add transparent session id propagation

  this uses a custom scanner which detects relative URIs and changes
  them appropiately.
  must be explicitly enabled with --enable-trans-sid
  (normal case is not affected, since session_adapt_uris defaults to
   void statement. Compiler should eliminate dead code in output.c then.)

9 files changed, 605 insertions, 9 deletions

diff --git a/ext/session/config.h.stub b/ext/session/config.h.stub
index 4dc5ec115c..24abf8b280 100644
--- a/ext/session/config.h.stub
+++ b/ext/session/config.h.stub
@@ -1,3 +1,4 @@
 /* define if you want to use the session extension */
 
+#undef TRANS_SID
 #undef HAVE_LIBMM
diff --git a/ext/session/config.m4 b/ext/session/config.m4
index ac39c53cf4..8db20b3324 100644
--- a/ext/session/config.m4
+++ b/ext/session/config.m4
@@ -2,8 +2,7 @@ dnl $Id$
 dnl config.m4 for extension session
 dnl don't forget to call PHP_EXTENSION(session)
 
-
-MM_RESULT=no
+RESULT=no
 AC_MSG_CHECKING(for mm support)
 AC_ARG_WITH(mm,
 [  --with-mm[=DIR]         Include mm support],
@@ -22,10 +21,22 @@ AC_ARG_WITH(mm,
 		AC_ADD_LIBRARY_WITH_PATH(mm, $MM_DIR/lib)
 		AC_ADD_INCLUDE($MM_DIR/include)
 		AC_DEFINE(HAVE_LIBMM, 1)
-		MM_RESULT=yes
+		RESULT=yes
 		PHP_EXTENSION(ps_mm)
 	fi
 ])
-AC_MSG_RESULT($MM_RESULT)
+AC_MSG_RESULT($RESULT)
+
+RESULT=no
+AC_MSG_CHECKING(whether to enable transparent session id propagation)
+AC_ARG_ENABLE(trans-sid,
+[  --enable-trans-sid      Enable transparent session id propagation],
+[
+  if test "$enableval" = "yes" ; then
+    AC_DEFINE(TRANS_SID, 1)
+	RESULT=yes
+  fi
+])
+AC_MSG_RESULT($RESULT)
 
 PHP_EXTENSION(session)
diff --git a/ext/session/php_session.h b/ext/session/php_session.h
index e54e8e2b85..62948d44d5 100644
--- a/ext/session/php_session.h
+++ b/ext/session/php_session.h
@@ -78,6 +78,7 @@ typedef struct {
 	char *entropy_file;
 	int entropy_length;
 	int lifetime;
+	zend_bool define_sid;
 	ps_module *mod;
 	void *mod_data;
 	HashTable vars;
@@ -141,4 +142,10 @@ typedef struct ps_serializer_struct {
 #define PS_SERIALIZER_ENTRY(x) \
 	{ #x, _ps_srlzr_encode_##x, _ps_srlzr_decode_##x }
 
+#ifdef TRANS_SID
+void session_adapt_uris(const char *, uint, char **, uint *);
+#else
+#define session_adapt_uris(a,b) do { } while(0)
+#endif
+	
 #endif
diff --git a/ext/session/session.c b/ext/session/session.c
index fc727bfa00..36957c1340 100644
--- a/ext/session/session.c
+++ b/ext/session/session.c
@@ -39,6 +39,7 @@
 #include "ext/standard/php3_var.h"
 #include "ext/standard/datetime.h"
 #include "ext/standard/php_lcg.h"
+#include "ext/standard/url_scanner.h"
 
 #ifdef ZTS
 int ps_globals_id;
@@ -504,6 +505,7 @@ static void _php_session_start(PSLS_D)
 	} else {
 		REGISTER_STRING_CONSTANT("SID", empty_string, 0);
 	}
+	PS(define_sid) = define_sid;
 
 	PS(nr_open_sessions)++;
 
@@ -528,7 +530,6 @@ static void _php_session_destroy(PSLS_D)
 	PS(mod)->destroy(&PS(mod_data), PS(id));
 	php_rshutdown_session_globals(PSLS_C);
 	php_rinit_session_globals(PSLS_C);
-	PS(nr_open_sessions)--;
 }
 
