Enable users to set the HTML tags to rewrite

through a configuration directive

ext/standard/basic_functions.c

@@ -718,6 +718,9 @@ PHP_MINIT_FUNCTION(basic)
 	PHP_MINIT(syslog)(INIT_FUNC_ARGS_PASSTHRU);
 	PHP_MINIT(array)(INIT_FUNC_ARGS_PASSTHRU);
 	PHP_MINIT(assert)(INIT_FUNC_ARGS_PASSTHRU);
+#ifdef TRANS_SID
+	PHP_MINIT(url_scanner_ex)(INIT_FUNC_ARGS_PASSTHRU);
+#endif
 
 	if(PG(allow_url_fopen)) {
 		if(FAILURE==php_register_url_wrapper("http",php_fopen_url_wrap_http)) {
@@ -759,7 +762,7 @@ PHP_MSHUTDOWN_FUNCTION(basic)
 	PHP_MSHUTDOWN(browscap)(SHUTDOWN_FUNC_ARGS_PASSTHRU);
 	PHP_MSHUTDOWN(array)(SHUTDOWN_FUNC_ARGS_PASSTHRU);
 	PHP_MSHUTDOWN(assert)(SHUTDOWN_FUNC_ARGS_PASSTHRU);
-
+	PHP_MSHUTDOWN(url_scanner_ex)(SHUTDOWN_FUNC_ARGS_PASSTHRU);
 
 	return SUCCESS;	
 }

ext/standard/url_scanner_ex.c

@@ -1,4 +1,4 @@
-/* Generated by re2c 0.5 on Wed Oct  4 20:04:55 2000 */
+/* Generated by re2c 0.5 on Mon Oct 16 00:15:52 2000 */
 #line 1 "/home/sas/src/php4/ext/standard/url_scanner_ex.re"
 /*
   +----------------------------------------------------------------------+
@@ -30,6 +30,7 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "php_ini.h"
 #include "php_globals.h"
 #define STATE_TAG SOME_OTHER_STATE_TAG
 #include "basic_functions.h"
@@ -40,6 +41,54 @@
 
 #include "php_smart_str.h"
 
+static PHP_INI_MH(OnUpdateTags)
+{
+	url_adapt_state_ex_t *ctx;
+	char *key;
+	char *lasts;
+	char *tmp;
+	BLS_FETCH();
+	
+	ctx = &BG(url_adapt_state_ex);
+	
+	tmp = estrndup(new_value, new_value_length);
+	
+	if (ctx->tags)
+		zend_hash_destroy(ctx->tags);
+	else
+		ctx->tags = malloc(sizeof(HashTable));
+	
+	zend_hash_init(ctx->tags, 0, NULL, NULL, 1);
+	
+	for (key = php_strtok_r(tmp, ",", &lasts);
+			key;
+			key = php_strtok_r(NULL, ",", &lasts)) {
+		char *val;
+
+		val = strchr(key, '=');
+		if (val) {
+			char *q;
+			int keylen;
+			
+			*val++ = '\0';
+			for (q = key; *q; q++)
+				*q = tolower(*q);
+			keylen = q - key;
+			/* key is stored withOUT NUL
+			   val is stored WITH    NUL */
+			zend_hash_add(ctx->tags, key, keylen, val, strlen(val)+1, NULL);
+		}
+	}
+
+	efree(tmp);
+
+	return SUCCESS;
+}
+
+PHP_INI_BEGIN()
+	STD_PHP_INI_ENTRY("url_rewriter.tags", "a=href,area=href,frame=src,form=fakeentry", PHP_INI_ALL, OnUpdateTags, url_adapt_state_ex, php_basic_globals, basic_globals)
+PHP_INI_END()
+
 static inline void append_modified_url(smart_str *url, smart_str *dest, smart_str *name, smart_str *val, const char *separator)
 {
 	register const char *p, *q;
@@ -62,6 +111,12 @@ static inline void append_modified_url(smart_str *url, smart_str *dest, smart_st
 		}
 	}
 
+	/* Don't modify URLs of the format "#mark" */
+	if (bash - url->c == 0) {
+		smart_str_append(dest, url);
+		return;
+	}
+
 	if (bash)
 		smart_str_appendl(dest, url->c, bash - url->c);
 	else
@@ -76,47 +131,20 @@ static inline void append_modified_url(smart_str *url, smart_str *dest, smart_st
 		smart_str_appendl(dest, bash, q - bash);
 }
 
