + The contents of STREAM from its current position to its end will be read. + The case that the last operation on STREAM was an 'ungetc' is not supported. + The resulting message digest number will be written into the 64 (48) bytes beginning at RESBLOCK. */ extern int sha512_stream (FILE *stream, void *resblock); extern int sha384_stream (FILE *stream, void *resblock); diff --git a/lib/sig2str.c b/lib/sig2str.c index 56114afde98..cd5bd4d70c7 100644 --- a/lib/sig2str.c +++ b/lib/sig2str.c @@ -1,7 +1,6 @@ /* sig2str.c -- convert between signal names and numbers - Copyright (C) 2002, 2004, 2006, 2009-2019 Free Software Foundation, - Inc. + Copyright (C) 2002, 2004, 2006, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/stat-time.h b/lib/stat-time.h index c7f5725f682..38a1f55a6cf 100644 --- a/lib/stat-time.h +++ b/lib/stat-time.h @@ -168,10 +168,10 @@ get_stat_birthtime (struct stat const *st _GL_UNUSED) #elif defined HAVE_STRUCT_STAT_ST_BIRTHTIMENSEC t.tv_sec = st->st_birthtime; t.tv_nsec = st->st_birthtimensec; -#elif (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +#elif defined _WIN32 && ! defined __CYGWIN__ /* Native Windows platforms (but not Cygwin) put the "file creation time" in st_ctime (!). See - <https://msdn.microsoft.com/en-us/library/14h5k7ff(VS.80).aspx>. */ + <https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/stat-functions>. */ # if _GL_WINDOWS_STAT_TIMESPEC t = st->st_ctim; # else @@ -213,7 +213,7 @@ stat_time_normalize (int result, struct stat *st _GL_UNUSED) #if defined __sun && defined STAT_TIMESPEC if (result == 0) { - long int timespec_resolution = 1000000000; + long int timespec_hz = 1000000000; short int const ts_off[] = { offsetof (struct stat, st_atim), offsetof (struct stat, st_mtim), offsetof (struct stat, st_ctim) }; @@ -221,11 +221,11 @@ stat_time_normalize (int result, struct stat *st _GL_UNUSED) for (i = 0; i < sizeof ts_off / sizeof *ts_off; i++) { struct timespec *ts = (struct timespec *) ((char *) st + ts_off[i]); - long int q = ts->tv_nsec / timespec_resolution; - long int r = ts->tv_nsec % timespec_resolution; + long int q = ts->tv_nsec / timespec_hz; + long int r = ts->tv_nsec % timespec_hz; if (r < 0) { - r += timespec_resolution; + r += timespec_hz; q--; } ts->tv_nsec = r; diff --git a/lib/stdio-impl.h b/lib/stdio-impl.h index 5005754a9a3..4260468b612 100644 --- a/lib/stdio-impl.h +++ b/lib/stdio-impl.h @@ -18,6 +18,12 @@ the same implementation of stdio extension API, except that some fields have different naming conventions, or their access requires some casts. */ +/* Glibc 2.28 made _IO_IN_BACKUP private. For now, work around this + problem by defining it ourselves. FIXME: Do not rely on glibc + internals. */ +#if !defined _IO_IN_BACKUP && defined _IO_EOF_SEEN +# define _IO_IN_BACKUP 0x100 +#endif /* BSD stdio derived implementations. */ @@ -54,25 +60,84 @@ # define _flags pub._flags # define _r pub._r # define _w pub._w +# elif defined __ANDROID__ /* Android */ +# ifdef __LP64__ +# define _gl_flags_file_t int +# else +# define _gl_flags_file_t short +# endif + /* Up to this commit from 2015-10-12 + <https://android.googlesource.com/platform/bionic.git/+/f0141dfab10a4b332769d52fa76631a64741297a> + the innards of FILE were public, and fp_ub could be defined like for OpenBSD, + see <https://android.googlesource.com/platform/bionic.git/+/e78392637d5086384a5631ddfdfa8d7ec8326ee3/libc/stdio/fileext.h> + and <https://android.googlesource.com/platform/bionic.git/+/e78392637d5086384a5631ddfdfa8d7ec8326ee3/libc/stdio/local.h>. + After this commit, the innards of FILE are hidden. */ +# define fp_ ((struct { unsigned char *_p; \ + int _r; \ + int _w; \ + _gl_flags_file_t _flags; \ + _gl_flags_file_t _file; \ + struct { unsigned char *_base; size_t _size; } _bf; \ + int _lbfsize; \ + void *_cookie; \ + void *_close; \ + void *_read; \ + void *_seek; \ + void *_write; \ + struct { unsigned char *_base; size_t _size; } _ext; \ + unsigned char *_up; \ + int _ur; \ + unsigned char _ubuf[3]; \ + unsigned char _nbuf[1]; \ + struct { unsigned char *_base; size_t _size; } _lb; \ + int _blksize; \ + fpos_t _offset; \ + /* More fields, not relevant here. */ \ + } *) fp) # else # define fp_ fp # endif -# if (defined __NetBSD__ && __NetBSD_Version__ >= 105270000) || defined __OpenBSD__ || defined __minix || defined __ANDROID__ /* NetBSD >= 1.5ZA, OpenBSD, Minix 3, Android */ +# if (defined __NetBSD__ && __NetBSD_Version__ >= 105270000) || defined __OpenBSD__ || defined __minix /* NetBSD >= 1.5ZA, OpenBSD, Minix 3 */ /* See <http://cvsweb.netbsd.org/bsdweb.cgi/src/lib/libc/stdio/fileext.h?rev=HEAD&content-type=text/x-cvsweb-markup> - and <https://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib/libc/stdio/fileext.h?rev=HEAD&content-type=text/x-cvsweb-markup> */ + and <https://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib/libc/stdio/fileext.h?rev=HEAD&content-type=text/x-cvsweb-markup> + and <https://github.com/Stichting-MINIX-Research-Foundation/minix/blob/master/lib/libc/stdio/fileext.h> */ struct __sfileext { struct __sbuf _ub; /* ungetc buffer */ /* More fields, not relevant here. */ }; # define fp_ub ((struct __sfileext *) fp->_ext._base)->_ub -# else /* FreeBSD, NetBSD <= 1.5Z, DragonFly, Mac OS X, Cygwin, Android */ +# elif defined __ANDROID__ /* Android */ + struct __sfileext + { + struct { unsigned char *_base; size_t _size; } _ub; /* ungetc buffer */ + /* More fields, not relevant here. */ + }; +# define fp_ub ((struct __sfileext *) fp_->_ext._base)->_ub +# else /* FreeBSD, NetBSD <= 1.5Z, DragonFly, Mac OS X, Cygwin */ # define fp_ub fp_->_ub # endif # define HASUB(fp) (fp_ub._base != NULL) +# if defined __ANDROID__ /* Android */ + /* Needed after this commit from 2016-01-25 + <https://android.googlesource.com/platform/bionic.git/+/e70e0e9267d069bf56a5078c99307e08a7280de7> */ +# ifndef __SEOF +# define __SLBF 1 +# define __SNBF 2 +# define __SRD 4 +# define __SWR 8 +# define __SRW 0x10 +# define __SEOF 0x20 +# define __SERR 0x40 +# endif +# ifndef __SOFF +# define __SOFF 0x1000 +# endif +# endif + #endif @@ -112,7 +177,7 @@ # define _flag __flag # endif -#elif (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ /* newer Windows with MSVC */ +#elif defined _WIN32 && ! defined __CYGWIN__ /* newer Windows with MSVC */ /* <stdio.h> does not define the innards of FILE any more. */ # define WINDOWS_OPAQUE_FILE diff --git a/lib/stdio.in.h b/lib/stdio.in.h index 012bcdd7d02..4a8aa55528b 100644 --- a/lib/stdio.in.h +++ b/lib/stdio.in.h @@ -118,11 +118,18 @@ # include <unistd.h> #endif +/* Android 4.3 declares renameat in <sys/stat.h>, not in <stdio.h>. */ +/* But in any case avoid namespace pollution on glibc systems. */ +#if (@GNULIB_RENAMEAT@ || defined GNULIB_POSIXCHECK) && defined __ANDROID__ \ + && ! defined __GLIBC__ +# include <sys/stat.h> +#endif + /* MSVC declares 'perror' in <stdlib.h>, not in <stdio.h>. We must include it before we #define perror rpl_perror. */ /* But in any case avoid namespace pollution on glibc systems. */ #if (@GNULIB_PERROR@ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) \ + && (defined _WIN32 && ! defined __CYGWIN__) \ && ! defined __GLIBC__ # include <stdlib.h> #endif @@ -133,7 +140,7 @@ it before we #define rename rpl_rename. */ /* But in any case avoid namespace pollution on glibc systems. */ #if (@GNULIB_REMOVE@ || @GNULIB_RENAME@ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) \ + && (defined _WIN32 && ! defined __CYGWIN__) \ && ! defined __GLIBC__ # include <io.h> #endif diff --git a/lib/stdlib.in.h b/lib/stdlib.in.h index 6e3866dd8a1..f829525c104 100644 --- a/lib/stdlib.in.h +++ b/lib/stdlib.in.h @@ -1,7 +1,6 @@ /* A GNU-like <stdlib.h>. - Copyright (C) 1995, 2001-2004, 2006-2019 Free Software Foundation, - Inc. + Copyright (C) 1995, 2001-2004, 2006-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -48,11 +47,14 @@ /* Solaris declares getloadavg() in <sys/loadavg.h>. */ #if (@GNULIB_GETLOADAVG@ || defined GNULIB_POSIXCHECK) && @HAVE_SYS_LOADAVG_H@ +/* OpenIndiana has a bug: <sys/time.h> must be included before + <sys/loadavg.h>. */ +# include <sys/time.h> # include <sys/loadavg.h> #endif /* Native Windows platforms declare mktemp() in <io.h>. */ -#if 0 && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) +#if 0 && (defined _WIN32 && ! defined __CYGWIN__) # include <io.h> #endif @@ -88,9 +90,10 @@ struct random_data # endif #endif -#if (@GNULIB_MKSTEMP@ || @GNULIB_MKSTEMPS@ || @GNULIB_GETSUBOPT@ || defined GNULIB_POSIXCHECK) && ! defined __GLIBC__ && !