Implement JIT request decoding support for $_GET and $_POST.

Author: Andrei Zmievski

Zend/zend.c

@@ -195,13 +195,17 @@ ZEND_INI_BEGIN()
 	STD_ZEND_INI_ENTRY("unicode.fallback_encoding",		NULL,	ZEND_INI_ALL, OnUpdateEncoding, fallback_encoding_conv, zend_unicode_globals, unicode_globals)
 	STD_ZEND_INI_ENTRY("unicode.runtime_encoding",		NULL,	ZEND_INI_ALL, OnUpdateEncoding, runtime_encoding_conv, zend_unicode_globals, unicode_globals)
 	STD_ZEND_INI_ENTRY("unicode.script_encoding",		NULL,	ZEND_INI_ALL, OnUpdateEncoding, script_encoding_conv, zend_unicode_globals, unicode_globals)
-	STD_ZEND_INI_ENTRY("unicode.http_input_encoding",	NULL,	ZEND_INI_ALL, OnUpdateEncoding, http_input_encoding_conv, zend_unicode_globals, unicode_globals)
 	STD_ZEND_INI_ENTRY("unicode.filesystem_encoding",	NULL,	ZEND_INI_ALL, OnUpdateEncoding, filesystem_encoding_conv, zend_unicode_globals, unicode_globals)
 	/*
 	 * This is used as a default for the stream contexts. It's not an actual
 	 * UConverter because each stream needs its own.
 	 */
 	STD_ZEND_INI_ENTRY("unicode.stream_encoding", "UTF-8", ZEND_INI_ALL, OnUpdateStringUnempty, stream_encoding, zend_unicode_globals, unicode_globals)
+	/*
+	 * This is used as a default for the request encoding. It's not an actual
+	 * UConverter because the request encoding converter is reset on each request.
+	 */
+	STD_ZEND_INI_ENTRY("unicode.request_encoding_default",	"UTF-8",	ZEND_INI_ALL, OnUpdateStringUnempty, request_encoding_def, zend_unicode_globals, unicode_globals)
 ZEND_INI_END()
 
 
@@ -1008,8 +1012,8 @@ static void unicode_globals_ctor(zend_unicode_globals *unicode_globals TSRMLS_DC
 	unicode_globals->runtime_encoding_conv = NULL;
 	unicode_globals->output_encoding_conv = NULL;
 	unicode_globals->script_encoding_conv = NULL;
-	unicode_globals->http_input_encoding_conv = NULL;
 	unicode_globals->filesystem_encoding_conv = NULL;
+	unicode_globals->request_encoding_conv = NULL;
 	zend_set_converter_encoding(&unicode_globals->utf8_conv, "UTF-8");
 	zend_set_converter_error_mode(unicode_globals->utf8_conv, ZEND_TO_UNICODE, ZEND_CONV_ERROR_STOP);
 	zend_set_converter_encoding(&unicode_globals->ascii_conv, "US-ASCII");
@@ -1061,11 +1065,6 @@ static void unicode_globals_dtor(zend_unicode_globals *unicode_globals TSRMLS_DC
 		unicode_globals->script_encoding_conv != unicode_globals->ascii_conv) {
 		ucnv_close(unicode_globals->script_encoding_conv);
 	}
-	if (unicode_globals->http_input_encoding_conv &&
-		unicode_globals->http_input_encoding_conv != unicode_globals->utf8_conv &&
-		unicode_globals->http_input_encoding_conv != unicode_globals->ascii_conv) {
-		ucnv_close(unicode_globals->http_input_encoding_conv);
-	}
 	if (unicode_globals->utf8_conv) {
 		ucnv_close(unicode_globals->utf8_conv);
 	}
@@ -1385,12 +1384,24 @@ static void init_unicode_request_globals(TSRMLS_D) /* {{{ */
 	UG(default_locale) = safe_estrdup(uloc_getDefault());
 	UG(default_collator) = NULL;
 
+	if (strcmp(UG(request_encoding_def), "binary") != 0) {
+		if (zend_set_converter_encoding(&UG(request_encoding_conv), UG(request_encoding_def)) == FAILURE) {
+			zend_error(E_CORE_ERROR, "Unrecognized encoding '%s' used for request_encoding", UG(request_encoding_def));
+			return;
+		}
+		zend_set_converter_error_mode(UG(request_encoding_conv), ZEND_TO_UNICODE, ZEND_CONV_ERROR_STOP);
+	}
+
 	zend_reset_locale_deps(TSRMLS_C);
 }
 /* }}} */
 
 static void shutdown_unicode_request_globals(TSRMLS_D) /* {{{ */
 {
+	if (UG(request_encoding_conv)) {
+		ucnv_close(UG(request_encoding_conv));
+		UG(request_encoding_conv) = NULL;
+	}
 	zend_collator_destroy(UG(default_collator));
 	efree(UG(default_locale));
 }

Zend/zend_compile.c

@@ -104,11 +104,25 @@ static void build_runtime_defined_function_key(zval *result, zend_uchar type, co
 }
 /* }}} */
 