-struct php_tag_arg {
-	char *tag;
-	int taglen;
-	char *arg;
-	int arglen;
-};
-
-#define TAG_ARG_ENTRY(a,b) {#a,sizeof(#a)-1,#b,sizeof(#b)-1},
-
-static struct php_tag_arg check_tag_arg[] = {
-	TAG_ARG_ENTRY(a, href)
-	TAG_ARG_ENTRY(area, href)
-	TAG_ARG_ENTRY(frame, src)
-	TAG_ARG_ENTRY(img, src)
-	TAG_ARG_ENTRY(input, src)
-	TAG_ARG_ENTRY(form, fake_entry_for_passing_on_form_tag)
-	{0}
-};
-
 static inline void tag_arg(url_adapt_state_ex_t *ctx PLS_DC)
 {
 	char f = 0;
-	int i;
 
-	for (i = 0; check_tag_arg[i].tag; i++) {
+	if (strncasecmp(ctx->arg.c, ctx->lookup_data, ctx->arg.len) == 0)
-		if (check_tag_arg[i].arglen == ctx->arg.len
-				&& check_tag_arg[i].taglen == ctx->tag.len
-				&& strncasecmp(ctx->tag.c, check_tag_arg[i].tag, ctx->tag.len) == 0
-				&& strncasecmp(ctx->arg.c, check_tag_arg[i].arg, ctx->arg.len) == 0) {
 		f = 1;
-			break;
-		}
-	}
 
-	smart_str_appends(&ctx->result, "\"");
+	smart_str_appendc(&ctx->result, '"');
 	if (f) {
 		append_modified_url(&ctx->val, &ctx->result, &ctx->q_name, &ctx->q_value, PG(arg_separator));
 	} else {
 		smart_str_append(&ctx->result, &ctx->val);
 	}
-	smart_str_appends(&ctx->result, "\"");
+	smart_str_appendc(&ctx->result, '"');
 }
 
 enum {
@@ -160,13 +188,10 @@ enum {
 	int ok = 0; \
 	int i; \
 	smart_str_copyl(&ctx->tag, start, YYCURSOR - start); \
-	for (i = 0; check_tag_arg[i].tag; i++) { \
+	for (i = 0; i < ctx->tag.len; i++) \
-		if (ctx->tag.len == check_tag_arg[i].taglen \
+		ctx->tag.c[i] = tolower(ctx->tag.c[i]); \
-				&& strncasecmp(ctx->tag.c, check_tag_arg[i].tag, ctx->tag.len) == 0) { \
+	if (zend_hash_find(ctx->tags, ctx->tag.c, ctx->tag.len, &ctx->lookup_data) == SUCCESS) \
 		ok = 1; \
-			break; \
-		} \
-	} \
 	STATE = ok ? STATE_NEXT_ARG : STATE_PLAIN; \
 }
 
@@ -197,7 +222,7 @@ static inline void mainloop(url_adapt_state_ex_t *ctx, const char *newdata, size
 	YYCURSOR = ctx->buf.c;
 	YYLIMIT = ctx->buf.c + ctx->buf.len;
 
-#line 201
+#line 226
 
 	
 	while(1) {
@@ -217,14 +242,14 @@ yy0:
 	if(yych != '<')	goto yy4;
 yy2:	yych = *++YYCURSOR;
 yy3:
-#line 210
+#line 235
 	{ PASSTHRU(); STATE = STATE_TAG; continue; }
 yy4:	yych = *++YYCURSOR;
 yy5:
-#line 211
+#line 236
 	{ PASSTHRU(); continue; }
 }
-#line 212
+#line 237
 
 			break;
 			
@@ -278,11 +303,11 @@ yy6:
 yy8:	yych = *++YYCURSOR;
 	goto yy13;
 yy9:
-#line 217
+#line 242
 	{ HANDLE_TAG() /* Sets STATE */; PASSTHRU(); continue; }
 yy10:	yych = *++YYCURSOR;
 yy11:
-#line 218
+#line 243
 	{ PASSTHRU(); STATE = STATE_PLAIN; continue; }
 yy12:	++YYCURSOR;
 	if(YYLIMIT == YYCURSOR) YYFILL(1);
@@ -290,7 +315,7 @@ yy12:	++YYCURSOR;
 yy13:	if(yybm[0+yych] & 128)	goto yy12;
 	goto yy9;
 }
-#line 219
+#line 244
 
   			break;
 			
