diff options
author | Roland McGrath <roland@gnu.org> | 1995-09-28 18:42:29 +0000 |
---|---|---|
committer | Roland McGrath <roland@gnu.org> | 1995-09-28 18:42:29 +0000 |
commit | 24906b43b9bd9108aef17d18dfbb8764212085f7 (patch) | |
tree | 48d641dc176c9941b3b036e66c383b00a6974bad /intl | |
parent | 91f62ce6b5797f04006ac5aff08416ecd61bf972 (diff) | |
download | glibc-24906b43b9bd9108aef17d18dfbb8764212085f7.tar.gz |
Thu Sep 28 13:05:54 1995 Roland McGrath <roland@churchy.gnu.ai.mit.edu>
Merge new message handling code from GNU gettext, by Drepper.
* intl: New directory.
* Makefile (subdirs): Add intl.
* sysdeps/generic/dl-sysdep.c (_dl_sysdep_start): Return
USER_ENTRY instead of storing it on our stack.
* elf/rtld.c (rtld_command): Variable removed.
(_dl_skip_args): New variable.
(dl_main): Increment _dl_skip_args instead of setting rtld_command.
If the link_map for the executable itself is not first in the chain,
make it so.
* sysdeps/i386/dl-machine.h (RTLD_START): Use _dl_skip_args as
count of args to skip.
Thu Sep 28 09:20:04 1995 Ulrich Drepper <drepper@gnu.ai.mit.edu>
* stdlib/strtod.c (STRTOF): Fix handling of numbers with lots of
leading zeroes.
Diffstat (limited to 'intl')
-rw-r--r-- | intl/Makefile | 9 | ||||
-rw-r--r-- | intl/bindtextdom.c | 172 | ||||
-rw-r--r-- | intl/dcgettext.c | 523 | ||||
-rw-r--r-- | intl/dgettext.c | 59 | ||||
-rw-r--r-- | intl/finddomain.c | 476 | ||||
-rw-r--r-- | intl/gettext.c | 70 | ||||
-rw-r--r-- | intl/gettext.h | 105 | ||||
-rw-r--r-- | intl/gettextP.h | 76 | ||||
-rw-r--r-- | intl/hash-string.h | 56 | ||||
-rw-r--r-- | intl/loadmsgcat.c | 184 | ||||
-rw-r--r-- | intl/localealias.c | 315 | ||||
-rw-r--r-- | intl/textdomain.c | 97 |
12 files changed, 2142 insertions, 0 deletions
diff --git a/intl/Makefile b/intl/Makefile new file mode 100644 index 0000000000..544daae8a2 --- /dev/null +++ b/intl/Makefile @@ -0,0 +1,9 @@ +subdir = intl +routines = bindtextdom dcgettext dgettext gettext \ + finddomain loadmsgcat localealias textdomain +distribute = gettext.h gettextP.h hash-string.h libgettext.h + +include ../Rules + +CPPFLAGS += -D'GNULOCALEDIR="$(localedir)"' \ + -D'LOCALE_ALIAS_PATH="$(localedir):$(nlsdir)"' diff --git a/intl/bindtextdom.c b/intl/bindtextdom.c new file mode 100644 index 0000000000..bd9f422654 --- /dev/null +++ b/intl/bindtextdom.c @@ -0,0 +1,172 @@ +/* bindtextdom.c -- implementation of the bindtextdomain(3) function + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#else +# ifdef HAVE_MALLOC_H +# include <malloc.h> +# else +void free (); +# endif +#endif + +#if defined HAVE_STRING_H || defined _LIBC +# include <string.h> +#else +# include <strings.h> +#endif + +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif +#include "gettext.h" +#include "gettextP.h" + +/* @@ end of prolog @@ */ + +/* Contains the default location of the message catalogs. */ +extern const char _nl_default_dirname[]; + +/* List with bindings of specific domains. */ +extern struct binding *_nl_domain_bindings; + + +/* Names for the libintl functions are a problem. They must not clash + with existing names and they should follow ANSI C. But this source + code is also used in GNU C Library where the names have a __ + prefix. So we have to make a difference here. */ +#ifdef _LIBC +# define BINDTEXTDOMAIN __bindtextdomain +#else +# define BINDTEXTDOMAIN bindtextdomain__ +#endif + +/* Specify that the DOMAINNAME message catalog will be found + in DIRNAME rather than in the system locale data base. */ +char * +BINDTEXTDOMAIN (domainname, dirname) + const char *domainname; + const char *dirname; +{ + struct binding *binding; + + /* Some sanity checks. */ + if (domainname == NULL || domainname[0] == '\0') + return NULL; + + for (binding = _nl_domain_bindings; binding != NULL; binding = binding->next) + { + int compare = strcmp (domainname, binding->domainname); + if (compare == 0) + /* We found it! */ + break; + if (compare < 0) + { + /* It is not in the list. */ + binding = NULL; + break; + } + } + + if (dirname == NULL) + /* The current binding has be to returned. */ + return binding == NULL ? (char *) _nl_default_dirname : binding->dirname; + + if (binding != NULL) + { + /* The domain is already bound. Replace the old binding. */ + char *new_dirname; + + if (strcmp (dirname, _nl_default_dirname) == 0) + new_dirname = (char *) _nl_default_dirname; + else + { + size_t len = strlen (dirname) + 1; + new_dirname = (char *) malloc (len); + if (new_dirname == NULL) + return NULL; + + memcpy (new_dirname, dirname, len); + } + + if (strcmp (binding->dirname, _nl_default_dirname) != 0) + free (binding->dirname); + + binding->dirname = new_dirname; + } + else + { + /* We have to create a new binding. */ + size_t len; + struct binding *new_binding = + (struct binding *) malloc (sizeof (*new_binding)); + + if (new_binding == NULL) + return NULL; + + len = strlen (domainname) + 1; + new_binding->domainname = (char *) malloc (len); + if (new_binding->domainname == NULL) + return NULL; + memcpy (new_binding->domainname, domainname, len); + + if (strcmp (dirname, _nl_default_dirname) == 0) + new_binding->dirname = (char *) _nl_default_dirname; + else + { + len = strlen (dirname) + 1; + new_binding->dirname = (char *) malloc (len); + if (new_binding->dirname == NULL) + return NULL; + memcpy (new_binding->dirname, dirname, len); + } + + /* Now enqueue it. */ + if (_nl_domain_bindings == NULL + || strcmp (domainname, _nl_domain_bindings->domainname) < 0) + { + new_binding->next = _nl_domain_bindings; + _nl_domain_bindings = new_binding; + } + else + { + binding = _nl_domain_bindings; + while (binding->next != NULL + && strcmp (domainname, binding->next->domainname) > 0) + binding = binding->next; + + new_binding->next = binding->next; + binding->next = new_binding; + } + + binding = new_binding; + } + + return binding->dirname; +} + +#ifdef _LIBC +/* Alias for function name in GNU C Library. */ +weak_alias (__bindtextdomain, bindtextdomain); +#endif diff --git a/intl/dcgettext.c b/intl/dcgettext.c new file mode 100644 index 0000000000..91025be89b --- /dev/null +++ b/intl/dcgettext.c @@ -0,0 +1,523 @@ +/* dcgettext.c -- implemenatation of the dcgettext(3) function + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#include <sys/types.h> + +#ifdef __GNUC__ +# define alloca __builtin_alloca +#else +# ifdef HAVE_ALLOCA_H || defined _LIBC +# include <alloca.h> +# else +# ifdef _AIX + #pragma alloca +# else +# ifndef alloca +char *alloca (); +# endif +# endif +# endif +#endif + +#include <errno.h> +#ifndef errno +extern int errno; +#endif + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#else +char *getenv (); +# ifdef HAVE_MALLOC_H +# include <malloc.h> +# else +void free (); +# endif +#endif + +#if defined HAVE_STRING_H || defined _LIBC +# include <string.h> +#else +# include <strings.h> +#endif +#if !HAVE_STRCHR && !defined _LIBC +# ifndef strchr +# define strchr index +# endif +#endif + +#if defined HAVE_UNISTD_H || defined _LIBC +# include <unistd.h> +#endif + +#include "gettext.h" +#include "gettextP.h" +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif +#include "hash-string.h" + +/* @@ end of prolog @@ */ + +#ifdef _LIBC +/* Rename the non ANSI C functions. This is required by the standard + because some ANSI C functions will require linking with this object + file and the name space must not be polluted. */ +# define getcwd __getcwd +# define stpcpy __stpcpy +#endif + +#if !defined HAVE_GETCWD && !defined _LIBC +char *getwd (); +# define getcwd(buf, max) getwd (buf) +#else +char *getcwd (); +#endif + +/* Amount to increase buffer size by in each try. */ +#define PATH_INCR 32 + +/* The following is from pathmax.h. */ +/* Non-POSIX BSD systems might have gcc's limits.h, which doesn't define + PATH_MAX but might cause redefinition warnings when sys/param.h is + later included (as on MORE/BSD 4.3). */ +#if defined(_POSIX_VERSION) || (defined(HAVE_LIMITS_H) && !defined(__GNUC__)) +# include <limits.h> +#endif + +#ifndef _POSIX_PATH_MAX +# define _POSIX_PATH_MAX 255 +#endif + +#if !defined(PATH_MAX) && defined(_PC_PATH_MAX) +# define PATH_MAX (pathconf ("/", _PC_PATH_MAX) < 1 ? 1024 : pathconf ("/", _PC_PATH_MAX)) +#endif + +/* Don't include sys/param.h if it already has been. */ +#if defined(HAVE_SYS_PARAM_H) && !defined(PATH_MAX) && !defined(MAXPATHLEN) +# include <sys/param.h> +#endif + +#if !defined(PATH_MAX) && defined(MAXPATHLEN) +# define PATH_MAX MAXPATHLEN +#endif + +#ifndef PATH_MAX +# define PATH_MAX _POSIX_PATH_MAX +#endif + +/* XPG3 defines the result of `setlocale (category, NULL)' as: + ``Directs `setlocale()' to query `category' and return the current + setting of `local'.'' + However it does not specify the exact format. And even worse: POSIX + defines this not at all. So we can use this feature only on selected + system (e.g. those using GNU C Library). */ +#ifdef _LIBC +# define HAVE_LOCALE_NULL +#endif + +/* Name of the default domain used for gettext(3) prior any call to + textdomain(3). The default value for this is "messages". */ +const char _nl_default_default_domain[] = "messages"; + +/* Value used as the default domain for gettext(3). */ +const char *_nl_current_default_domain = _nl_default_default_domain; + +/* Contains the default location of the message catalogs. */ +const char _nl_default_dirname[] = GNULOCALEDIR; + +/* List with bindings of specific domains created by bindtextdomain() + calls. */ +struct binding *_nl_domain_bindings; + +/* Prototypes for local functions. */ +static char *find_msg __P ((struct loaded_domain *domain, const char *msgid)); +static const char *category_to_name __P((int category)); +static const char *guess_category_value __P((int category, + const char *categoryname)); + + +/* Names for the libintl functions are a problem. They must not clash + with existing names and they should follow ANSI C. But this source + code is also used in GNU C Library where the names have a __ + prefix. So we have to make a difference here. */ +#ifdef _LIBC +# define DCGETTEXT __dcgettext +#else +# define DCGETTEXT dcgettext__ +#endif + +/* Look up MSGID in the DOMAINNAME message catalog for the current CATEGORY + locale. */ +char * +DCGETTEXT (domainname, msgid, category) + const char *domainname; + const char *msgid; + int category; +{ + struct loaded_domain *domain; + struct binding *binding; + const char *categoryname; + const char *categoryvalue; + char *dirname, *xdomainname; + char *single_locale; + char *retval; + + /* If no real MSGID is given return NULL. */ + if (msgid == NULL) + return NULL; + + /* If DOMAINNAME is NULL, we are interested in the default domain. If + CATEGORY is not LC_MESSAGES this might not make much sense but the + defintion left this undefined. */ + if (domainname == NULL) + domainname = _nl_current_default_domain; + + /* First find matching binding. */ + for (binding = _nl_domain_bindings; binding != NULL; binding = binding->next) + { + int compare = strcmp (domainname, binding->domainname); + if (compare == 0) + /* We found it! */ + break; + if (compare < 0) + { + /* It is not in the list. */ + binding = NULL; + break; + } + } + + if (binding == NULL) + dirname = (char *) _nl_default_dirname; + else if (binding->dirname[0] == '/') + dirname = binding->dirname; + else + { + /* We have a relative path. Make it absolute now. */ + size_t dirname_len = strlen (binding->dirname) + 1; + size_t path_max; + char *ret; + + path_max = (unsigned) PATH_MAX; + path_max += 2; /* The getcwd docs say to do this. */ + + dirname = (char *) alloca (path_max + dirname_len); + + errno = 0; + while ((ret = getcwd (dirname, path_max)) == NULL && errno == ERANGE) + { + path_max += PATH_INCR; + dirname = (char *) alloca (path_max + dirname_len); + errno = 0; + } + + if (ret == NULL) + /* We cannot get the current working directory. Don't signal an + error but simply return the default string. */ + return (char *) msgid; + + /* We don't want libintl.a to depend on any other library. So + we avoid the non-standard function stpcpy. In GNU C Library + this function is available, though. Also allow the symbol + HAVE_STPCPY to be defined. */ +#if defined _LIBC || defined HAVE_STPCPY + stpcpy (stpcpy (strchr (dirname, '\0'), "/"), binding->dirname); +#else + strcat (dirname, "/"); + strcat (dirname, binding->dirname); +#endif + } + + /* Now determine the symbolic name of CATEGORY and its value. */ + categoryname = category_to_name (category); + categoryvalue = guess_category_value (category, categoryname); + + xdomainname = (char *) alloca (strlen (categoryname) + + strlen (domainname) + 5); + /* We don't want libintl.a to depend on any other library. So we + avoid the non-standard function stpcpy. In GNU C Library this + function is available, though. Also allow the symbol HAVE_STPCPY + to be defined. */ +#if