 /* {{{ proto string session_name([string newname])
@@ -756,6 +757,22 @@ PHP_FUNCTION(session_destroy)
 }
 /* }}} */
 
+#ifdef TRANS_SID
+void session_adapt_uris(const char *src, uint srclen, char **new, uint *newlen)
+{
+	char *data;
+	size_t len;
+	char buf[512];
+	PSLS_FETCH();
+
+	if(PS(define_sid) && PS(nr_open_sessions) > 0) {
+		snprintf(buf, sizeof(buf), "%s=%s", PS(session_name), PS(id));
+		data = url_adapt(src, srclen, buf, &len);
+		*new = data;
+		*newlen = len;
+	}
+}
+#endif
 
 /* {{{ proto session_unset()
    Unset all registered variables */
@@ -784,6 +801,7 @@ static void php_rinit_session_globals(PSLS_D)
 		_php_find_ps_serializer(INI_STR("session.serialize_handler") PSLS_CC);
 		
 	zend_hash_init(&PS(vars), 0, NULL, NULL, 0);
+	PS(define_sid) = 0;
 	PS(save_path) = estrdup(INI_STR("session.save_path"));
 	PS(session_name) = estrdup(INI_STR("session.name"));
 	PS(entropy_file) = estrdup(INI_STR("session.entropy_file"));
diff --git a/ext/standard/Makefile.am b/ext/standard/Makefile.am
index a0ca94e0aa..db72a28491 100644
--- a/ext/standard/Makefile.am
+++ b/ext/standard/Makefile.am
@@ -8,7 +8,7 @@ libphpext_standard_a_SOURCES=\
 	pack.c pageinfo.c rand.c reg.c soundex.c string.c \
 	syslog.c type.c uniqid.c url.c iptc.c var.c quot_print.c \
 	cyr_convert.c flock_compat.c crypt.c dl.c head.c post.c \
-	parsedate.y lcg.c
+	parsedate.y lcg.c url_scanner.c
 