@@ -324,22 +349,22 @@ yy14:
 	}
 yy16:	yych = *++YYCURSOR;
 yy17:
-#line 224
+#line 249
 	{ PASSTHRU(); HANDLE_FORM(); STATE = STATE_PLAIN; continue; }
 yy18:	yych = *++YYCURSOR;
 yy19:
-#line 225
+#line 250
 	{ PASSTHRU(); continue; }
 yy20:	yych = *++YYCURSOR;
 yy21:
-#line 226
+#line 251
 	{ YYCURSOR--; STATE = STATE_ARG; continue; }
 yy22:	yych = *++YYCURSOR;
 yy23:
-#line 227
+#line 252
 	{ PASSTHRU(); continue; }
 }
-#line 228
+#line 253
 
  	 		break;
 
@@ -393,11 +418,11 @@ yy24:
 yy26:	yych = *++YYCURSOR;
 	goto yy31;
 yy27:
-#line 233
+#line 258
 	{ PASSTHRU(); HANDLE_ARG(); STATE = STATE_BEFORE_VAL; continue; }
 yy28:	yych = *++YYCURSOR;
 yy29:
-#line 234
+#line 259
 	{ PASSTHRU(); STATE = STATE_NEXT_ARG; continue; }
 yy30:	++YYCURSOR;
 	if(YYLIMIT == YYCURSOR) YYFILL(1);
@@ -405,7 +430,7 @@ yy30:	++YYCURSOR;
 yy31:	if(yybm[0+yych] & 128)	goto yy30;
 	goto yy27;
 }
-#line 235
+#line 260
 
 
 		case STATE_BEFORE_VAL:
@@ -459,12 +484,12 @@ yy34:	yyaccept = 0;
 	if(yych == ' ')	goto yy41;
 	if(yych == '=')	goto yy39;
 yy35:
-#line 240
+#line 265
 	{ YYCURSOR--; STATE = STATE_NEXT_ARG; continue; }
 yy36:	yych = *++YYCURSOR;
 	goto yy40;
 yy37:
-#line 239
+#line 264
 	{ PASSTHRU(); STATE = STATE_VAL; continue; }
 yy38:	yych = *++YYCURSOR;
 	goto yy35;
@@ -483,7 +508,7 @@ yy43:	YYCURSOR = YYMARKER;
 	case 0:	goto yy35;
 	}
 }
-#line 241
+#line 266
 
 			break;
 
@@ -552,13 +577,13 @@ yy46:	yyaccept = 0;
 	yych = *(YYMARKER = ++YYCURSOR);
 	if(yych != '>')	goto yy63;
 yy47:
-#line 249
+#line 274
 	{ PASSTHRU(); STATE = STATE_NEXT_ARG; continue; }
 yy48:	yyaccept = 1;
 	yych = *(YYMARKER = ++YYCURSOR);
 	goto yy55;
 yy49:
-#line 248
+#line 273
 	{ HANDLE_VAL(0); STATE = STATE_NEXT_ARG; continue; }
 yy50:	yych = *++YYCURSOR;
 	goto yy53;
@@ -579,7 +604,7 @@ yy55:	if(yybm[0+yych] & 32)	goto yy54;
 yy56:	yych = *++YYCURSOR;
 	if(yybm[0+yych] & 16)	goto yy52;
 yy57:
-#line 247
+#line 272
 	{ HANDLE_VAL(1); STATE = STATE_NEXT_ARG; continue; }
 yy58:	++YYCURSOR;
 	if(YYLIMIT == YYCURSOR) YYFILL(1);
@@ -600,10 +625,10 @@ yy63:	if(yybm[0+yych] & 128)	goto yy62;
 	if(yych >= '>')	goto yy60;
 yy64:	yych = *++YYCURSOR;
 yy65:
-#line 246
+#line 271
 	{ HANDLE_VAL(1); STATE = STATE_NEXT_ARG; continue; }
 }
-#line 250
+#line 275
 
 			break;
 	}
@@ -650,7 +675,7 @@ PHP_RINIT_FUNCTION(url_scanner)
 	
 	ctx = &BG(url_adapt_state_ex);
 