defined _LIBC || defined HAVE_STPCPY + stpcpy (stpcpy (stpcpy (stpcpy (xdomainname, categoryname), "/"), + domainname), + ".mo"); +#else + strcpy (xdomainname, categoryname); + strcat (xdomainname, "/"); + strcat (xdomainname, domainname); + strcat (xdomainname, ".mo"); +#endif + + /* Creating working area. */ + single_locale = (char *) alloca (strlen (categoryvalue) + 1); + + + /* Search for the given string. This is a loop because we perhaps + got an ordered list of languages to consider for th translation. */ + while (1) + { + /* Make CATEGORYVALUE point to the next element of the list. */ + while (categoryvalue[0] != '\0' && categoryvalue[0] == ':') + ++categoryvalue; + if (categoryvalue[0] == '\0') + { + /* The whole contents of CATEGORYVALUE has been searched but + no valid entry has been found. We solve this situation + by implicitely appending a "C" entry, i.e. no translation + will take place. */ + single_locale[0] = 'C'; + single_locale[1] = '\0'; + } + else + { + char *cp = single_locale; + while (categoryvalue[0] != '\0' && categoryvalue[0] != ':') + *cp++ = *categoryvalue++; + *cp = '\0'; + } + + /* If the current locale value is C (or POSIX) we don't load a + domain. Return the MSGID. */ + if (strcmp (single_locale, "C") == 0 + || strcmp (single_locale, "POSIX") == 0) + return (char *) msgid; + + + /* Find structure describing the message catalog matching the + DOMAINNAME and CATEGORY. */ + domain = _nl_find_domain (dirname, single_locale, xdomainname); + + if (domain != NULL) + { + retval = find_msg (domain, msgid); + + if (retval == NULL) + { + int cnt; + + for (cnt = 6; cnt >= 0 && retval == NULL; --cnt) + if (domain->successor[cnt] != NULL) + { + retval = find_msg (domain->successor[cnt], msgid); + + if (domain->successor[cnt]->data == NULL) + domain->successor[cnt] = NULL; + } + } + + if (retval != NULL) + return retval; + } + } + /* NOTREACHED */ +} + +#ifdef _LIBC +/* Alias for function name in GNU C Library. */ +weak_alias (__dcgettext, dcgettext); +#endif + + +static char * +find_msg (domain, msgid) + struct loaded_domain *domain; + const char *msgid; +{ + size_t top, act, bottom; + + if (domain->decided == 0) + _nl_load_domain (domain); + + if (domain->data == NULL) + return NULL; + + /* Locate the MSGID and its translation. */ + if (domain->hash_size > 2 && domain->hash_tab != NULL) + { + /* Use the hashing table. */ + nls_uint32 len = strlen (msgid); + nls_uint32 hash_val = hash_string (msgid); + nls_uint32 idx = hash_val % domain->hash_size; + nls_uint32 incr = 1 + (hash_val % (domain->hash_size - 2)); + nls_uint32 nstr = W (domain->must_swap, domain->hash_tab[idx]); + + if (nstr == 0) + /* Hash table entry is empty. */ + return NULL; + + if (W (domain->must_swap, domain->orig_tab[nstr - 1].length) == len + && strcmp (msgid, + domain->data + W (domain->must_swap, + domain->orig_tab[nstr - 1].offset)) == 0) + return (char *) domain->data + W (domain->must_swap, + domain->trans_tab[nstr - 1].offset); + + while (1) + { + if (idx >= W (domain->must_swap, domain->hash_size) - incr) + idx -= W (domain->must_swap, domain->hash_size) - incr; + else + idx += incr; + + nstr = W (domain->must_swap, domain->hash_tab[idx]); + if (nstr == 0) + /* Hash table entry is empty. */ + return NULL; + + if (W (domain->must_swap, domain->orig_tab[nstr - 1].length) == len + && strcmp (msgid, + domain->data + W (domain->must_swap, + domain->orig_tab[nstr - 1].offset)) + == 0) + return (char *) domain->data + + W (domain->must_swap, domain->trans_tab[nstr - 1].offset); + } + /* NOTREACHED */ + } + + /* Now we try the default method: binary search in the sorted + array of messages. */ + bottom = 0; + top = domain->nstrings; + while (bottom < top) + { + int cmp_val; + + act = (bottom + top) / 2; + cmp_val = strcmp (msgid, domain->data + + W (domain->must_swap, + domain->orig_tab[act].offset)); + if (cmp_val < 0) + top = act; + else if (cmp_val > 0) + bottom = act + 1; + else + break; + } + + /* If an translation is found return this. */ + return bottom >= top ? NULL : (char *) domain->data + + W (domain->must_swap, + domain->trans_tab[act].offset); +} + + +/* Return string representation of locale CATEGORY. */ +static const char *category_to_name (category) + int category; +{ + const char *retval; + + switch (category) + { +#ifdef LC_COLLATE + case LC_COLLATE: + retval = "LC_COLLATE"; + break; +#endif +#ifdef LC_CTYPE + case LC_CTYPE: + retval = "LC_CTYPE"; + break; +#endif +#ifdef LC_MONETARY + case LC_MONETARY: + retval = "LC_MONETARY"; + break; +#endif +#ifdef LC_NUMERIC + case LC_NUMERIC: + retval = "LC_NUMERIC"; + break; +#endif +#ifdef LC_TIME + case LC_TIME: + retval = "LC_TIME"; + break; +#endif +#ifdef LC_MESSAGES + case LC_MESSAGES: + retval = "LC_MESSAGES"; + break; +#endif +#ifdef LC_RESPONSE + case LC_RESPONSE: + retval = "LC_RESPONSE"; + break; +#endif +#ifdef LC_ALL + case LC_ALL: + /* This might not make sense but is perhaps better than any other + value. */ + retval = "LC_ALL"; + break; +#endif + default: + /* If you have a better idea for a default value let me know. */ + retval = "LC_XXX"; + } + + return retval; +} + +/* Guess value of current locale from value of the environment variables. */ +static const char *guess_category_value (category, categoryname) + int category; + const char *categoryname; +{ + const char *retval; + + /* The highest priority value is the `LANGUAGE' environment + variable. This is a GNU extension. */ + retval = getenv ("LANGUAGE"); + if (retval != NULL && retval[0] != '\0') + return retval; + + /* `LANGUAGE' is not set. So we have to proceed with the POSIX + methods of looking to `LC_ALL', `LC_xxx', and `LANG'. On some + systems this can be done by the `setlocale' function itself. */ +#if defined HAVE_SETLOCALE && defined HAVE_LC_MESSAGES && defined HAVE_LOCALE_NULL + return setlocale (category, NULL); +#else + /* Setting of LC_ALL overwrites all other. */ + retval = getenv ("LC_ALL"); + if (retval != NULL && retval[0] != '\0') + return retval; + + /* Next comes the name of the desired category. */ + retval = getenv (categoryname); + if (retval != NULL && retval[0] != '\0') + return retval; + + /* Last possibility is the LANG environment variable. */ + retval = getenv ("LANG"); + if (retval != NULL && retval[0] != '\0') + return retval; + + /* We use C as the default domain. POSIX says this is implementation + defined. */ + return "C"; +#endif +} diff --git a/intl/dgettext.c b/intl/dgettext.c new file mode 100644 index 0000000000..2fde6770f7 --- /dev/null +++ b/intl/dgettext.c @@ -0,0 +1,59 @@ +/* dgettext.c -- implementation of the dgettext(3) function + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#if defined HAVE_LOCALE_H || defined _LIBC +# include <locale.h> +#endif + +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif + +/* @@ end of prolog @@ */ + +/* Names for the libintl functions are a problem. They must not clash + with existing names and they should follow ANSI C. But this source + code is also used in GNU C Library where the names have a __ + prefix. So we have to make a difference here. */ +#ifdef _LIBC +# define DGETTEXT __dgettext +# define DCGETTEXT __dcgettext +#else +# define DGETTEXT dgettext__ +# define DCGETTEXT dcgettext__ +#endif + +/* Look up MSGID in the DOMAINNAME message catalog of the current + LC_MESSAGES locale. */ +char * +DGETTEXT (domainname, msgid) + const char *domainname; + const char *msgid; +{ + return DCGETTEXT (domainname, msgid, LC_MESSAGES); +} + +#ifdef _LIBC +/* Alias for function name in GNU C Library. */ +weak_alias (__dgettext, dgettext); +#endif diff --git a/intl/finddomain.c b/intl/finddomain.c new file mode 100644 index 0000000000..19cf2d7581 --- /dev/null +++ b/intl/finddomain.c @@ -0,0 +1,476 @@ +/* finddomain.c -- handle list of needed message catalogs + Copyright (C) 1995 Software Foundation, Inc. + Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995. + +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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#include <errno.h> +#include <stdio.h> +#include <sys/types.h> + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#else +# ifdef HAVE_MALLOC_H +# include <malloc.h> +# else +void free (); +# endif +#endif + +#if defined HAVE_STRING_H || defined _LIBC +# include <string.h> +#else +# include <strings.h> +#endif +#if !HAVE_STRCHR && !defined _LIBC +# ifndef strchr +# define strchr index +# endif +#endif + +#if defined HAVE_UNISTD_H || defined _LIBC +# include <unistd.h> +#endif + +#include "gettext.h" +#include "gettextP.h" +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif + +/* @@ end of prolog @@ */ + +#ifdef _LIBC +/* Rename the non ANSI C functions. This is required by the standard + because some ANSI C functions will require linking with this object + file and the name space must not be polluted. */ +# define stpcpy __stpcpy +#endif + +/* Encoding of locale name parts. */ +#define CEN_REVISION 1 +#define CEN_SPONSOR 2 +#define CEN_SPECIAL 4 +#define XPG_CODESET 8 +#define TERRITORY 16 +#define CEN_AUDIENCE 32 +#define XPG_MODIFIER 64 + +#define CEN_SPECIFIC (CEN_REVISION|CEN_SPONSOR|CEN_SPECIAL|CEN_AUDIENCE) +#define XPG_SPECIFIC (XPG_CODESET|XPG_MODIFIER) + + +/* List of already loaded domains. */ +static struct loaded_domain *_nl_loaded_domains; + +/* Prototypes for local functions. */ +static struct loaded_domain *make_entry_rec __P ((const char *dirname, + int mask, + const char *language, + const char *territory, + const char *codeset, + const char *modifier, + const char *special, + const char *sponsor, + const char *revision, + const char *domainname, + int do_allocate)); + + +/* Return a data structure describing the message catalog described by + the DOMAINNAME and CATEGORY parameters with respect to the currently + established bindings. */ +struct loaded_domain * +_nl_find_domain (dirname, locale, domainname) + const char *dirname; + char *locale; + const char *domainname; +{ + enum { undecided, xpg, cen } syntax; + struct loaded_domain *retval; + const char *language; + const char *modifier = NULL; + const char *territory = NULL; + const char *codeset = NULL; + const char *special = NULL; + const char *sponsor = NULL; + const char *revision = NULL; + const char *alias_value = NULL; + char *cp; + int mask; + + /* CATEGORYVALUE now possibly contains a colon separated list of + locales. Each single locale can consist of up to four recognized + parts for the XPG syntax: + + language[_territory[.codeset]][@modifier] + + and six parts for the CEN syntax: + + language[_territory][+audience][+special][,sponsor][_revision] + + Beside the first all of them are allowed to be missing. If the + full specified locale is not found, the less specific one are + looked for. The various part will be stripped of according to + the following order: + (1) revision + (2) sponsor + (3) special + (4) codeset + (5) territory + (6) audience/modifier + */ + + /* If we have already tested for this locale entry there has to + be one data set in the list of loaded domains. */ + retval = make_entry_rec (dirname, 0, locale, NULL, NULL, NULL, + NULL, NULL, NULL, domainname, 0); + if (retval != NULL) + { + /* We know something about this locale. */ + int cnt; + + if (retval->decided == 0) + _nl_load_domain (retval); /* @@@ */ + + if (retval->data != NULL) + return retval; + + for (cnt = 6; cnt >= 0; --cnt) + { + if (retval->successor[cnt] == 0) + _nl_load_domain (retval->successor[cnt]); + + if (retval->successor[cnt]->data != NULL) + break; + } + + /* We really found some usable information. */ + return cnt >= 0 ? retval : NULL; + /* NOTREACHED */ + } + + /* See whether the locale value is an alias. If yes its value + *overwrites* the alias name. No test for the original value is + done. */ + alias_value = _nl_expand_alias (locale); + if (alias_value != NULL) + { + size_t len = strlen (alias_value) + 1; + locale = (char *) malloc (len); + if (locale == NULL) + return NULL; + + memcpy (locale, alias_value, len); + } + + /* Now we determine the single parts of the locale name. First + look for the language. Termination symbols are `_' and `@' if + we use XPG4 style, and `_', `+', and `,' if we use CEN syntax. */ + mask = 0; + syntax = undecided; + language = cp = locale; + while (cp[0] != '\0' && cp[0] != '_' && cp[0] != '@' + && cp[0] != '+' && cp[0] != ',') + ++cp; + + if (language == cp) + /* This does not make sense: language has to be specified. Use + this entry as it is without exploding. Perhaps it is an alias. */ + cp = strchr (language, '\0'); + else if (cp[0] == '_') + { + /* Next is the territory. */ + cp[0] = '\0'; + territory = ++cp; + + while (cp[0] != '\0' && cp[0] != '.' && cp[0] != '@' + && cp[0] != '+' && cp[0] != ',' && cp[0] != '_') + ++cp; + + mask |= TERRITORY; + + if (cp[0] == '.') + { + /* Next