((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) +#if (@GNULIB_MKSTEMP@ || @GNULIB_MKSTEMPS@ || @GNULIB_MKOSTEMP@ || @GNULIB_MKOSTEMPS@ || @GNULIB_GETSUBOPT@ || defined GNULIB_POSIXCHECK) && ! defined __GLIBC__ && !(defined _WIN32 && ! defined __CYGWIN__) /* On Mac OS X 10.3, only <unistd.h> declares mkstemp. */ /* On Mac OS X 10.5, only <unistd.h> declares mkstemps. */ +/* On Mac OS X 10.13, only <unistd.h> declares mkostemp and mkostemps. */ /* On Cygwin 1.7.1, only <unistd.h> declares getsubopt. */ /* But avoid namespace pollution on glibc systems and native Windows. */ # include <unistd.h> @@ -303,9 +306,18 @@ _GL_WARN_ON_USE (malloc, "malloc is not POSIX compliant everywhere - " _GL_FUNCDECL_RPL (mbtowc, int, (wchar_t *pwc, const char *s, size_t n)); _GL_CXXALIAS_RPL (mbtowc, int, (wchar_t *pwc, const char *s, size_t n)); # else +# if !@HAVE_MBTOWC@ +_GL_FUNCDECL_SYS (mbtowc, int, (wchar_t *pwc, const char *s, size_t n)); +# endif _GL_CXXALIAS_SYS (mbtowc, int, (wchar_t *pwc, const char *s, size_t n)); # endif _GL_CXXALIASWARN (mbtowc); +#elif defined GNULIB_POSIXCHECK +# undef mbtowc +# if HAVE_RAW_DECL_MBTOWC +_GL_WARN_ON_USE (mbtowc, "mbtowc is not portable - " + "use gnulib module mbtowc for portability"); +# endif #endif #if @GNULIB_MKDTEMP@ @@ -570,10 +582,19 @@ _GL_WARN_ON_USE (qsort_r, "qsort_r is not portable - " #if @GNULIB_RANDOM@ -# if !@HAVE_RANDOM@ +# if @REPLACE_RANDOM@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef random +# define random rpl_random +# endif +_GL_FUNCDECL_RPL (random, long, (void)); +_GL_CXXALIAS_RPL (random, long, (void)); +# else +# if !@HAVE_RANDOM@ _GL_FUNCDECL_SYS (random, long, (void)); -# endif +# endif _GL_CXXALIAS_SYS (random, long, (void)); +# endif _GL_CXXALIASWARN (random); #elif defined GNULIB_POSIXCHECK # undef random @@ -584,10 +605,19 @@ _GL_WARN_ON_USE (random, "random is unportable - " #endif #if @GNULIB_RANDOM@ -# if !@HAVE_RANDOM@ +# if @REPLACE_RANDOM@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef srandom +# define srandom rpl_srandom +# endif +_GL_FUNCDECL_RPL (srandom, void, (unsigned int seed)); +_GL_CXXALIAS_RPL (srandom, void, (unsigned int seed)); +# else +# if !@HAVE_RANDOM@ _GL_FUNCDECL_SYS (srandom, void, (unsigned int seed)); -# endif +# endif _GL_CXXALIAS_SYS (srandom, void, (unsigned int seed)); +# endif _GL_CXXALIASWARN (srandom); #elif defined GNULIB_POSIXCHECK # undef srandom @@ -598,31 +628,52 @@ _GL_WARN_ON_USE (srandom, "srandom is unportable - " #endif #if @GNULIB_RANDOM@ -# if !@HAVE_RANDOM@ || !@HAVE_DECL_INITSTATE@ +# if @REPLACE_INITSTATE@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef initstate +# define initstate rpl_initstate +# endif +_GL_FUNCDECL_RPL (initstate, char *, + (unsigned int seed, char *buf, size_t buf_size) + _GL_ARG_NONNULL ((2))); +_GL_CXXALIAS_RPL (initstate, char *, + (unsigned int seed, char *buf, size_t buf_size)); +# else +# if !@HAVE_INITSTATE@ || !@HAVE_DECL_INITSTATE@ _GL_FUNCDECL_SYS (initstate, char *, (unsigned int seed, char *buf, size_t buf_size) _GL_ARG_NONNULL ((2))); -# endif +# endif _GL_CXXALIAS_SYS (initstate, char *, (unsigned int seed, char *buf, size_t buf_size)); +# endif _GL_CXXALIASWARN (initstate); #elif defined GNULIB_POSIXCHECK # undef initstate -# if HAVE_RAW_DECL_INITSTATE_R +# if HAVE_RAW_DECL_INITSTATE _GL_WARN_ON_USE (initstate, "initstate is unportable - " "use gnulib module random for portability"); # endif #endif #if @GNULIB_RANDOM@ -# if !@HAVE_RANDOM@ || !@HAVE_DECL_SETSTATE@ +# if @REPLACE_SETSTATE@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef setstate +# define setstate rpl_setstate +# endif +_GL_FUNCDECL_RPL (setstate, char *, (char *arg_state) _GL_ARG_NONNULL ((1))); +_GL_CXXALIAS_RPL (setstate, char *, (char *arg_state)); +# else +# if !@HAVE_SETSTATE@ || !@HAVE_DECL_SETSTATE@ _GL_FUNCDECL_SYS (setstate, char *, (char *arg_state) _GL_ARG_NONNULL ((1))); -# endif +# endif _GL_CXXALIAS_SYS (setstate, char *, (char *arg_state)); +# endif _GL_CXXALIASWARN (setstate); #elif defined GNULIB_POSIXCHECK # undef setstate -# if HAVE_RAW_DECL_SETSTATE_R +# if HAVE_RAW_DECL_SETSTATE _GL_WARN_ON_USE (setstate, "setstate is unportable - " "use gnulib module random for portability"); # endif @@ -878,6 +929,7 @@ _GL_WARN_ON_USE (setenv, "setenv is unportable - " # if !(defined __cplusplus && defined GNULIB_NAMESPACE) # define strtod rpl_strtod # endif +# define GNULIB_defined_strtod_function 1 _GL_FUNCDECL_RPL (strtod, double, (const char *str, char **endp) _GL_ARG_NONNULL ((1))); _GL_CXXALIAS_RPL (strtod, double, (const char *str, char **endp)); @@ -897,6 +949,32 @@ _GL_WARN_ON_USE (strtod, "strtod is unportable - " # endif #endif +#if @GNULIB_STRTOLD@ + /* Parse a 'long double' from STRING, updating ENDP if appropriate. */ +# if @REPLACE_STRTOLD@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define strtold rpl_strtold +# endif +# define GNULIB_defined_strtold_function 1 +_GL_FUNCDECL_RPL (strtold, long double, (const char *str, char **endp) + _GL_ARG_NONNULL ((1))); +_GL_CXXALIAS_RPL (strtold, long double, (const char *str, char **endp)); +# else +# if !@HAVE_STRTOLD@ +_GL_FUNCDECL_SYS (strtold, long double, (const char *str, char **endp) + _GL_ARG_NONNULL ((1))); +# endif +_GL_CXXALIAS_SYS (strtold, long double, (const char *str, char **endp)); +# endif +_GL_CXXALIASWARN (strtold); +#elif defined GNULIB_POSIXCHECK +# undef strtold +# if HAVE_RAW_DECL_STRTOLD +_GL_WARN_ON_USE (strtold, "strtold is unportable - " + "use gnulib module strtold for portability"); +# endif +#endif + #if @GNULIB_STRTOLL@ /* Parse a signed integer whose textual representation starts at STRING. The integer is expected to be in base BASE (2 <= BASE <= 36); if BASE == 0, diff --git a/lib/str-two-way.h b/lib/str-two-way.h new file mode 100644 index 00000000000..7078c34bdc7 --- /dev/null +++ b/lib/str-two-way.h @@ -0,0 +1,452 @@ +/* Byte-wise substring search, using the Two-Way algorithm. + Copyright (C) 2008-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Written by Eric Blake <ebb9@byu.net>, 2008. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, see <https://www.gnu.org/licenses/>. */ + +/* Before including this file, you need to include <config.h> and + <string.h>, and define: + RETURN_TYPE A macro that expands to the return type. + AVAILABLE(h, h_l, j, n_l) + A macro that returns nonzero if there are + at least N_L bytes left starting at H[J]. + H is 'unsigned char *', H_L, J, and N_L + are 'size_t'; H_L is an lvalue. For + NUL-terminated searches, H_L can be + modified each iteration to avoid having + to compute the end of H up front. + + For case-insensitivity, you may optionally define: + CMP_FUNC(p1, p2, l) A macro that returns 0 iff the first L + characters of P1 and P2 are equal. + CANON_ELEMENT(c) A macro that canonicalizes an element right after + it has been fetched from one of the two strings. + The argument is an 'unsigned char'; the result + must be an 'unsigned char' as well. + + This file undefines the macros documented above, and defines + LONG_NEEDLE_THRESHOLD. +*/ + +#include <limits.h> +#include <stdint.h> + +/* We use the Two-Way string matching algorithm (also known as + Chrochemore-Perrin), which guarantees linear complexity with + constant space. Additionally, for long needles, we also use a bad + character shift table similar to the Boyer-Moore algorithm to + achieve improved (potentially sub-linear) performance. + + See http://www-igm.univ-mlv.fr/~lecroq/string/node26.html#SECTION00260, + https://en.wikipedia.org/wiki/Boyer-Moore_string_search_algorithm, + https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.34.6641&rep=rep1&type=pdf +*/ + +/* Point at which computing a bad-byte shift table is likely to be + worthwhile. Small needles should not compute a table, since it + adds (1 << CHAR_BIT) + NEEDLE_LEN computations of preparation for a + speedup no greater than a factor of NEEDLE_LEN. The larger the + needle, the better the potential performance gain. On the other + hand, on non-POSIX systems with CHAR_BIT larger than eight, the + memory required for the table is prohibitive. */ +#if CHAR_BIT < 10 +# define LONG_NEEDLE_THRESHOLD 32U +#else +# define LONG_NEEDLE_THRESHOLD SIZE_MAX +#endif + +#ifndef MAX +# define MAX(a, b) ((a < b) ? (b) : (a)) +#endif + +#ifndef CANON_ELEMENT +# define CANON_ELEMENT(c) c +#endif +#ifndef CMP_FUNC +# define CMP_FUNC memcmp +#endif + +/* Perform a critical factorization of NEEDLE, of length NEEDLE_LEN. + Return the index of the first byte in the right half, and set + *PERIOD to the global period of the right half. + + The global period of a string is the smallest index (possibly its + length) at which all remaining bytes in the string are repetitions + of the prefix (the last repetition may be a subset of the prefix). + + When NEEDLE is factored into two halves, a local period is the + length of the smallest word that shares a suffix with the left half + and shares a prefix with the right half. All factorizations of a + non-empty NEEDLE have a local period of at least 1 and no greater + than NEEDLE_LEN. + + A critical factorization has the property that the local period + equals the global period. All strings have at least one critical + factorization with the left half smaller than the global period. + And while some strings have more than one critical factorization, + it is provable that with an ordered alphabet, at least one of the + critical factorizations corresponds to a maximal suffix. + + Given an ordered alphabet, a critical factorization can be computed + in linear time, with 2 * NEEDLE_LEN comparisons, by computing the + shorter of two ordered maximal suffixes. The ordered maximal + suffixes are determined by lexicographic comparison while tracking + periodicity. */ +static size_t +critical_factorization (const unsigned char *needle, size_t needle_len, + size_t *period) +{ + /* Index of last byte of left half, or SIZE_MAX. */ + size_t max_suffix, max_suffix_rev; + size_t j; /* Index into NEEDLE for current candidate suffix. */ + size_t k; /* Offset into current period. */ + size_t p; /* Intermediate period. */ + unsigned char a, b; /* Current comparison bytes. */ + + /* Special case NEEDLE_LEN of 1 or 2 (all callers already filtered + out 0-length needles. */ + if (needle_len < 3) + { + *period = 1; + return needle_len - 1; + } + + /* Invariants: + 0 <= j < NEEDLE_LEN - 1 + -1 <= max_suffix{,_rev} < j (treating SIZE_MAX as if it were signed) + min(max_suffix, max_suffix_rev) < global period of NEEDLE + 1 <= p <= global period of NEEDLE + p == global period of the substring NEEDLE[max_suffix{,_rev}+1...j] + 1 <= k <= p + */ + + /* Perform lexicographic search. */ + max_suffix = SIZE_MAX; + j = 0; + k = p = 1; + while (j + k < needle_len) + { + a = CANON_ELEMENT (needle[j + k]); + b = CANON_ELEMENT (needle[max_suffix + k]); + if (a < b) + { + /* Suffix is smaller, period is entire prefix so far. */ + j += k; + k = 1; + p = j - max_suffix; + } + else if (a == b) + { + /* Advance through repetition of the current period. */ + if (k != p) + ++k; + else + { + j += p; + k = 1; + } + } + else /* b < a */ + { + /* Suffix is larger, start over from current location. */ + max_suffix = j++; + k = p = 1; + } + } + *period = p; + + /* Perform reverse lexicographic search. */ + max_suffix_rev = SIZE_MAX; + j = 0; + k = p = 1; + while (j + k < needle_len) + { + a = CANON_ELEMENT (needle[j + k]); + b = CANON_ELEMENT (needle[max_suffix_rev + k]); + if (b < a) + { + /* Suffix is smaller, period is entire prefix so far. */ + j += k; + k = 1; + p = j - max_suffix_rev; + } + else if (a == b) + { + /* Advance through repetition of the current period. */ + if (k != p) + ++k; + else + { + j += p; + k = 1; + } + } + else /* a < b */ + { + /* Suffix is larger, start over from current location. */ + max_suffix_rev = j++; + k = p = 1; + } + } + + /* Choose the shorter suffix. Return the index of the first byte of + the right half, rather than the last byte of the left half. + + For some examples, 'banana' has two critical factorizations, both + exposed by the two lexicographic extreme suffixes of 'anana' and + 'nana', where both suffixes have a period of 2. On the other + hand, with 'aab' and 'bba', both strings have a single critical + factorization of the last byte, with the suffix having a period + of 1. While the maximal lexicographic suffix of 'aab' is 'b', + the maximal lexicographic suffix of 'bba' is 'ba', which is not a + critical factorization. Conversely, the maximal reverse + lexicographic suffix of 'a' works for 'bba', but not 'ab' for + 'aab'. The shorter suffix of the two will always be a critical + factorization. */ + if (max_suffix_rev + 1 < max_suffix + 1) + return max_suffix + 1; + *period = p; + return max_suffix_rev + 1; +} + +/* Return the first location of non-empty NEEDLE within HAYSTACK, or + NULL. HAYSTACK_LEN is the minimum known length of HAYSTACK. This + method is optimized for NEEDLE_LEN < LONG_NEEDLE_THRESHOLD. + Performance is guaranteed to be linear, with an initialization cost + of 2 * NEEDLE_LEN comparisons. + + If AVAILABLE does not modify HAYSTACK_LEN (as in memmem), then at + most 2 * HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching. + If AVAILABLE modifies HAYSTACK_LEN (as in strstr), then at most 3 * + HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching. */ +static RETURN_TYPE +two_way_short_needle (const unsigned char *haystack, size_t haystack_len, + const unsigned char *needle, size_t needle_len) +{ + size_t i; /* Index into current byte of NEEDLE. */ + size_t j; /* Index into current window of HAYSTACK. */ + size_t period; /* The period of the right half of needle. */ + size_t suffix; /* The index of the right half of needle. */ + + /* Factor the needle into two halves, such that the left half is + smaller than the global period, and the right half is + periodic (with a period as large as NEEDLE_LEN - suffix). */ + suffix = critical_factorization (needle, needle_len, &period); + + /* Perform the search. Each iteration compares the right half + first. */ + if (CMP_FUNC (needle, needle + period, suffix) == 0) + { + /* Entire needle is periodic; a mismatch in the left half can + only advance by the period, so use memory to avoid rescanning + known occurrences of the period in the right half. */ + size_t memory = 0; + j = 0; + while (AVAILABLE (haystack, haystack_len, j, needle_len)) + { + /* Scan for matches in right half. */ + i = MAX (suffix, memory); + while (i < needle_len && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + ++i; + if (needle_len <= i) + { + /* Scan for matches in left half. */ + i = suffix - 1; + while (memory < i + 1 && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + --i; + if (i + 1 < memory + 1) + return (RETURN_TYPE) (haystack + j); + /* No match, so remember how many repetitions of period + on the right half were scanned. */ + j += period; + memory = needle_len - period; + } + else + { + j += i - suffix + 1; + memory = 0; + } + } + } + else + { + /* The two halves of needle