-	memset(ctx, 0, sizeof(*ctx));
+	memset(ctx, 0, ((size_t) &((url_adapt_state_ex_t *)0)->tags));
 
 	return SUCCESS;
 }
@@ -670,4 +695,18 @@ PHP_RSHUTDOWN_FUNCTION(url_scanner)
 	return SUCCESS;
 }
 
+PHP_MINIT_FUNCTION(url_scanner)
+{
+	REGISTER_INI_ENTRIES();
+	return SUCCESS;
+}
+
+PHP_MSHUTDOWN_FUNCTION(url_scanner)
+{
+	UNREGISTER_INI_ENTRIES();
+	zend_hash_destroy(BG(url_adapt_state_ex).tags);
+	free(BG(url_adapt_state_ex).tags);
+	return SUCCESS;
+}
+
 #endif

ext/standard/url_scanner_ex.h

@@ -19,6 +19,8 @@
 #ifndef URL_SCANNER_EX_H
 #define URL_SCANNER_EX_H
 
+PHP_MINIT_FUNCTION(url_scanner_ex);
+PHP_MSHUTDOWN_FUNCTION(url_scanner_ex);
 PHP_RSHUTDOWN_FUNCTION(url_scanner_ex);
 PHP_RINIT_FUNCTION(url_scanner_ex);
 
@@ -40,7 +42,11 @@ typedef struct {
 	smart_str q_name;
 	smart_str q_value;
 
+	char *lookup_data;
 	int state;
+	
+	/* Everything above is zeroed in RINIT */
+	HashTable *tags;
 } url_adapt_state_ex_t;
 
 #endif

ext/standard/url_scanner_ex.re

@@ -28,6 +28,7 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "php_ini.h"
 #include "php_globals.h"
 #define STATE_TAG SOME_OTHER_STATE_TAG
 #include "basic_functions.h"
@@ -38,6 +39,54 @@
 
 #include "php_smart_str.h"
 
+static PHP_INI_MH(OnUpdateTags)
+{
+	url_adapt_state_ex_t *ctx;
+	char *key;
+	char *lasts;
+	char *tmp;
+	BLS_FETCH();
+	
+	ctx = &BG(url_adapt_state_ex);
+	
+	tmp = estrndup(new_value, new_value_length);
+	
+	if (ctx->tags)
+		zend_hash_destroy(ctx->tags);
+	else
+		ctx->tags = malloc(sizeof(HashTable));
+	
+	zend_hash_init(ctx->tags, 0, NULL, NULL, 1);
+	
+	for (key = php_strtok_r(tmp, ",", &lasts);
+			key;
+			key = php_strtok_r(NULL, ",", &lasts)) {
+		char *val;
+
+		val = strchr(key, '=');
+		if (val) {
+			char *q;
+			int keylen;
+			
+			*val++ = '\0';
+			for (q = key; *q; q++)
+				*q = tolower(*q);
+			keylen = q - key;
+			/* key is stored withOUT NUL
+			   val is stored WITH    NUL */
+			zend_hash_add(ctx->tags, key, keylen, val, strlen(val)+1, NULL);
+		}
+	}
+
+	efree(tmp);
+
+	return SUCCESS;
+}
+
+PHP_INI_BEGIN()
+	STD_PHP_INI_ENTRY("url_rewriter.tags", "a=href,area=href,frame=src,form=fakeentry", PHP_INI_ALL, OnUpdateTags, url_adapt_state_ex, php_basic_globals, basic_globals)
+PHP_INI_END()
+
 static inline void append_modified_url(smart_str *url, smart_str *dest, smart_str *name, smart_str *val, const char *separator)
 {
 	register const char *p, *q;
@@ -60,6 +109,12 @@ static inline void append_modified_url(smart_str *url, smart_str *dest, smart_st
 		}
 	}
 
+	/* Don't modify URLs of the format "#mark" */
+	if (bash - url->c == 0) {
+		smart_str_append(dest, url);
+		return;
+	}
+
 	if (bash)
 		smart_str_appendl(dest, url->c, bash - url->c);
 	else
@@ -74,47 +129,20 @@ static inline void append_modified_url(smart_str *url, smart_str *dest, smart_st
 		smart_str_appendl(dest, bash, q - bash);
 }
 