is the codeset. */ + syntax = xpg; + cp[0] = '\0'; + codeset = ++cp; + + while (cp[0] != '\0' && cp[0] != '@') + ++cp; + + mask |= XPG_CODESET; + } + } + + if (cp[0] == '@' || (syntax != xpg && cp[0] == '+')) + { + /* Next is the modifier. */ + syntax = cp[0] == '@' ? xpg : cen; + cp[0] = '\0'; + modifier = ++cp; + + while (syntax == cen && cp[0] != '\0' && cp[0] != '+' + && cp[0] != ',' && cp[0] != '_') + ++cp; + + mask |= XPG_MODIFIER | CEN_AUDIENCE; + } + + if (syntax != xpg && (cp[0] == '+' || cp[0] == ',' || cp[0] == '_')) + { + syntax = cen; + + if (cp[0] == '+') + { + /* Next is special application (CEN syntax). */ + cp[0] = '\0'; + special = ++cp; + + while (cp[0] != '\0' && cp[0] != ',' && cp[0] != '_') + ++cp; + + mask |= CEN_SPECIAL; + } + + if (cp[0] == ',') + { + /* Next is sponsor (CEN syntax). */ + cp[0] = '\0'; + sponsor = ++cp; + + while (cp[0] != '\0' && cp[0] != '_') + ++cp; + + mask |= CEN_SPONSOR; + } + + if (cp[0] == '_') + { + /* Next is revision (CEN syntax). */ + cp[0] = '\0'; + revision = ++cp; + + mask |= CEN_REVISION; + } + } + + /* For CEN sytnax values it might be important to have the + separator character in the file name, not for XPG syntax. */ + if (syntax == xpg) + { + if (territory[0] == '\0') + mask &= ~TERRITORY; + + if (codeset[0] == '\0') + mask &= ~XPG_CODESET; + + if (modifier[0] == '\0') + mask &= ~XPG_MODIFIER; + } + + /* Create all possible locale entries which might be interested in + generalzation. */ + retval = make_entry_rec (dirname, mask, language, territory, codeset, + modifier, special, sponsor, revision, + domainname, 1); + if (retval == NULL) + /* This means we are out of core. */ + return NULL; + + if (retval->decided == 0) + _nl_load_domain (retval); + if (retval->data == NULL) + { + int cnt; + for (cnt = 0; retval->successor[cnt] != NULL; ++cnt) + { + if (retval->successor[cnt]->decided == 0) + _nl_load_domain (retval->successor[cnt]); + if (retval->successor[cnt]->data != NULL) + break; + + /* Signal that locale is not available. */ + retval->successor[cnt] = NULL; + } + if (retval->successor[cnt] == NULL) + retval = NULL; + } + + /* The room for an alias was dynamically allocated. Free it now. */ + if (alias_value != NULL) + free (locale); + + return retval; +} + + +static struct loaded_domain * +make_entry_rec (dirname, mask, language, territory, codeset, modifier, + special, sponsor, revision, domain, do_allocate) + const char *dirname; + int mask; + const char *language; + const char *territory; + const char *codeset; + const char *modifier; + const char *special; + const char *sponsor; + const char *revision; + const char *domain; + int do_allocate; +{ + struct loaded_domain *retval, *last; + char *filename, *cp; + size_t entries; + int cnt; + + /* Allocate room for the full file name. */ + filename = (char *) malloc (strlen (dirname) + 1 + + strlen (language) + + ((mask & TERRITORY) != 0 + ? strlen (territory) : 0) + + ((mask & XPG_CODESET) != 0 + ? strlen (codeset) : 0) + + ((mask & XPG_MODIFIER) != 0 ? + strlen (modifier) : 0) + + ((mask & CEN_SPECIAL) != 0 + ? strlen (special) : 0) + + ((mask & CEN_SPONSOR) != 0 + ? strlen (sponsor) : 0) + + ((mask & CEN_REVISION) != 0 + ? strlen (revision) : 0) + 1 + + strlen (domain) + 1); + + if (filename == NULL) + return NULL; + + retval = NULL; + last = NULL; + + /* We don't want libintl.a to depend on any other library. So we + avoid the non-standard function stpcpy. In GNU C Library this + function is available, though. Also allow the symbol HAVE_STPCPY + to be defined. */ +#if !defined _LIBC && !defined HAVE_STPCPY +# define stpcpy(p, s) \ + (strcpy (p, s), strchr (p, '\0')) +#endif + + /* Construct file name. */ + cp = stpcpy (filename, dirname); + *cp++ = '/'; + cp = stpcpy (cp, language); + + if ((mask & TERRITORY) != 0) + { + *cp++ = '_'; + cp = stpcpy (cp, territory); + } + if ((mask & XPG_CODESET) != 0) + { + *cp++ = '.'; + cp = stpcpy (cp, codeset); + } + if ((mask & (XPG_MODIFIER | CEN_AUDIENCE)) != 0) + { + /* This component can be part of both syntaces but has different + leading characters. For CEN we use `+', else `@'. */ + *cp++ = (mask & CEN_AUDIENCE) != 0 ? '+' : '@'; + cp = stpcpy (cp, modifier); + } + if ((mask & CEN_SPECIAL) != 0) + { + *cp++ = '+'; + cp = stpcpy (cp, special); + } + if ((mask & CEN_SPONSOR) != 0) + { + *cp++ = ','; + cp = stpcpy (cp, sponsor); + } + if ((mask & CEN_REVISION) != 0) + { + *cp++ = '_'; + cp = stpcpy (cp, revision); + } + + *cp++ = '/'; + stpcpy (cp, domain); + + /* Look in list of already loaded domains whether it is already + available. */ + last = NULL; + for (retval = _nl_loaded_domains; retval != NULL; retval = retval->next) + { + int compare = strcmp (retval->filename, filename); + if (compare == 0) + /* We found it! */ + break; + if (compare < 0) + { + /* It's not in the list. */ + retval = NULL; + break; + } + + last = retval; + } + + if (retval != NULL || do_allocate == 0) + { + free (filename); + return retval; + } + + retval = (struct loaded_domain *) malloc (sizeof (*retval)); + if (retval == NULL) + return NULL; + + retval->filename = filename; + retval->decided = 0; + + if (last == NULL) + { + retval->next = _nl_loaded_domains; + _nl_loaded_domains = retval; + } + else + { + retval->next = last->next; + last->next = retval; + } + + entries = 0; + for (cnt = 126; cnt >= 0; --cnt) + if (cnt < mask && (cnt & ~mask) == 0 + && ((cnt & CEN_SPECIFIC) == 0 || (cnt & XPG_SPECIFIC) == 0)) + retval->successor[entries++] = make_entry_rec (dirname, cnt, + language, territory, + codeset, modifier, + special, sponsor, + revision, domain, 1); + retval->successor[entries] = NULL; + + return retval; +} diff --git a/intl/gettext.c b/intl/gettext.c new file mode 100644 index 0000000000..204d85ecd4 --- /dev/null +++ b/intl/gettext.c @@ -0,0 +1,70 @@ +/* gettext.c -- implementation of gettext(3) function + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#ifdef _LIBC +# define __need_NULL +# include <stddef.h> +#else +# ifdef STDC_HEADERS +# include <stdlib.h> /* Just for NULL. */ +# else +# ifdef HAVE_STRING_H +# include <string.h> +# else +# define NULL 0 +# endif +# endif +#endif + +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif + +/* @@ end of prolog @@ */ + +/* Names for the libintl functions are a problem. They must not clash + with existing names and they should follow ANSI C. But this source + code is also used in