 #number.o: number.c
 #	$(CC) $(CFLAGS) -w@WARNING_LEVEL@ -c $< -o $@
diff --git a/ext/standard/url_scanner.c b/ext/standard/url_scanner.c
new file mode 100644
index 0000000000..4e191b21cc
--- /dev/null
+++ b/ext/standard/url_scanner.c
@@ -0,0 +1,375 @@
+/* Generated by re2c 0.5 on Sun Sep 12 00:56:23 1999 */
+#line 1 "xx.re"
+
+#include <sys/types.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define YYCTYPE char
+#define YYCURSOR state->crs
+#define YYLIMIT state->end
+#define YYMARKER state->ptr
+#define YYFILL(n)
+
+typedef enum {
+	INITIAL,
+	REF
+} state;
+
+typedef struct {
+	state state;
+	const char *crs;
+	const char *end;
+	const char *ptr;
+	const char *start;
+	char *target;
+	size_t targetsize;
+	const char *data;
+} lexdata;
+
+#define FINISH { catchup(state); goto finish; }
+
+#define BEGIN(x) \
+		switch(state->state) { \
+			case INITIAL: \
+				catchup(state); \
+				break; \
+			case REF: \
+				screw_url(state); \
+				break; \
+		} \
+		state->state = x; \
+		state->start = state->crs; \
+		goto nextiter
+
+#define ATTACH(s, n) { \
+	size_t _newlen = state->targetsize + n; \
+	state->target = realloc(state->target, _newlen + 1); \
+	memcpy(state->target + state->targetsize, s, n); \
+	state->targetsize = _newlen; \
+	state->target[_newlen] = '\0'; \
+	}
+	
+#define URLLEN 512
+	
+static void screw_url(lexdata *state)
+{
+	char *url;
+	int len;
+	char buf[URLLEN];
+	const char *p, *q;
+	char c;
+
+	/* search outer limits for URI */
+	for(p = state->start; p < state->crs && (c = *p); p++)
+		if(c != '"' && c != ' ') break;
+
+	/*  
+	 *  we look at q-1, because q points to the character behind the last
+	 *  character we are going to copy and the decision is based on that last
+	 *  character 
+	 */
+
+	for(q = state->crs; q > state->start && (c = *(q-1)); q--)
+		if(c != '"' && c != ' ') break;
+
+	/* attach beginning */
+	
+	ATTACH(state->start, p-state->start);
+	
+	/* copy old URI */
+	len = q - p;
+	url = malloc(len + 1);
+	memcpy(url, p, len);
+	url[len] = '\0';
+	
+	/* construct new URI */
+	len = snprintf(buf, sizeof(buf), "%s%c%s", url,
+			memchr(state->start, '?', len) ? '&' : '?',
+			state->data);
+	free(url);
+
+	/* attach new URI */
+	ATTACH(buf, len);
+	
+	/* attach rest */
+	ATTACH(q, state->crs - q);
+}
+
+static void catchup(lexdata *state) 
+{
+	ATTACH(state->start, (state->crs - state->start));
+}
+
+#line 115
+
+
+static void url_scanner(lexdata *state)
+{
+	while(state->crs < state->end) {
+	
+	switch(state->state) {
+		case INITIAL: 
+{
+	YYCTYPE yych;
+	unsigned int yyaccept;
+	goto yy0;
+yy1:	++YYCURSOR;
+yy0:
+	if((YYLIMIT - YYCURSOR) < 7) YYFILL(7);
+	yych = *YYCURSOR;
+	switch(yych){
+	case '\000':	goto yy7;
+	case '<':	goto yy2;
+	default:	goto yy4;
+	}
+yy2:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'A':	case 'a':	goto yy9;
+	case 'F':	case 'f':	goto yy10;
+	default:	goto yy3;
+	}
+yy3:yy4:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy5:	switch(yych){
+	case '\000':	case '<':	goto yy6;
+	default:	goto yy4;
+	}
+yy6:
+#line 126
+	{ BEGIN(INITIAL); }
+yy7:	yych = *++YYCURSOR;
+yy8:
+#line 127
+	{ FINISH; }
+yy9:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'H':	case 'h':	goto yy3;
+	default:	goto yy25;
+	}
+yy10:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'R':	case 'r':	goto yy11;
+	default:	goto yy3;
+	}
+yy11:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'A':	case 'a':	goto yy12;
+	default:	goto yy3;
+	}
+yy12:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'M':	case 'm':	goto yy13;
+	default:	goto yy3;
+	}
+yy13:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'E':	case 'e':	goto yy14;
+	default:	goto yy3;
+	}
+yy14:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'S':	case 's':	goto yy3;
+	default:	goto yy16;
+	}
+yy15:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy16:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy15;
+	case 'S':	case 's':	goto yy17;
+	default:	goto yy3;
+	}
+yy17:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'R':	case 'r':	goto yy18;
+	default:	goto yy3;
+	}