are distinct; no extra memory is + required, and any mismatch results in a maximal shift. */ + period = MAX (suffix, needle_len - suffix) + 1; + j = 0; + while (AVAILABLE (haystack, haystack_len, j, needle_len)) + { + /* Scan for matches in right half. */ + i = suffix; + while (i < needle_len && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + ++i; + if (needle_len <= i) + { + /* Scan for matches in left half. */ + i = suffix - 1; + while (i != SIZE_MAX && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + --i; + if (i == SIZE_MAX) + return (RETURN_TYPE) (haystack + j); + j += period; + } + else + j += i - suffix + 1; + } + } + return NULL; +} + +/* Return the first location of non-empty NEEDLE within HAYSTACK, or + NULL. HAYSTACK_LEN is the minimum known length of HAYSTACK. This + method is optimized for LONG_NEEDLE_THRESHOLD <= NEEDLE_LEN. + Performance is guaranteed to be linear, with an initialization cost + of 3 * NEEDLE_LEN + (1 << CHAR_BIT) operations. + + If AVAILABLE does not modify HAYSTACK_LEN (as in memmem), then at + most 2 * HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching, + and sublinear performance O(HAYSTACK_LEN / NEEDLE_LEN) is possible. + If AVAILABLE modifies HAYSTACK_LEN (as in strstr), then at most 3 * + HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching, and + sublinear performance is not possible. */ +static RETURN_TYPE +two_way_long_needle (const unsigned char *haystack, size_t haystack_len, + const unsigned char *needle, size_t needle_len) +{ + size_t i; /* Index into current byte of NEEDLE. */ + size_t j; /* Index into current window of HAYSTACK. */ + size_t period; /* The period of the right half of needle. */ + size_t suffix; /* The index of the right half of needle. */ + size_t shift_table[1U << CHAR_BIT]; /* See below. */ + + /* Factor the needle into two halves, such that the left half is + smaller than the global period, and the right half is + periodic (with a period as large as NEEDLE_LEN - suffix). */ + suffix = critical_factorization (needle, needle_len, &period); + + /* Populate shift_table. For each possible byte value c, + shift_table[c] is the distance from the last occurrence of c to + the end of NEEDLE, or NEEDLE_LEN if c is absent from the NEEDLE. + shift_table[NEEDLE[NEEDLE_LEN - 1]] contains the only 0. */ + for (i = 0; i < 1U << CHAR_BIT; i++) + shift_table[i] = needle_len; + for (i = 0; i < needle_len; i++) + shift_table[CANON_ELEMENT (needle[i])] = needle_len - i - 1; + + /* Perform the search. Each iteration compares the right half + first. */ + if (CMP_FUNC (needle, needle + period, suffix) == 0) + { + /* Entire needle is periodic; a mismatch in the left half can + only advance by the period, so use memory to avoid rescanning + known occurrences of the period in the right half. */ + size_t memory = 0; + size_t shift; + j = 0; + while (AVAILABLE (haystack, haystack_len, j, needle_len)) + { + /* Check the last byte first; if it does not match, then + shift to the next possible match location. */ + shift = shift_table[CANON_ELEMENT (haystack[j + needle_len - 1])]; + if (0 < shift) + { + if (memory && shift < period) + { + /* Since needle is periodic, but the last period has + a byte out of place, there can be no match until + after the mismatch. */ + shift = needle_len - period; + } + memory = 0; + j += shift; + continue; + } + /* Scan for matches in right half. The last byte has + already been matched, by virtue of the shift table. */ + i = MAX (suffix, memory); + while (i < needle_len - 1 && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + ++i; + if (needle_len - 1 <= i) + { + /* Scan for matches in left half. */ + i = suffix - 1; + while (memory < i + 1 && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + --i; + if (i + 1 < memory + 1) + return (RETURN_TYPE) (haystack + j); + /* No match, so remember how many repetitions of period + on the right half were scanned. */ + j += period; + memory = needle_len - period; + } + else + { + j += i - suffix + 1; + memory = 0; + } + } + } + else + { + /* The two halves of needle are distinct; no extra memory is + required, and any mismatch results in a maximal shift. */ + size_t shift; + period = MAX (suffix, needle_len - suffix) + 1; + j = 0; + while (AVAILABLE (haystack, haystack_len, j, needle_len)) + { + /* Check the last byte first; if it does not match, then + shift to the next possible match location. */ + shift = shift_table[CANON_ELEMENT (haystack[j + needle_len - 1])]; + if (0 < shift) + { + j += shift; + continue; + } + /* Scan for matches in right half. The last byte has + already been matched, by virtue of the shift table. */ + i = suffix; + while (i < needle_len - 1 && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + ++i; + if (needle_len - 1 <= i) + { + /* Scan for matches in left half. */ + i = suffix - 1; + while (i != SIZE_MAX && (CANON_ELEMENT (needle[i]) + == CANON_ELEMENT (haystack[i + j]))) + --i; + if (i == SIZE_MAX) + return (RETURN_TYPE) (haystack + j); + j += period; + } + else + j += i - suffix + 1; + } + } + return NULL; +} + +#undef AVAILABLE +#undef CANON_ELEMENT +#undef CMP_FUNC +#undef MAX +#undef RETURN_TYPE diff --git a/lib/strtoimax.c b/lib/strtoimax.c index 0d602e2f5e0..87b080c4f35 100644 --- a/lib/strtoimax.c +++ b/lib/strtoimax.c @@ -1,7 +1,7 @@ /* Convert string representation of a number into an intmax_t value. - Copyright (C) 1999, 2001-2004, 2006, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1999, 2001-2004, 2006, 2009-2019 Free Software Foundation, + Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/strtol.c b/lib/strtol.c index e5dc5659bcd..1d920aff544 100644 --- a/lib/strtol.c +++ b/lib/strtol.c @@ -1,7 +1,7 @@ /* Convert string representation of a number into an integer value. - Copyright (C) 1991-1992, 1994-1999, 2003, 2005-2007, 2009-2019 Free - Software Foundation, Inc. + Copyright (C) 1991-1992, 1994-1999, 2003, 2005-2007, 2009-2019 Free Software + Foundation, Inc. NOTE: The canonical source of this file is maintained with the GNU C Library. Bugs can be reported to bug-glibc@gnu.org. @@ -117,35 +117,6 @@ # define STRTOL_LONG_MIN LLONG_MIN # define STRTOL_LONG_MAX LLONG_MAX # define STRTOL_ULONG_MAX ULLONG_MAX - -/* The extra casts in the following macros work around compiler bugs, - e.g., in Cray C 5.0.3.0. */ - -/* True if the arithmetic type T is signed. */ -# define TYPE_SIGNED(t) (! ((t) 0 < (t) -1)) - -/* Minimum and maximum values for integer types. - These macros have undefined behavior for signed types that either - have padding bits or do not use two's complement. If this is a - problem for you, please let us know how to fix it for your host. */ - -/* The maximum and minimum values for the integer type T. */ -# define TYPE_MINIMUM(t) ((t) ~ TYPE_MAXIMUM (t)) -# define TYPE_MAXIMUM(t) \ - ((t) (! TYPE_SIGNED (t) \ - ? (t) -1 \ - : ((((t) 1 << (sizeof (t) * CHAR_BIT - 2)) - 1) * 2 + 1))) - -# ifndef ULLONG_MAX -# define ULLONG_MAX TYPE_MAXIMUM (unsigned long long) -# endif -# ifndef LLONG_MAX -# define LLONG_MAX TYPE_MAXIMUM (long long int) -# endif -# ifndef LLONG_MIN -# define LLONG_MIN TYPE_MINIMUM (long long int) -# endif - # if __GNUC__ == 2 && __GNUC_MINOR__ < 7 /* Work around gcc bug with using this constant. */ static const unsigned long long int maxquad = ULLONG_MAX; diff --git a/lib/strtoll.c b/lib/strtoll.c index c89554813a5..038362a308b 100644 --- a/lib/strtoll.c +++ b/lib/strtoll.c @@ -1,6 +1,6 @@ /* Function to parse a 'long long int' from text. - Copyright (C) 1995-1997, 1999, 2001, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1995-1997, 1999, 2001, 2009-2019 Free Software Foundation, + Inc. This file is part of the GNU C Library. This program is free software: you can redistribute it and/or modify diff --git a/lib/sys_stat.in.h b/lib/sys_stat.in.h index b25a6d46b25..9ddd1a8d004 100644 --- a/lib/sys_stat.in.h +++ b/lib/sys_stat.in.h @@ -54,16 +54,23 @@ /* The definition of _GL_WARN_ON_USE is copied here. */ +/* Before doing "#define mknod rpl_mknod" below, we need to include all + headers that may declare mknod(). OS/2 kLIBC declares mknod() in + <unistd.h>, not in <sys/stat.h>. */ +#ifdef __KLIBC__ +# include <unistd.h> +#endif + /* Before doing "#define mkdir rpl_mkdir" below, we need to include all headers that may declare mkdir(). Native Windows platforms declare mkdir - in <io.h> and/or <direct.h>, not in <unistd.h>. */ -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ + in <io.h> and/or <direct.h>, not in <sys/stat.h>. */ +#if defined _WIN32 && ! defined __CYGWIN__ # include <io.h> /* mingw32, mingw64 */ # include <direct.h> /* mingw64, MSVC 9 */ #endif /* Native Windows platforms declare umask() in <io.h>. */ -#if 0 && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) +#if 0 && (defined _WIN32 && ! defined __CYGWIN__) # include <io.h> #endif @@ -576,7 +583,7 @@ _GL_CXXALIAS_RPL (mkdir, int, (char const *name, mode_t mode)); Additionally, it declares _mkdir (and depending on compile flags, an alias mkdir), only in the nonstandard includes <direct.h> and <io.h>, which are included above. */ -# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ +# if defined _WIN32 && ! defined __CYGWIN__ # if !GNULIB_defined_rpl_mkdir static int diff --git a/lib/sys_types.in.h b/lib/sys_types.in.h index 0e85241714a..237e2068cf8 100644 --- a/lib/sys_types.in.h +++ b/lib/sys_types.in.h @@ -20,6 +20,17 @@ #endif @PRAGMA_COLUMNS@ +#if defined _WIN32 && !defined __CYGWIN__ \ + && (defined __need_off_t || defined __need___off64_t \ + || defined __need_ssize_t || defined __need_time_t) + +/* Special invocation convention inside mingw header files. */ + +#@INCLUDE_NEXT@ @NEXT_SYS_TYPES_H@ + +#else +/* Normal invocation convention. */ + #ifndef _@GUARD_PREFIX@_SYS_TYPES_H /* The include_next requires a split double-inclusion guard. */ @@ -86,10 +97,10 @@ typedef unsigned long long int rpl_ino_t; /* MSVC 9 defines size_t in <stddef.h>, not in <sys/types.h>. */ /* But avoid namespace pollution on glibc systems. */ -#if ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) \ - && ! defined __GLIBC__ +#if (defined _WIN32 && ! defined __CYGWIN__) && ! defined __GLIBC__ # include <stddef.h> #endif #endif /* _@GUARD_PREFIX@_SYS_TYPES_H */ #endif /* _@GUARD_PREFIX@_SYS_TYPES_H */ +#endif /* __need_XXX */ diff --git a/lib/tempname.c b/lib/tempname.c index 43aaef86d19..be62ed9513c 100644 --- a/lib/tempname.c +++ b/lib/tempname.c @@ -1,7 +1,6 @@ /* tempname.c - generate the name of a temporary file. - Copyright (C) 1991-2003, 2005-2007, 2009-2019 Free Software - Foundation, Inc. + Copyright (C) 1991-2003, 2005-2007, 2009-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/time.in.h b/lib/time.in.h index f9e422ffcad..dd3b21273c9 100644 --- a/lib/time.in.h +++ b/lib/time.in.h @@ -48,7 +48,7 @@ /* The definition of _GL_WARN_ON_USE is copied here. */ -/* Some systems don't define struct timespec (e.g., AIX 4.1, Ultrix 4.3). +/* Some systems don't define struct timespec (e.g., AIX 4.1). Or they define it with the wrong member names or define it in <sys/time.h> (e.g., FreeBSD circa 1997). Stock Mingw prior to 3.0 does not define it, but the pthreads-win32 library defines it in <pthread.h>. */ @@ -212,7 +212,7 @@ _GL_CXXALIASWARN (gmtime_r); # define localtime rpl_localtime # endif _GL_FUNCDECL_RPL (localtime, struct tm *, (time_t const *__timer) - _GL_ARG_NONNULL ((1))); + _GL_ARG_NONNULL ((1))); _GL_CXXALIAS_RPL (localtime, struct tm *, (time_t const *__timer)); # else _GL_CXXALIAS_SYS (localtime, struct tm *, (time_t const *__timer)); diff --git a/lib/time_r.c b/lib/time_r.c index abec93020d3..a701ccc5b82 100644 --- a/lib/time_r.c +++ b/lib/time_r.c @@ -1,7 +1,6 @@ /* Reentrant time functions like localtime_r. - Copyright (C) 2003, 2006-2007, 2010-2019 Free Software Foundation, - Inc. + Copyright (C) 2003, 2006-2007, 2010-2019 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by diff --git a/lib/time_rz.c b/lib/time_rz.c index a1dd1cf2481..42ae3d3649f 100644 --- a/lib/time_rz.c +++ b/lib/time_rz.c @@ -286,6 +286,21 @@ revert_tz (timezone_t tz) struct tm * localtime_rz (timezone_t tz, time_t const *t, struct tm *tm) { +#ifdef HAVE_LOCALTIME_INFLOOP_BUG + /* The -67768038400665599 comes from: + https://lists.gnu.org/r/bug-gnulib/2017-07/msg00142.html + On affected platforms the greatest POSIX-compatible time_t value + that could return nonnull is 67768036191766798 (when + TZ="XXX24:59:59" it resolves to the year 2**31 - 1 + 1900, on + 12-31 at 23:59:59), so test for that too while we're in the + neighborhood. */ + if (! (-67768038400665599 <= *t && *t <= 67768036191766798)) + { + errno = EOVERFLOW; + return NULL; + } +#endif + if (!tz) return gmtime_r (t, tm); else diff --git a/lib/timegm.c b/lib/timegm.c index 8532a00342b..c440480cb2d 100644 --- a/lib/timegm.c +++ b/lib/timegm.c @@ -1,40 +1,58 @@ /* Convert UTC calendar time to simple time. Like mktime but assumes UTC. - Copyright (C) 1994, 1997, 2003-2004, 2006-2007, 2009-2019 Free - Software Foundation, Inc. This file is part of the GNU C Library. + Copyright (C) 1994-2019 Free Software Foundation, Inc. + This file is part of the GNU C Library. - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 3, or (at your option) - any later version. + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU General Public + License as published by the Free Software Foundation; either + version 3 of the License, or (at your option) any later version. - This program is distributed in the hope that it will be useful, + The GNU C Library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. - You should have received a copy of the GNU General Public License - along with this program; if not, see <https://www.gnu.org/licenses/>. */ + You should have received a copy of the GNU General Public + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ #ifndef _LIBC -# include <config.h> +# include <libc-config.h> #endif #include <time.h> +#include <errno.h> -#ifdef _LIBC -typedef time_t mktime_offset_t; -#else -# undef __gmtime_r -# define __gmtime_r gmtime_r -# define __mktime_internal mktime_internal -# include "mktime-internal.h" -#endif +#include "mktime-internal.h" -time_t -timegm (struct tm *tmp) +__time64_t +__timegm64 (struct tm *tmp) { static mktime_offset_t gmtime_offset; tmp->tm_isdst = 0; - return __mktime_internal (tmp, __gmtime_r, &gmtime_offset); + return __mktime_internal (tmp, __gmtime64_r, &gmtime_offset); +} + +#if defined _LIBC && __TIMESIZE != 64 + +libc_hidden_def (__timegm64) + +time_t +timegm (struct tm *tmp) +{ + struct tm tm = *tmp; + __time64_t t = __timegm64 (&tm); + if (in_time_t_range (t)) + { + *tmp = tm; + return t; + } + else + { + __set_errno (EOVERFLOW); + return -1; + } } + +#endif diff --git a/lib/timespec-add.c b/lib/timespec-add.c index ded923a355b..e0a9f12e14c 100644 --- a/lib/timespec-add.c +++ b/lib/timespec-add.c @@ -18,7 +18,7 @@ /* Written by Paul Eggert. */ /* Return the sum of two timespec values A and B. On overflow, return - an extremal value. This assumes 0 <= tv_nsec < TIMESPEC_RESOLUTION. */ + an extremal value. This assumes 0 <= tv_nsec < TIMESPEC_HZ. */ #include <config.h> #include "timespec.h" @@ -31,7 +31,7 @@ timespec_add (struct timespec a, struct timespec b) time_t rs = a.tv_sec; time_t bs = b.tv_sec; int ns = a.tv_nsec + b.tv_nsec; - int nsd = ns - TIMESPEC_RESOLUTION; + int nsd = ns - TIMESPEC_HZ; int rns = ns; time_t tmin = TYPE_MINIMUM (time_t); time_t tmax = TYPE_MAXIMUM (time_t); @@ -63,7 +63,7 @@ timespec_add (struct timespec a, struct timespec b) { high_overflow: rs = tmax; - rns = TIMESPEC_RESOLUTION - 1; + rns = TIMESPEC_HZ - 1; } } diff --git a/lib/timespec-sub.c b/lib/timespec-sub.c index 664e6340a7b..48434e81506 100644 --- a/lib/timespec-sub.c +++ b/lib/timespec-sub.c @@ -19,7 +19,7 @@ /* Return the difference between two timespec values A and B. On overflow, return an extremal value. This assumes 0 <= tv_nsec < - TIMESPEC_RESOLUTION. */ + TIMESPEC_HZ. */ #include <config.h> #include "timespec.h" @@ -38,7 +38,7 @@ timespec_sub (struct timespec a, struct timespec b) if (ns < 0) { - rns = ns + TIMESPEC_RESOLUTION; + rns = ns + TIMESPEC_HZ; if (bs < tmax) bs++; else if (- TYPE_SIGNED (time_t) < rs) @@ -63,7 +63,7 @@ timespec_sub (struct timespec a, struct timespec b) else { rs = tmax; - rns = TIMESPEC_RESOLUTION - 1; + rns = TIMESPEC_HZ - 1; } } diff --git a/lib/timespec.h b/lib/timespec.h index 8d4232a98c2..26f1bc1a4c7 100644 --- a/lib/timespec.h +++ b/lib/timespec.h @@ -17,9 +17,9 @@ along with this program. If not, see <https://www.gnu.org/licenses/>. */ #if ! defined TIMESPEC_H -# define TIMESPEC_H +#define TIMESPEC_H -# include <time.h> +#include <time.h> #ifndef _GL_INLINE_HEADER_BEGIN #error "Please include config.h first." @@ -33,13 +33,20 @@ _GL_INLINE_HEADER_BEGIN extern "C" { #endif +#include "arg-nonnull.h" #include "verify.h" -/* Resolution of timespec timestamps (in units per second), and log - base 10 of the resolution. */ +/* Inverse resolution of timespec timestamps (in units per second), + and log base 10 of the inverse resolution. */ -enum { TIMESPEC_RESOLUTION = 1000000000 }; -enum { LOG10_TIMESPEC_RESOLUTION = 9 }; +enum { TIMESPEC_HZ = 1000000000 }; +enum { LOG10_TIMESPEC_HZ = 9 }; + +/* Obsolescent names for backward compatibility. + They are misnomers, because TIMESPEC_RESOLUTION is not a resolution. */ + +enum { TIMESPEC_RESOLUTION = TIMESPEC_HZ }; +enum { LOG10_TIMESPEC_RESOLUTION = LOG10_TIMESPEC_HZ }; /* Return a timespec with seconds S and nanoseconds NS. */ @@ -87,9 +94,9 @@ timespec_cmp (struct timespec a, struct timespec b) return 1; /* Pacify gcc -Wstrict-overflow (bleeding-edge circa 2017-10-02). See: - http://lists.gnu.org/r/bug-gnulib/2017-10/msg00006.html */ - assume (-1 <= a.tv_nsec && a.tv_nsec <= 2 * TIMESPEC_RESOLUTION); - assume (-1 <= b.tv_nsec && b.tv_nsec <= 2 * TIMESPEC_RESOLUTION); + https://lists.gnu.org/r/bug-gnulib/2017-10/msg00006.html */ + assume (-1 <= a.tv_nsec && a.tv_nsec <= 2 * TIMESPEC_HZ); + assume (-1 <= b.tv_nsec && b.tv_nsec <= 2 * TIMESPEC_HZ); return a.tv_nsec - b.tv_nsec; } @@ -116,8 +123,9 @@ timespectod (struct timespec a) return a.tv_sec + a.tv_nsec / 1e9; } -void gettime (struct timespec *); -int settime (struct timespec const *); +struct timespec current_timespec (void); +void gettime (struct timespec *) _GL_ARG_NONNULL ((1)); +int settime (struct timespec const *) _GL_ARG_NONNULL ((1)); #ifdef __cplusplus } diff --git a/lib/unistd.in.h b/lib/unistd.in.h index d83553805f2..7778d25dc7e 100644 --- a/lib/unistd.in.h +++ b/lib/unistd.in.h @@ -61,16 +61,18 @@ /* But avoid namespace pollution on glibc systems. */ #if (!(defined SEEK_CUR && defined SEEK_END && defined SEEK_SET) \ || ((@GNULIB_UNLINK@ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__)) \ + && (defined _WIN32 && ! defined __CYGWIN__)) \ || ((@GNULIB_SYMLINKAT@ || defined GNULIB_POSIXCHECK) \ && defined __CYGWIN__)) \ && ! defined __GLIBC__ # include <stdio.h> #endif -/* Cygwin 1.7.1 declares unlinkat in <fcntl.h>, not in <unistd.h>. */ +/* Cygwin 1.7.1 and Android 4.3 declare unlinkat in <fcntl.h>, not in + <unistd.h>. */ /* But avoid namespace pollution on glibc systems. */ -#if (@GNULIB_UNLINKAT@ || defined GNULIB_POSIXCHECK) && defined __CYGWIN__ \ +#if (@GNULIB_UNLINKAT@ || defined GNULIB_POSIXCHECK) \ + && (defined __CYGWIN__ || defined __ANDROID__) \ && ! defined __GLIBC__ # include <fcntl.h> #endif @@ -94,13 +96,13 @@ lseek(), read(), unlink(), write() in <io.h>. */ #if ((@GNULIB_CHDIR@ || @GNULIB_GETCWD@ || @GNULIB_RMDIR@ \ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__)) + && (defined _WIN32 && ! defined __CYGWIN__)) # include <io.h> /* mingw32, mingw64 */ # include <direct.h> /* mingw64, MSVC 9 */ #elif (@GNULIB_CLOSE@ || @GNULIB_DUP@ || @GNULIB_DUP2@ || @GNULIB_ISATTY@ \ || @GNULIB_LSEEK@ || @GNULIB_READ@ || @GNULIB_UNLINK@ || @GNULIB_WRITE@ \ || defined GNULIB_POSIXCHECK) \ - && ((defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__) + && (defined _WIN32 && ! defined __CYGWIN__) # include <io.h> #endif @@ -113,6 +115,13 @@ # include <netdb.h> #endif +/* Android 4.3 declares fchownat in <sys/stat.h>, not in <unistd.h>. */ +/* But avoid namespace pollution on glibc systems. */ +#if (@GNULIB_FCHOWNAT@ || defined GNULIB_POSIXCHECK) && defined __ANDROID__ \ + && !defined __GLIBC__ +# include <sys/stat.h> +#endif + /* MSVC defines off_t in <sys/types.h>. May also define off_t to a 64-bit type on native Windows. */ #if !@HAVE_UNISTD_H@ || @WINDOWS_64_BIT_OFF_T@ @@ -400,6 +409,13 @@ _GL_WARN_ON_USE (dup3, "dup3 is unportable - " #if @GNULIB_ENVIRON@ +# if defined __CYGWIN__ && !defined __i386__ +/* The 'environ' variable is defined in a DLL. Therefore its declaration needs + the '__declspec(dllimport)' attribute, but the system's <unistd.h> lacks it. + This leads to a link error on 64-bit Cygwin when the option + -Wl,--disable-auto-import is in use. */ +_GL_EXTERN_C __declspec(dllimport) char **environ; +# endif # if !@HAVE_DECL_ENVIRON@ /* Set of environment variables and values. An array of strings of the form "VARIABLE=VALUE", terminated with a NULL. */ @@ -425,12 +441,12 @@ extern char **environ; #elif defined GNULIB_POSIXCHECK # if HAVE_RAW_DECL_ENVIRON _GL_UNISTD_INLINE char *** +_GL_WARN_ON_USE_ATTRIBUTE ("environ is unportable - " + "use gnulib module environ for portability") rpl_environ (void) { return &environ; } -_GL_WARN_ON_USE (rpl_environ, "environ is unportable - " - "use gnulib module environ for portability"); # undef environ # define environ (*rpl_environ ()) # endif @@ -928,6 +944,36 @@ _GL_WARN_ON_USE (getpagesize, "getpagesize is unportable - " #endif +#if @GNULIB_GETPASS@ +/* Function getpass() from module 'getpass': + Read a password from /dev/tty or stdin. + Function getpass() from module 'getpass-gnu': + Read a password of arbitrary length from /dev/tty or stdin. */ +# if @REPLACE_GETPASS@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef getpass +# define getpass rpl_getpass +# endif +_GL_FUNCDECL_RPL (getpass, char *, (const char *prompt) + _GL_ARG_NONNULL ((1))); +_GL_CXXALIAS_RPL (getpass, char *, (const char *prompt)); +# else +# if !@HAVE_GETPASS@ +_GL_FUNCDECL_SYS (getpass, char *, (const char *prompt) + _GL_ARG_NONNULL ((1))); +# endif +_GL_CXXALIAS_SYS (getpass, char *, (const char *prompt)); +# endif +_GL_CXXALIASWARN (getpass); +#elif defined GNULIB_POSIXCHECK +# undef getpass +# if HAVE_RAW_DECL_GETPASS +_GL_WARN_ON_USE (getpass, "getpass is unportable - " + "use gnulib module getpass or getpass-gnu for portability"); +# endif +#endif + + #if @GNULIB_GETUSERSHELL@ /* Return the next valid login shell on the system, or NULL when the end of the list has been reached. */ @@ -1482,7 +1528,7 @@ _GL_FUNCDECL_RPL (truncate, int, (const char *filename, off_t length) _GL_ARG_NONNULL ((1))); _GL_CXXALIAS_RPL (truncate, int, (const char *filename, off_t length)); # else -# if !@HAVE_TRUNCATE@ +# if !