GNU C Library where the names have a __ + prefix. So we have to make a difference here. */ +#ifdef _LIBC +# define GETTEXT __gettext +# define DGETTEXT __dgettext +#else +# define GETTEXT gettext__ +# define DGETTEXT dgettext__ +#endif + +/* Look up MSGID in the current default message catalog for the current + LC_MESSAGES locale. If not found, returns MSGID itself (the default + text). */ +char * +GETTEXT (msgid) + const char *msgid; +{ + return DGETTEXT (NULL, msgid); +} + +#ifdef _LIBC +/* Alias for function name in GNU C Library. */ +weak_alias (__gettext, gettext); +#endif diff --git a/intl/gettext.h b/intl/gettext.h new file mode 100644 index 0000000000..5190f09447 --- /dev/null +++ b/intl/gettext.h @@ -0,0 +1,105 @@ +/* gettext.h - internal header for GNU gettext internationalization functions + Copyright (C) 1995 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 2, 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 Library General Public +License along with the GNU C Library; see the file COPYING.LIB. If +not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, +Boston, MA 02111-1307, USA. */ + +#ifndef _GETTEXT_H +#define _GETTEXT_H 1 + +#include <stdio.h> + +#if HAVE_LIMITS_H || _LIBC +# include <limits.h> +#endif + +/* @@ end of prolog @@ */ + +/* The magic number of the GNU message catalog format. */ +#define _MAGIC 0x950412de +#define _MAGIC_SWAPPED 0xde120495 + +/* Revision number of the currently used .mo (binary) file format. */ +#define MO_REVISION_NUMBER 0 + +/* The following contortions are an attempt to use the C preprocessor + to determine an unsigned integral type that is 32 bits wide. An + alternative approach is to use autoconf's AC_CHECK_SIZEOF macro, but + doing that would require that the configure script compile and *run* + the resulting executable. Locally running cross-compiled executables + is usually not possible. */ + +#if __STDC__ +# define UINT_MAX_32_BITS 4294967295U +#else +# define UINT_MAX_32_BITS 0xFFFFFFFF +#endif + +/* If UINT_MAX isn't defined, assume it's a 32-bit type. + This should be valid for all systems GNU cares about because + that doesn't include 16-bit systems, and only modern systems + (that certainly have <limits.h>) have 64+-bit integral types. */ + +#ifndef UINT_MAX +# define UINT_MAX UINT_MAX_32_BITS +#endif + +#if UINT_MAX == UINT_MAX_32_BITS +typedef unsigned nls_uint32; +#else +# if USHRT_MAX == UINT_MAX_32_BITS +typedef unsigned short nls_uint32; +# else +# if ULONG_MAX == UINT_MAX_32_BITS +typedef unsigned long nls_uint32; +# else + /* The following line is intended to throw an error. Using #error is + not portable enough. */ + "Cannot determine unsigned 32-bit data type." +# endif +# endif +#endif + + +/* Header for binary .mo file format. */ +struct mo_file_header +{ + /* The magic number. */ + nls_uint32 magic; + /* The revision number of the file format. */ + nls_uint32 revision; + /* The number of strings pairs. */ + nls_uint32 nstrings; + /* Offset of table with start offsets of original strings. */ + nls_uint32 orig_tab_offset; + /* Offset of table with start offsets of translation strings. */ + nls_uint32 trans_tab_offset; + /* Size of hashing table. */ + nls_uint32 hash_tab_size; + /* Offset of first hashing entry. */ + nls_uint32 hash_tab_offset; +}; + +struct string_desc +{ + /* Length of addressed string. */ + nls_uint32 length; + /* Offset of string in file. */ + nls_uint32 offset; +}; + +/* @@ begin of epilog @@ */ + +#endif /* gettext.h */ diff --git a/intl/gettextP.h b/intl/gettextP.h new file mode 100644 index 0000000000..12031c605b --- /dev/null +++ b/intl/gettextP.h @@ -0,0 +1,76 @@ +/* gettextP.h -- header describing internals of gettext library + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifndef _GETTEXTP_H +#define _GETTEXTP_H + +/* @@ end of prolog @@ */ + +#ifndef __P +# if __STDC__ +# define __P(args) args +# else +# define __P(args) () +# endif +#endif + +#ifndef W +# define W(flag, data) ((flag) ? SWAP (data) : (data)) +#endif + +static inline nls_uint32 +SWAP (i) + nls_uint32 i; +{ + return (i << 24) | ((i & 0xff00) << 8) | ((i >> 8) & 0xff00) | (i >> 24); +} + + +struct loaded_domain +{ + struct loaded_domain *next; + struct loaded_domain *successor[31]; + + const char *filename; + int decided; + + const char *data; + int must_swap; + nls_uint32 nstrings; + struct string_desc *orig_tab; + struct string_desc *trans_tab; + nls_uint32 hash_size; + nls_uint32 *hash_tab; +}; + +struct binding +{ + struct binding *next; + char *domainname; + char *dirname; +}; + +struct loaded_domain *_nl_find_domain __P ((const char *__dirname, + char *__locale, + const char *__domainname)); +void _nl_load_domain __P ((struct loaded_domain *__domain)); + +const char *_nl_expand_alias __P ((const char *__name)); + +/* @@ begin of epilog @@ */ + +#endif /* gettextP.h */ diff --git a/intl/hash-string.h b/intl/hash-string.h new file mode 100644 index 0000000000..e846b507ac --- /dev/null +++ b/intl/hash-string.h @@ -0,0 +1,56 @@ +/* hash-string - Implements a string hashing function. + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_VALUES_H +# include <values.h> +#endif + +/* @@ end of prolog @@ */ + +#ifndef BITSPERBYTE +# define BITSPERBYTE 8 +#endif + +#ifndef LONGBITS +# define LONGBITS (sizeof (long) * BITSPERBYTE) +#endif /* LONGBITS */ + +/* Defines the so called `hashpjw' function by P.J. Weinberger + [see Aho/Sethi/Ullman, COMPILERS: Principles, Techniques and Tools, + 1986, 1987 Bell Telephone Laboratories, Inc.] */ +static inline unsigned long +hash_string (str_param) + const char *str_param; +{ + unsigned long hval, g; + const char *str = str_param; + + /* Compute the hash value for the given string. */ + hval = 0; + while (*str != '\0') + { + hval <<= 4; + hval += (unsigned long) *str++; + g = hval & ((unsigned long) 0xf << (LONGBITS - 4)); + if (g != 0) + { + hval ^= g >> (LONGBITS - 8); + hval ^= g; + } + } + return hval; +} diff --git a/intl/loadmsgcat.c b/intl/loadmsgcat.c new file mode 100644 index 0000000000..b05c630724 --- /dev/null +++ b/intl/loadmsgcat.c @@ -0,0 +1,184 @@ +/* loadmsgcat.c -- load needed message catalogs + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#include <fcntl.h> +#include <sys/types.h> +#include <sys/stat.h> + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#endif + +#if defined HAVE_UNISTD_H || defined _LIBC +# include <unistd.h> +#endif + +#if (defined HAVE_MMAP && defined HAVE_MUNMAP) || defined _LIBC +# include <sys/mman.h> +#endif + +#include "gettext.h" +#include "gettextP.h" + +/* @@ end of prolog @@ */ + +#ifdef _LIBC +/* Rename the non ANSI C functions. This is required by the standard + because some ANSI C functions will require linking with this object + file and the name space must not be polluted. */ +# define fstat __fstat +# define open __open +# define close __close +# define read __read +# define mmap __mmap +# define munmap __munmap +#endif + +/* We need a sign, whether a new catalog was loaded, which can be associated + with all translations. This is important if the translations are + cached by one of GCC's features. */ +int _nl_msg_cat_cntr; + + +/* Load the message catalogs specified by FILENAME. If it is no valid + message catalog do nothing. */ +void +_nl_load_domain (domain) + struct loaded_domain *domain; +{ + int fd; + struct stat st; + struct mo_file_header *data = (struct mo_file_header *) -1; +#if (defined HAVE_MMAP && defined HAVE_MUNMAP && !defined DISALLOW_MMAP) \ + || defined _LIBC + int use_mmap = 0; +#endif + + domain->decided = 1; + domain->data = NULL; + + /* Try to open the addressed file. */ + fd = open (domain->filename, O_RDONLY); + if (fd == -1) + return; + + /* We must know about the size of the file. */ + if (fstat (fd, &st) != 0 + && st.st_size < (off_t) sizeof (struct mo_file_header)) + { + /* Something went wrong. */ + close (fd); + return; + } + +#if (defined HAVE_MMAP && defined HAVE_MUNMAP && !defined DISALLOW_MMAP) \ + || defined _LIBC + /* Now we are ready to load the file. If mmap() is available we try + this first. If not available or it failed we try to load it. */ + data = (struct mo_file_header *) mmap (NULL, st.st_size, PROT_READ, + MAP_PRIVATE, fd, 0); + + if (data != (struct mo_file_header *) -1) + { + /* mmap() call was successful. */ + close (fd); + use_mmap = 1; + } +#endif + + /* If the data is not yet available (i.e. mmap'ed) we try to load + it manually. */ + if (data == (struct mo_file_header *) -1) + { + off_t to_read; + char *read_ptr; + + data = (struct mo_file_header *) malloc (st.st_size); + if (data == NULL) + return; + + to_read = st.st_size; + read_ptr = (char *) data; + do + { + long int nb = (long int) read (fd, read_ptr, to_read); + if (nb == -1) + { + close (fd); + return; + } + + read_ptr += nb; + to_read -= nb; + } + while (to_read > 0); + + close (fd); + } + + /* Using the magic number we can test whether it really is a message + catalog file. */ + if (data->magic != _MAGIC && data->magic != _MAGIC_SWAPPED) + { + /* The magic number is wrong: not a message catalog file. */ +#if (defined HAVE_MMAP && defined HAVE_MUNMAP && !defined DISALLOW_MMAP) \ + || defined _LIBC + if (use_mmap) + munmap ((caddr_t) data, st.st_size); + else +#endif + free (data); + return; + } + + domain->data = (char *) data; + domain->must_swap = data->magic != _MAGIC; + + /* Fill in the information about the available tables. */ + switch (W (domain->must_swap, data->revision)) + { + case 0: + domain->nstrings = W (domain->must_swap, data->nstrings); + domain->orig_tab = (struct string_desc *) + ((char *) data + W (domain->must_swap, data->orig_tab_offset)); + domain->trans_tab = (struct string_desc *) + ((char *) data + W (domain->must_swap, data->trans_tab_offset)); + domain->hash_size = W (domain->must_swap, data->hash_tab_size); + domain->hash_tab = (nls_uint32 *) + ((char *) data + W (domain->must_swap, data->hash_tab_offset)); + break; + default: + /* This is an illegal revision. */ +#if (defined HAVE_MMAP && defined HAVE_MUNMAP && !defined DISALLOW_MMAP) \ + || defined _LIBC + if (use_mmap) + munmap ((caddr_t) data, st.st_size); + else +#endif + free (data); + domain->data = NULL; + return; + } + + /* Show that one domain is changed. This might make some cached + translation invalid. */ + ++_nl_msg_cat_cntr; +} diff --git a/intl/localealias.c b/intl/localealias.c new file mode 100644 index 0000000000..716657c844 --- /dev/null +++ b/intl/localealias.c @@ -0,0 +1,315 @@ +/* localealias.c -- handle aliases for locale names + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#include <ctype.h> +#include <stdio.h> +#include <sys/types.h> + +#ifdef __GNUC__ +# define alloca __builtin_alloca +#else +# if defined HAVE_ALLOCA_H || defined _LIBC +# include <alloca.h> +# else +# ifdef _AIX + #pragma alloca +# else +# ifndef alloca +char *alloca (); +# endif +# endif +# endif +#endif + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#else +char *getenv (); +# ifdef HAVE_MALLOC_H +# include <malloc.h> +# else +void free (); +# endif +#endif + +#if defined HAVE_STRING_H || defined _LIBC +# include <string.h> +#else +# include <strings.h> +#endif +#if !HAVE_STRCHR && !defined _LIBC +# ifndef strchr +# define strchr index +# endif +#endif + +#include "gettext.h" +#include "gettextP.h" + +/* @@ end of prolog @@ */ + +#ifdef _LIBC +/* Rename the non ANSI C functions. This is required by the standard + because some ANSI C functions will require linking with this object + file and the name space must not be polluted. */ +# define strcasecmp __strcasecmp +#endif + +struct alias_map +{ + const char *alias; + const char *value; +}; + + +static struct alias_map *map; +static size_t nmap = 0; +static size_t maxmap = 0; + + +/* Prototypes for local functions. */ +static size_t read_alias_file __P ((const char *fname, int fname_len)); +static void extend_alias_table __P ((void)); +static int alias_compare __P ((const struct alias_map *map1, + const struct alias_map *map2)); + + +const char * +_nl_expand_alias (name) + const char *name; +{ + static const char *locale_alias_path = LOCALE_ALIAS_PATH; + struct alias_map *retval; + size_t added; + + do + { + struct alias_map item; + + item.alias = name; + + if (nmap > 0) + retval = (struct alias_map *) bsearch (&item, map, nmap, + sizeof (struct alias_map), + (int (*) (const void *, + const void *)) + alias_compare); + else + retval = NULL; + + /* We really found an alias. Return the value. */ + if (retval != NULL) + return retval->value; + + /* Perhaps we can find another alias file. */ + added = 0; + while (added == 0 && locale_alias_path[0] != '\0') + { + const char *start; + + while (locale_alias_path[0] != '\0' && locale_alias_path[0] == ':') + ++locale_alias_path; + start = locale_alias_path; + + while (locale_alias_path[0] != '\0' && locale_alias_path[0] != ':') + ++locale_alias_path; + + if (start < locale_alias_path) + added = read_alias_file (start, locale_alias_path - start); + } + } + while (added != 0); + + return NULL; +} + + +static size_t +read_alias_file (fname, fname_len) + const char *fname; + int fname_len; +{ + FILE *fp; + char *full_fname; + size_t added; + + full_fname = (char *) alloca (fname_len + sizeof ("/locale.alias")); + sprintf (full_fname, "%.