-struct php_tag_arg {
-	char *tag;
-	int taglen;
-	char *arg;
-	int arglen;
-};
-
-#define TAG_ARG_ENTRY(a,b) {#a,sizeof(#a)-1,#b,sizeof(#b)-1},
-
-static struct php_tag_arg check_tag_arg[] = {
-	TAG_ARG_ENTRY(a, href)
-	TAG_ARG_ENTRY(area, href)
-	TAG_ARG_ENTRY(frame, src)
-	TAG_ARG_ENTRY(img, src)
-	TAG_ARG_ENTRY(input, src)
-	TAG_ARG_ENTRY(form, fake_entry_for_passing_on_form_tag)
-	{0}
-};
-
 static inline void tag_arg(url_adapt_state_ex_t *ctx PLS_DC)
 {
 	char f = 0;
-	int i;
 
-	for (i = 0; check_tag_arg[i].tag; i++) {
+	if (strncasecmp(ctx->arg.c, ctx->lookup_data, ctx->arg.len) == 0)
-		if (check_tag_arg[i].arglen == ctx->arg.len
-				&& check_tag_arg[i].taglen == ctx->tag.len
-				&& strncasecmp(ctx->tag.c, check_tag_arg[i].tag, ctx->tag.len) == 0
-				&& strncasecmp(ctx->arg.c, check_tag_arg[i].arg, ctx->arg.len) == 0) {
 		f = 1;
-			break;
-		}
-	}
 
-	smart_str_appends(&ctx->result, "\"");
+	smart_str_appendc(&ctx->result, '"');
 	if (f) {
 		append_modified_url(&ctx->val, &ctx->result, &ctx->q_name, &ctx->q_value, PG(arg_separator));
 	} else {
 		smart_str_append(&ctx->result, &ctx->val);
 	}
-	smart_str_appends(&ctx->result, "\"");
+	smart_str_appendc(&ctx->result, '"');
 }
 
 enum {
@@ -158,13 +186,10 @@ enum {
 	int ok = 0; \
 	int i; \
 	smart_str_copyl(&ctx->tag, start, YYCURSOR - start); \
-	for (i = 0; check_tag_arg[i].tag; i++) { \
+	for (i = 0; i < ctx->tag.len; i++) \
-		if (ctx->tag.len == check_tag_arg[i].taglen \
+		ctx->tag.c[i] = tolower(ctx->tag.c[i]); \
-				&& strncasecmp(ctx->tag.c, check_tag_arg[i].tag, ctx->tag.len) == 0) { \
+	if (zend_hash_find(ctx->tags, ctx->tag.c, ctx->tag.len, &ctx->lookup_data) == SUCCESS) \
 		ok = 1; \
-			break; \
-		} \
-	} \
 	STATE = ok ? STATE_NEXT_ARG : STATE_PLAIN; \
 }
 
@@ -293,7 +318,7 @@ PHP_RINIT_FUNCTION(url_scanner)
 	
 	ctx = &BG(url_adapt_state_ex);
 
-	memset(ctx, 0, sizeof(*ctx));
+	memset(ctx, 0, ((size_t) &((url_adapt_state_ex_t *)0)->tags));
 
 	return SUCCESS;
 }
@@ -313,4 +338,18 @@ PHP_RSHUTDOWN_FUNCTION(url_scanner)
 	return SUCCESS;
 }
 
+PHP_MINIT_FUNCTION(url_scanner)
+{
+	REGISTER_INI_ENTRIES();
+	return SUCCESS;
+}
+
+PHP_MSHUTDOWN_FUNCTION(url_scanner)
+{
+	UNREGISTER_INI_ENTRIES();
+	zend_hash_destroy(BG(url_adapt_state_ex).tags);
+	free(BG(url_adapt_state_ex).tags);
+	return SUCCESS;
+}
+
 #endif

php.ini-dist

@@ -450,6 +450,7 @@ session.cache_limiter     = nocache ; set to {nocache,private,public} to
 session.cache_expire      = 180     ; document expires after n minutes
 session.use_trans_sid     = 1       ; use transient sid support if enabled
                                     ; by compiling with --enable-trans-sid
+url_rewriter.tags         = a=href,area=href,frame=src,input=src,form=fakeentry
 
 [MSSQL]
 ;extension=php_mssql.dll