@HAVE_DECL_TRUNCATE@ _GL_FUNCDECL_SYS (truncate, int, (const char *filename, off_t length) _GL_ARG_NONNULL ((1))); # endif diff --git a/lib/utimens.c b/lib/utimens.c index c7599afdb88..c9b65ef4c20 100644 --- a/lib/utimens.c +++ b/lib/utimens.c @@ -39,8 +39,7 @@ GNU Emacs, which arranges for this in some other way and which defines WIN32_LEAN_AND_MEAN itself. */ -#if ((defined _WIN32 || defined __WIN32__) \ - && ! defined __CYGWIN__ && ! defined EMACS_CONFIGURATION) +#if defined _WIN32 && ! defined __CYGWIN__ && ! defined EMACS_CONFIGURATION # define USE_SETFILETIME # define WIN32_LEAN_AND_MEAN # include <windows.h> @@ -92,11 +91,11 @@ validate_timespec (struct timespec timespec[2]) if ((timespec[0].tv_nsec != UTIME_NOW && timespec[0].tv_nsec != UTIME_OMIT && ! (0 <= timespec[0].tv_nsec - && timespec[0].tv_nsec < TIMESPEC_RESOLUTION)) + && timespec[0].tv_nsec < TIMESPEC_HZ)) || (timespec[1].tv_nsec != UTIME_NOW && timespec[1].tv_nsec != UTIME_OMIT && ! (0 <= timespec[1].tv_nsec - && timespec[1].tv_nsec < TIMESPEC_RESOLUTION))) + && timespec[1].tv_nsec < TIMESPEC_HZ))) { errno = EINVAL; return -1; @@ -289,8 +288,8 @@ fdutimens (int fd, char const *file, struct timespec const timespec[2]) #ifdef USE_SETFILETIME /* On native Windows, use SetFileTime(). See - <https://msdn.microsoft.com/en-us/library/ms724933.aspx> - <https://msdn.microsoft.com/en-us/library/ms724284.aspx> */ + <https://docs.microsoft.com/en-us/windows/desktop/api/fileapi/nf-fileapi-setfiletime> + <https://docs.microsoft.com/en-us/windows/desktop/api/minwinbase/ns-minwinbase-filetime> */ if (0 <= fd) { HANDLE handle; @@ -308,10 +307,10 @@ fdutimens (int fd, char const *file, struct timespec const timespec[2]) if (ts == NULL || ts[0].tv_nsec == UTIME_NOW || ts[1].tv_nsec == UTIME_NOW) { /* GetSystemTimeAsFileTime - <https://msdn.microsoft.com/en-us/library/ms724397.aspx>. + <https://docs.microsoft.com/en-us/windows/desktop/api/sysinfoapi/nf-sysinfoapi-getsystemtimeasfiletime>. It would be overkill to use GetSystemTimePreciseAsFileTime - <https://msdn.microsoft.com/en-us/library/hh706895.aspx>. */ + <https://docs.microsoft.com/en-us/windows/desktop/api/sysinfoapi/nf-sysinfoapi-getsystemtimepreciseasfiletime>. */ GetSystemTimeAsFileTime (¤t_time); } diff --git a/lib/verify.h b/lib/verify.h index 4cd2f4af8a3..f8e4eff026d 100644 --- a/lib/verify.h +++ b/lib/verify.h @@ -21,31 +21,37 @@ #define _GL_VERIFY_H -/* Define _GL_HAVE__STATIC_ASSERT to 1 if _Static_assert works as per C11. - This is supported by GCC 4.6.0 and later, in C mode, and its use - here generates easier-to-read diagnostics when verify (R) fails. - - Define _GL_HAVE_STATIC_ASSERT to 1 if static_assert works as per C++11. - This will likely be supported by future GCC versions, in C++ mode. - - Use this only with GCC. If we were willing to slow 'configure' - down we could also use it with other compilers, but since this - affects only the quality of diagnostics, why bother? */ -#if (4 < __GNUC__ + (6 <= __GNUC_MINOR__) \ - && (201112L <= __STDC_VERSION__ || !defined __STRICT_ANSI__) \ - && !defined __cplusplus) -# define _GL_HAVE__STATIC_ASSERT 1 -#endif -/* The condition (99 < __GNUC__) is temporary, until we know about the - first G++ release that supports static_assert. */ -#if (99 < __GNUC__) && defined __cplusplus -# define _GL_HAVE_STATIC_ASSERT 1 +/* Define _GL_HAVE__STATIC_ASSERT to 1 if _Static_assert (R, DIAGNOSTIC) + works as per C11. This is supported by GCC 4.6.0 and later, in C + mode. + + Define _GL_HAVE__STATIC_ASSERT1 to 1 if _Static_assert (R) works as + per C2X, and define _GL_HAVE_STATIC_ASSERT1 if static_assert (R) + works as per C++17. This is supported by GCC 9.1 and later. + + Support compilers claiming conformance to the relevant standard, + and also support GCC when not pedantic. If we were willing to slow + 'configure' down we could also use it with other compilers, but + since this affects only the quality of diagnostics, why bother? */ +#ifndef __cplusplus +# if (201112L <= __STDC_VERSION__ \ + || (!defined __STRICT_ANSI__ && 4 < __GNUC__ + (6 <= __GNUC_MINOR__))) +# define _GL_HAVE__STATIC_ASSERT 1 +# endif +# if (202000L <= __STDC_VERSION__ \ + || (!defined __STRICT_ANSI__ && 9 <= __GNUC__)) +# define _GL_HAVE__STATIC_ASSERT1 1 +# endif +#else +# if 201703L <= __cplusplus || 9 <= __GNUC__ +# define _GL_HAVE_STATIC_ASSERT1 1 +# endif #endif /* FreeBSD 9.1 <sys/cdefs.h>, included by <stddef.h> and lots of other system headers, defines a conflicting _Static_assert that is no better than ours; override it. */ -#ifndef _GL_HAVE_STATIC_ASSERT +#ifndef _GL_HAVE__STATIC_ASSERT # include <stddef.h> # undef _Static_assert #endif @@ -143,9 +149,9 @@ which do not support _Static_assert, also do not warn about the last declaration mentioned above. - * GCC warns if -Wnested-externs is enabled and verify() is used + * GCC warns if -Wnested-externs is enabled and 'verify' is used within a function body; but inside a function, you can always - arrange to use verify_expr() instead. + arrange to use verify_expr instead. * In C++, any struct definition inside sizeof is invalid. Use a template type to work around the problem. */ @@ -169,11 +175,9 @@ #define _GL_GENSYM(prefix) _GL_CONCAT (prefix, _GL_COUNTER) /* Verify requirement R at compile-time, as an integer constant expression - that returns 1. If R is false, fail at compile-time, preferably - with a diagnostic that includes the string-literal DIAGNOSTIC. */ + that returns 1. If R is false, fail at compile-time. */ -#define _GL_VERIFY_TRUE(R, DIAGNOSTIC) \ - (!!sizeof (_GL_VERIFY_TYPE (R, DIAGNOSTIC))) +#define _GL_VERIFY_TRUE(R) (!!sizeof (_GL_VERIFY_TYPE (R))) #ifdef __cplusplus # if !GNULIB_defined_struct__gl_verify_type @@ -183,40 +187,43 @@ template <int w> }; # define GNULIB_defined_struct__gl_verify_type 1 # endif -# define _GL_VERIFY_TYPE(R, DIAGNOSTIC) \ - _gl_verify_type<(R) ? 1 : -1> -#elif defined _GL_HAVE__STATIC_ASSERT -# define _GL_VERIFY_TYPE(R, DIAGNOSTIC) \ +# define _GL_VERIFY_TYPE(R) _gl_verify_type<(R) ? 1 : -1> +#elif defined _GL_HAVE__STATIC_ASSERT1 +# define _GL_VERIFY_TYPE(R) \ struct { \ - _Static_assert (R, DIAGNOSTIC); \ + _Static_assert (R); \ int _gl_dummy; \ } #else -# define _GL_VERIFY_TYPE(R, DIAGNOSTIC) \ +# define _GL_VERIFY_TYPE(R) \ struct { unsigned int _gl_verify_error_if_negative: (R) ? 1 : -1; } #endif /* Verify requirement R at compile-time, as a declaration without a - trailing ';'. If R is false, fail at compile-time, preferably - with a diagnostic that includes the string-literal DIAGNOSTIC. + trailing ';'. If R is false, fail at compile-time. + + This macro requires three or more arguments but uses at most the first + two, so that the _Static_assert macro optionally defined below supports + both the C11 two-argument syntax and the C2X one-argument syntax. Unfortunately, unlike C11, this implementation must appear as an ordinary declaration, and cannot appear inside struct { ... }. */ -#ifdef _GL_HAVE__STATIC_ASSERT -# define _GL_VERIFY _Static_assert +#if defined _GL_HAVE__STATIC_ASSERT +# define _GL_VERIFY(R, DIAGNOSTIC, ...) _Static_assert (R, DIAGNOSTIC) #else -# define _GL_VERIFY(R, DIAGNOSTIC) \ +# define _GL_VERIFY(R, DIAGNOSTIC, ...) \ extern int (*_GL_GENSYM (_gl_verify_function) (void)) \ - [_GL_VERIFY_TRUE (R, DIAGNOSTIC)] + [_GL_VERIFY_TRUE (R)] #endif /* _GL_STATIC_ASSERT_H is defined if this code is copied into assert.h. */ #ifdef _GL_STATIC_ASSERT_H -# if !defined _GL_HAVE__STATIC_ASSERT && !defined _Static_assert -# define _Static_assert(R, DIAGNOSTIC) _GL_VERIFY (R, DIAGNOSTIC) +# if !defined _GL_HAVE__STATIC_ASSERT1 && !defined _Static_assert +# define _Static_assert(...) \ + _GL_VERIFY (__VA_ARGS__, "static assertion failed", -) # endif -# if !defined _GL_HAVE_STATIC_ASSERT && !defined static_assert +# if !defined _GL_HAVE_STATIC_ASSERT1 && !defined static_assert # define static_assert _Static_assert /* C11 requires this #define. */ # endif #endif @@ -228,31 +235,24 @@ template <int w> assert (R), there is no run-time overhead. There are two macros, since no single macro can be used in all - contexts in C. verify_true (R) is for scalar contexts, including + contexts in C. verify_expr (R, E) is for scalar contexts, including integer constant expression contexts. verify (R) is for declaration contexts, e.g., the top level. */ -/* Verify requirement R at compile-time, as an integer constant expression. - Return 1. This is equivalent to verify_expr (R, 1). - - verify_true is obsolescent; please use verify_expr instead. */ - -#define