*s/locale.alias", fname_len, fname); + + fp = fopen (full_fname, "r"); + if (fp == NULL) + return 0; + + added = 0; + while (!feof (fp)) + { + /* It is a reasonable approach to use a fix buffer here because + a) we are only interested in the first two fields + b) these fields must be usable as file names and so must not + be that long + */ + char buf[BUFSIZ]; + char *alias; + char *value; + char *cp; + + if (fgets (buf, BUFSIZ, fp) == NULL) + /* EOF reached. */ + break; + + cp = buf; + /* Ignore leading white space. */ + while (isspace (cp[0])) + ++cp; + + /* A leading '#' signals a comment line. */ + if (cp[0] != '\0' && cp[0] != '#') + { + alias = cp++; + while (cp[0] != '\0' && !isspace (cp[0])) + ++cp; + /* Terminate alias name. */ + if (cp[0] != '\0') + *cp++ = '\0'; + + /* Now look for the beginning of the value. */ + while (isspace (cp[0])) + ++cp; + + if (cp[0] != '\0') + { + char *tp; + size_t len; + + value = cp++; + while (cp[0] != '\0' && !isspace (cp[0])) + ++cp; + /* Terminate value. */ + if (cp[0] == '\n') + { + /* This has to be done to make the following test + for the end of line possible. We are looking for + the terminating '\n' which do not overwrite here. */ + *cp++ = '\0'; + *cp = '\n'; + } + else if (cp[0] != '\0') + *cp++ = '\0'; + + if (nmap >= maxmap) + extend_alias_table (); + + /* We cannot depend on strdup available in the libc. Sigh! */ + len = strlen (alias) + 1; + tp = (char *) malloc (len); + if (tp == NULL) + return added; + memcpy (tp, alias, len); + map[nmap].alias = tp; + + len = strlen (value) + 1; + tp = (char *) malloc (len); + if (tp == NULL) + return added; + memcpy (tp, value, len); + map[nmap].value = tp; + + ++nmap; + ++added; + } + } + + /* Possibily not the whole line fits into the buffer. Ignore + the rest of the line. */ + while (strchr (cp, '\n') == NULL) + { + cp = buf; + if (fgets (buf, BUFSIZ, fp) == NULL) + /* Make sure the inner loop will be left. The outer loop + will exit at the `feof' test. */ + *cp = '\n'; + } + } + + /* Should we test for ferror()? I think we have to silently ignore + errors. --drepper */ + fclose (fp); + + if (added > 0) + qsort (map, nmap, sizeof (struct alias_map), + (int (*) (const void *, const void *)) alias_compare); + + return added; +} + + +static void +extend_alias_table () +{ + size_t new_size; + struct alias_map *new_map; + + new_size = maxmap == 0 ? 100 : 2 * maxmap; + new_map = (struct alias_map *) malloc (new_size + * sizeof (struct alias_map)); + if (new_map == NULL) + /* Simply don't extend: we don't have any more core. */ + return; + + memcpy (new_map, map, nmap * sizeof (struct alias_map)); + + if (maxmap != 0) + free (map); + + map = new_map; + maxmap = new_size; +} + + +static int +alias_compare (map1, map2) + const struct alias_map *map1; + const struct alias_map *map2; +{ +#if defined _LIBC || defined HAVE_STRCASECMP + return strcasecmp (map1->alias, map2->alias); +#else + const unsigned char *p1 = (const unsigned char *) map1->alias; + const unsigned char *p2 = (const unsigned char *) map2->alias; + unsigned char c1, c2; + + if (p1 == p2) + return 0; + + do + { + /* I know this seems to be odd but the tolower() function in + some systems libc cannot handle nonalpha characters. */ + c1 = isalpha (*p1) ? tolower (*p1) : *p1; + c2 = isalpha (*p2) ? tolower (*p2) : *p2; + if (c1 == '\0') + break; + } + while (c1 == c2); + + return c1 - c2; +#endif +} diff --git a/intl/textdomain.c b/intl/textdomain.c new file mode 100644 index 0000000000..c06642b268 --- /dev/null +++ b/intl/textdomain.c @@ -0,0 +1,97 @@ +/* textdomain.c -- implementation of the textdomain(3) function + Copyright (C) 1995 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 2, 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, write to the Free Software +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ + +#ifdef HAVE_CONFIG_H +# include <config.h> +#endif + +#if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +#endif + +#if defined STDC_HEADERS || defined HAVE_STRING_H || defined _LIBC +# include <string.h> +#else +# include <strings.h> +#endif + +#ifdef _LIBC +# include <libintl.h> +#else +# include "libgettext.h" +#endif + +/* @@ end of prolog @@ */ + +/* Name of the default text domain. */ +extern const char _nl_default_default_domain[]; + +/* Default text domain in which entries for gettext(3) are to be found. */ +extern const char *_nl_current_default_domain; + + +/* Names for the libintl functions are a problem. They must not clash + with existing names and they should follow ANSI C. But this source + code is also used in GNU C Library where the names have a __ + prefix. So we have to make a difference here. */ +#ifdef _LIBC +# define TEXTDOMAIN __textdomain +#else +# define TEXTDOMAIN textdomain__ +#endif + +/* Set the current default message catalog to DOMAINNAME. + If DOMAINNAME is null, return the current default. + If DOMAINNAME is "", reset to the default of "messages". */ +char * +TEXTDOMAIN (domainname) + const char *domainname; +{ + char *old; + + /* A NULL pointer requests the current setting. */ + if (domainname == NULL) + return (char *) _nl_current_default_domain; + + old = (char *) _nl_current_default_domain; + + /* If domain name is the null string set to default domain "messages". */ + if (domainname[0] == '\0' + || strcmp (domainname, _nl_default_default_domain) == 0) + _nl_current_default_domain = _nl_default_default_domain; + else + { + /* If the following malloc fails `_nl_current_default_domain' + will be NULL. This value will be returned and so signals we + are out of core. */ + size_t len = strlen (domainname) + 1; + char *cp = (char *) malloc (len); + if (cp != NULL) + memcpy (cp, domainname, len); + _nl_current_default_domain = cp; + } + + if (old != _nl_default_default_domain) + free (old); + + return (char *) _nl_current_default_domain; +} + +#ifdef _LIBC +/* Alias for function name in GNU C Library. */ +weak_alias (__textdomain, textdomain); +#endif |