+yy18:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'C':	case 'c':	goto yy19;
+	default:	goto yy3;
+	}
+yy19:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy20:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy19;
+	case '=':	goto yy21;
+	default:	goto yy3;
+	}
+yy21:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy22:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy21;
+	default:	goto yy23;
+	}
+yy23:
+#line 124
+	{ BEGIN(REF); }
+yy24:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy25:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy24;
+	case 'H':	case 'h':	goto yy26;
+	default:	goto yy3;
+	}
+yy26:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'R':	case 'r':	goto yy27;
+	default:	goto yy3;
+	}
+yy27:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'E':	case 'e':	goto yy28;
+	default:	goto yy3;
+	}
+yy28:	yych = *++YYCURSOR;
+	switch(yych){
+	case 'F':	case 'f':	goto yy29;
+	default:	goto yy3;
+	}
+yy29:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy30:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy29;
+	case '=':	goto yy31;
+	default:	goto yy3;
+	}
+yy31:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy32:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy31;
+	default:	goto yy33;
+	}
+yy33:
+#line 125
+	{ BEGIN(REF); }
+}
+#line 128
+
+			break;
+		case REF: 
+{
+	YYCTYPE yych;
+	unsigned int yyaccept;
+	goto yy34;
+yy35:	++YYCURSOR;
+yy34:
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+	switch(yych){
+	case '\000':	case '#':	case '>':	goto yy36;
+	case '\t':	case '\v':
+	case '\f':	case ' ':	case '"':	goto yy37;
+	case ':':	goto yy42;
+	default:	goto yy39;
+	}
+yy36:yy37:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy38:	switch(yych){
+	case '\000':	case '"':
+	case '#':	case '>':	goto yy36;
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy37;
+	case ':':	goto yy42;
+	default:	goto yy39;
+	}
+yy39:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy40:	switch(yych){
+	case '\000':	case '>':	goto yy41;
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy48;
+	case '"':	goto yy50;
+	case '#':	goto yy51;
+	case ':':	goto yy42;
+	default:	goto yy39;
+	}
+yy41:
+#line 132
+	{ BEGIN(INITIAL); }
+yy42:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy43:	switch(yych){
+	case '\000':	case '#':	case '>':	goto yy44;
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy45;
+	case '"':	goto yy47;
+	default:	goto yy42;
+	}
+yy44:
+#line 134
+	{ 
+			/* don't modify absolute links */
+			state->state = INITIAL; BEGIN(INITIAL); 
+	}
+yy45:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy46:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy45;
+	case '"':	goto yy47;
+	default:	goto yy44;
+	}
+yy47:	yych = *++YYCURSOR;
+	goto yy44;
+yy48:	++YYCURSOR;
+	if(YYLIMIT == YYCURSOR) YYFILL(1);
+	yych = *YYCURSOR;
+yy49:	switch(yych){
+	case '\t':	case '\v':
+	case '\f':	case ' ':	goto yy48;
+	case '"':	goto yy50;
+	default:	goto yy41;
+	}
+yy50:	yych = *++YYCURSOR;
+	goto yy41;
+yy51:	yych = *++YYCURSOR;
+yy52:	YYCURSOR -= 1;
+#line 133
+	{ BEGIN(INITIAL); }
+}
+#line 138
+
+  			break;
+	}
+nextiter:
+	}
+finish:
+}
+
+char *url_adapt(const char *src, size_t srclen, const char *data, size_t *newlen)
+{
+	lexdata state;
+
+	state.state = INITIAL;
+	state.start = state.crs = src;
+	state.end = src + srclen;
+	state.ptr = NULL;
+	state.target = NULL;
+	state.targetsize = 0;
+	state.data = data;
+
+	url_scanner(&state);
+
+	if(newlen) *newlen = state.targetsize;
+
+	return state.target;
+}
diff --git a/ext/standard/url_scanner.h b/ext/standard/url_scanner.h
new file mode 100644
index 0000000000..4779cc10a2
--- /dev/null
+++ b/ext/standard/url_scanner.h
@@ -0,0 +1,6 @@
+#ifndef URI_SCANNER_H
+#define URI_SCANNER_H
+
+char *url_adapt(const char *src, size_t srclen, const char *data, size_t *newlen);
+
+#endif
diff --git a/ext/standard/url_scanner.re b/ext/standard/url_scanner.re
new file mode 100644
index 0000000000..4436862e7f
--- /dev/null
+++ b/ext/standard/url_scanner.re
@@ -0,0 +1,163 @@
+
+#include <sys/types.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define YYCTYPE char