verify_true(R) _GL_VERIFY_TRUE (R, "verify_true (" #R ")") - /* Verify requirement R at compile-time. Return the value of the expression E. */ -#define verify_expr(R, E) \ - (_GL_VERIFY_TRUE (R, "verify_expr (" #R ", " #E ")") ? (E) : (E)) +#define verify_expr(R, E) (_GL_VERIFY_TRUE (R) ? (E) : (E)) /* Verify requirement R at compile-time, as a declaration without a - trailing ';'. */ + trailing ';'. verify (R) acts like static_assert (R) except that + it is portable to C11/C++14 and earlier, and its name is shorter + and may be more convenient. */ -#ifdef __GNUC__ -# define verify(R) _GL_VERIFY (R, "verify (" #R ")") +#ifdef _GL_HAVE__STATIC_ASSERT1 +# define verify(R) _Static_assert (R) #else -/* PGI barfs if R is long. Play it safe. */ -# define verify(R) _GL_VERIFY (R, "verify (...)") +# define verify(R) _GL_VERIFY (R, "verify (...)", -) #endif #ifndef __has_builtin @@ -276,7 +276,8 @@ template <int w> when 'assume' silences warnings even with older GCCs. */ # define assume(R) ((R) ? (void) 0 : __builtin_trap ()) #else -# define assume(R) ((void) (0 && (R))) + /* Some tools grok NOTREACHED, e.g., Oracle Studio 12.6. */ +# define assume(R) ((R) ? (void) 0 : /*NOTREACHED*/ (void) 0) #endif /* @assert.h omit end@ */ diff --git a/lib/vla.h b/lib/vla.h index f6ebba0ede3..8f5dea76f61 100644 --- a/lib/vla.h +++ b/lib/vla.h @@ -17,6 +17,20 @@ Written by Paul Eggert. */ +/* The VLA_ELEMS macro does not allocate variable-length arrays (VLAs), + so it does not have the security or performance issues commonly + associated with VLAs. VLA_ELEMS is for exploiting a C11 feature + where a function can start like this: + + double scan_array (int n, double v[static n]) + + to require a caller to pass a vector V with at least N elements; + this allows better static checking and performance in some cases. + In C11 this feature means that V is a VLA, so the feature is + supported only if __STDC_NO_VLA__ is defined, and for compatibility + to platforms that do not support VLAs, VLA_ELEMS (n) expands to + nothing when __STDC_NO_VLA__ is not defined. */ + /* A function's argument must point to an array with at least N elements. Example: 'int main (int argc, char *argv[VLA_ELEMS (argc)]);'. */ @@ -25,3 +39,15 @@ #else # define VLA_ELEMS(n) static n #endif + +/* Although C99 requires support for variable-length arrays (VLAs), + some C compilers never supported VLAs and VLAs are optional in C11. + VLAs are controversial because their allocation may be unintended + or awkward to support, and large VLAs might cause security or + performance problems. GCC can diagnose the use of VLAs via the + -Wvla and -Wvla-larger-than warnings options, and defining the + macro GNULIB_NO_VLA disables the allocation of VLAs in Gnulib code. + + The VLA_ELEMS macro is unaffected by GNULIB_NO_VLA, since it does + not allocate VLAs. Programs that use VLA_ELEMS should be compiled + with 'gcc -Wvla-larger-than' instead of with 'gcc -Wvla'. */ diff --git a/lib/warn-on-use.h b/lib/warn-on-use.h index 9f17e5bfd4b..7d11a156911 100644 --- a/lib/warn-on-use.h +++ b/lib/warn-on-use.h @@ -20,23 +20,32 @@ supported by the compiler. If the compiler does not support this feature, the macro expands to an unused extern declaration. - This macro is useful for marking a function as a potential + _GL_WARN_ON_USE_ATTRIBUTE ("literal string") expands to the + attribute used in _GL_WARN_ON_USE. If the compiler does not support + this feature, it expands to empty. + + These macros are useful for marking a function as a potential portability trap, with the intent that "literal string" include instructions on the replacement function that should be used - instead. However, one of the reasons that a function is a - portability trap is if it has the wrong signature. Declaring - FUNCTION with a different signature in C is a compilation error, so - this macro must use the same type as any existing declaration so - that programs that avoid the problematic FUNCTION do not fail to - compile merely because they included a header that poisoned the - function. But this implies that _GL_WARN_ON_USE is only safe to - use if FUNCTION is known to already have a declaration. Use of - this macro implies that there must not be any other macro hiding - the declaration of FUNCTION; but undefining FUNCTION first is part - of the poisoning process anyway (although for symbols that are - provided only via a macro, the result is a compilation error rather - than a warning containing "literal string"). Also note that in - C++, it is only safe to use if FUNCTION has no overloads. + instead. + _GL_WARN_ON_USE is for functions with 'extern' linkage. + _GL_WARN_ON_USE_ATTRIBUTE is for functions with 'static' or 'inline' + linkage. + + However, one of the reasons that a function is a portability trap is + if it has the wrong signature. Declaring FUNCTION with a different + signature in C is a compilation error, so this macro must use the + same type as any existing declaration so that programs that avoid + the problematic FUNCTION do not fail to compile merely because they + included a header that poisoned the function. But this implies that + _GL_WARN_ON_USE is only safe to use if FUNCTION is known to already + have a declaration. Use of this macro implies that there must not + be any other macro hiding the declaration of FUNCTION; but + undefining FUNCTION first is part of the poisoning process anyway + (although for symbols that are provided only via a macro, the result + is a compilation error rather than a warning containing + "literal string"). Also note that in C++, it is only safe to use if + FUNCTION has no overloads. For an example, it is possible to poison 'getline' by: - adding a call to gl_WARN_ON_USE_PREPARE([[#include <stdio.h>]], @@ -54,12 +63,21 @@ (less common usage, like &environ, will cause a compilation error rather than issue the nice warning, but the end result of informing the developer about their portability problem is still achieved): - #if HAVE_RAW_DECL_ENVIRON - static char ***rpl_environ (void) { return &environ; } - _GL_WARN_ON_USE (rpl_environ, "environ is not always properly declared"); - # undef environ - # define environ (*rpl_environ ()) - #endif + #if HAVE_RAW_DECL_ENVIRON + static char *** + rpl_environ (void) { return &environ; } + _GL_WARN_ON_USE (rpl_environ, "environ is not always properly declared"); + # undef environ + # define environ (*rpl_environ ()) + #endif + or better (avoiding contradictory use of 'static' and 'extern'): + #if HAVE_RAW_DECL_ENVIRON + static char *** + _GL_WARN_ON_USE_ATTRIBUTE ("environ is not always properly declared") + rpl_environ (void) { return &environ; } + # undef environ + # define environ (*rpl_environ ()) + #endif */ #ifndef _GL_WARN_ON_USE @@ -67,13 +85,17 @@ /* A compiler attribute is available in gcc versions 4.3.0 and later. */ # define _GL_WARN_ON_USE(function, message) \ extern __typeof__ (function) function __attribute__ ((__warning__ (message))) +# define _GL_WARN_ON_USE_ATTRIBUTE(message) \ + __attribute__ ((__warning__ (message))) # elif __GNUC__ >= 3 && GNULIB_STRICT_CHECKING /* Verify the existence of the function. */ # define _GL_WARN_ON_USE(function, message) \ extern __typeof__ (function) function +# define _GL_WARN_ON_USE_ATTRIBUTE(message) # else /* Unsupported. */ # define _GL_WARN_ON_USE(function, message) \ _GL_WARN_EXTERN_C int _gl_warn_on_use +# define _GL_WARN_ON_USE_ATTRIBUTE(message) # endif #endif diff --git a/lib/xalloc-oversized.h b/lib/xalloc-oversized.h index 1bbde4e6eed..e3068c83c48 100644 --- a/lib/xalloc-oversized.h +++ b/lib/xalloc-oversized.h @@ -1,7 +1,6 @@ /* xalloc-oversized.h -- memory allocation size checking - Copyright (C) 1990-2000, 2003-2004, 2006-2019 Free Software - Foundation, Inc. + Copyright (C) 1990-2000, 2003-2004, 2006-2019 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by |