+#define YYCURSOR state->crs
+#define YYLIMIT state->end
+#define YYMARKER state->ptr
+#define YYFILL(n)
+
+typedef enum {
+	INITIAL,
+	REF
+} state;
+
+typedef struct {
+	state state;
+	const char *crs;
+	const char *end;
+	const char *ptr;
+	const char *start;
+	char *target;
+	size_t targetsize;
+	const char *data;
+} lexdata;
+
+#define FINISH { catchup(state); goto finish; }
+
+#define BEGIN(x) \
+		switch(state->state) { \
+			case INITIAL: \
+				catchup(state); \
+				break; \
+			case REF: \
+				screw_url(state); \
+				break; \
+		} \
+		state->state = x; \
+		state->start = state->crs; \
+		goto nextiter
+
+#define ATTACH(s, n) { \
+	size_t _newlen = state->targetsize + n; \
+	state->target = realloc(state->target, _newlen + 1); \
+	memcpy(state->target + state->targetsize, s, n); \
+	state->targetsize = _newlen; \
+	state->target[_newlen] = '\0'; \
+	}
+	
+#define URLLEN 512
+	
+static void screw_url(lexdata *state)
+{
+	char *url;
+	int len;
+	char buf[URLLEN];
+	const char *p, *q;
+	char c;
+
+	/* search outer limits for URI */
+	for(p = state->start; p < state->crs && (c = *p); p++)
+		if(c != '"' && c != ' ') break;
+
+	/*  
+	 *  we look at q-1, because q points to the character behind the last
+	 *  character we are going to copy and the decision is based on that last
+	 *  character 
+	 */
+
+	for(q = state->crs; q > state->start && (c = *(q-1)); q--)
+		if(c != '"' && c != ' ') break;
+
+	/* attach beginning */
+	
+	ATTACH(state->start, p-state->start);
+	
+	/* copy old URI */
+	len = q - p;
+	url = malloc(len + 1);
+	memcpy(url, p, len);
+	url[len] = '\0';
+	
+	/* construct new URI */
+	len = snprintf(buf, sizeof(buf), "%s%c%s", url,
+			memchr(state->start, '?', len) ? '&' : '?',
+			state->data);
+	free(url);
+
+	/* attach new URI */
+	ATTACH(buf, len);
+	
+	/* attach rest */
+	ATTACH(q, state->crs - q);
+}
+
+static void catchup(lexdata *state) 
+{
+	ATTACH(state->start, (state->crs - state->start));
+}
+
+/*!re2c
+all = [\001-\377];
+eof = [\000];
+ws = [ \t\v\f];
+A = [aA];
+C = [cC];
+E = [eE];
+F = [fF];
+H = [hH];
+M = [mM];
+R = [rR];
+S = [sS];
+*/
+
+static void url_scanner(lexdata *state)
+{
+	while(state->crs < state->end) {
+	
+	switch(state->state) {
+		case INITIAL: 
+/*!re2c
+  	"<" F R A M E ws+ S R C ws* "=" ws*		{ BEGIN(REF); }
+  	"<" A ws+ H R E F ws* "="	ws*			{ BEGIN(REF); }
+	(all\[<])+							{ BEGIN(INITIAL); }
+	eof									{ FINISH; }
+*/
+			break;
+		case REF: 
+/*!re2c
+    ["]? ws* (all\[> \t\v\f":#])+ ws* ["]? { BEGIN(INITIAL); }
+    ["]? ws* (all\[> \t\v\f":#])+/[#]  { BEGIN(INITIAL); }
+    ["]? ws* (all\[> \t\v\f"#])+ ws* ["]? { 
+			/* don't modify absolute links */
+			state->state = INITIAL; BEGIN(INITIAL); 
+	}
+*/
+  			break;
+	}
+nextiter:
+	}
+finish:
+}
+
+char *url_adapt(const char *src, size_t srclen, const char *data, size_t *newlen)
+{
+	lexdata state;
+
+	state.state = INITIAL;
+	state.start = state.crs = src;
+	state.end = src + srclen;
+	state.ptr = NULL;
+	state.target = NULL;
+	state.targetsize = 0;
+	state.data = data;
+
+	url_scanner(&state);
+
+	if(newlen) *newlen = state.targetsize;
+
+	return state.target;
+}
diff --git a/output.c b/output.c
index 7a3037c0cf..3739d41d9e 100644
--- a/output.c
+++ b/output.c
@@ -20,6 +20,7 @@
 
 #include "php.h"
 #include "ext/standard/head.h"
+#include "ext/session/php_session.h"
 #include "SAPI.h"
 
 /* output functions */
@@ -189,17 +190,31 @@ static int zend_ub_body_write_no_header(const char *str, uint str_length)
 
 static int zend_ub_body_write(const char *str, uint str_length)
 {
+	char *newstr = NULL;
+	uint new_length;
+	int result = 0;
 	SLS_FETCH();
 
 	if (SG(request_info).headers_only) {
 		zend_bailout();
 	}
 	if (php3_header()) {
+		session_adapt_uris(str, str_length, &newstr, &new_length);
+		
+		if (newstr) {
+			str = newstr;
+			str_length = new_length;
+		}
+		
 		zend_body_write = zend_ub_body_write_no_header;
-		return zend_header_write(str, str_length);
-	} else {
-		return 0;
+		result = zend_header_write(str, str_length);
+		
+		if (newstr) {
+			free(newstr);
+		}
 	}
+
+	return result;
 }