-int zend_auto_global_arm(zend_auto_global *auto_global TSRMLS_DC) /* {{{ */
+ZEND_API int zend_auto_global_arm(zend_auto_global *auto_global TSRMLS_DC) /* {{{ */
 {
 	auto_global->armed = (auto_global->auto_global_callback ? 1 : 0);
 	return 0;
 }
+
+ZEND_API int zend_auto_global_arm_by_name(const char *name, zend_uint name_len TSRMLS_DC) /* {{{ */
+{
+	zend_auto_global *auto_global;
+
+	if (zend_hash_find(CG(auto_globals), name, name_len+1, (void **) &auto_global)==FAILURE) {
+		return FAILURE;
+	}
+
+	auto_global->armed = 1;
+	CG(auto_globals_cache)[auto_global->index] = NULL;
+
+	return SUCCESS;
+}
 /* }}} */
 
 int zend_auto_global_disable_jit(const char *varname, zend_uint varname_length TSRMLS_DC) /* {{{ */

Zend/zend_compile.h

@@ -625,6 +625,8 @@ ZEND_API int zend_register_auto_global_ex(const char *name, uint name_len, zend_
 ZEND_API zend_bool zend_is_auto_global(const char *name, uint name_len TSRMLS_DC);
 ZEND_API zend_bool zend_u_is_auto_global(zend_uchar type, zstr name, uint name_len TSRMLS_DC);
 ZEND_API zend_bool zend_u_is_auto_global_ex(zend_uchar type, zstr name, uint name_len, zend_bool runtime, zend_auto_global **ret TSRMLS_DC);
+ZEND_API int zend_auto_global_arm(zend_auto_global *auto_global TSRMLS_DC);
+ZEND_API int zend_auto_global_arm_by_name(const char *name, zend_uint name_len TSRMLS_DC);
 ZEND_API int zend_auto_global_disable_jit(const char *varname, zend_uint varname_length TSRMLS_DC);
 ZEND_API size_t zend_dirname(char *path, size_t len);
 ZEND_API size_t zend_u_dirname(UChar *path, size_t len);

Zend/zend_globals.h

@@ -292,15 +292,16 @@ struct _zend_unicode_globals {
 	UConverter *fallback_encoding_conv;  /* converter for default encoding for IS_STRING type */
 	UConverter *runtime_encoding_conv;   /* runtime encoding converter */
 	UConverter *output_encoding_conv;    /* output layer converter */
+	UConverter *request_encoding_conv;   /* http request encoding converter */
 	UConverter *script_encoding_conv;    /* default script encoding converter */
-	UConverter *http_input_encoding_conv;/* http input encoding converter */
 	UConverter *filesystem_encoding_conv;/* default filesystem converter (entries, not contents) */ 
 	UConverter *utf8_conv;				 /* all-purpose UTF-8 converter */
 	UConverter *ascii_conv;				 /* all-purpose ASCII converter */
 
 	char *stream_encoding;		    /* default stream encoding (contents, not FS entries)
 						Uses name of encoding rather than a real converter
 						because each stream needs its own instance */
+	char *request_encoding_def;   /* the default http request encoding */
 
 	uint16_t from_error_mode;
 	UChar from_subst_char[3];

ext/filter/filter.c

@@ -139,7 +139,7 @@ zend_module_entry filter_module_entry = {
 	filter_functions,
 	PHP_MINIT(filter),
 	PHP_MSHUTDOWN(filter),
-	NULL,
+	PHP_RINIT(filter),
 	PHP_RSHUTDOWN(filter),
 	PHP_MINFO(filter),
 	"0.11.0",
@@ -287,6 +287,20 @@ PHP_MSHUTDOWN_FUNCTION(filter)
 }
 /* }}} */
 
+/* {{{ PHP_RINIT_FUNCTION
+ */
+PHP_RINIT_FUNCTION(filter)
+{
+	IF_G(get_array) = NULL;
+	IF_G(post_array) = NULL;
+	IF_G(cookie_array) = NULL;
+	IF_G(server_array) = NULL;
+	IF_G(env_array) = NULL;
+	IF_G(session_array) = NULL;
+	return SUCCESS;
+}
+/* }}} */
+
 /* {{{ PHP_RSHUTDOWN_FUNCTION
  */
 #define VAR_ARRAY_COPY_DTOR(a)   \

ext/mbstring/mb_gpc.c

@@ -326,7 +326,7 @@ enum mbfl_no_encoding _php_mb_encoding_handler_ex(const php_mb_encoding_handler_
 		val = estrndup(val, val_len);
 		if (sapi_module.input_filter(info->data_type, var, &val, val_len, &new_val_len TSRMLS_CC)) {
 			/* add variable to symbol table */
-			php_register_variable_safe(var, val, new_val_len, array_ptr TSRMLS_CC);
+			php_register_variable_safe(IS_STRING, ZSTR(var), ZSTR(val), new_val_len, array_ptr TSRMLS_CC);
 		}
 		efree(val);
 

ext/pcre/php_pcre.c

@@ -996,7 +996,7 @@ static int preg_do_eval(char *eval_str, int eval_str_len, char *subject,
 					match = subject + offsets[backref<<1];
 					match_len = offsets[(backref<<1)+1] - offsets[backref<<1];
 					if (match_len) {
-						esc_match = php_addslashes_ex(match, match_len, &esc_match_len, 0, 1 TSRMLS_CC);
+						esc_match = php_addslashes_ex(match, match_len, &esc_match_len, 0 TSRMLS_CC);
 					} else {
 						esc_match = match;
 						esc_match_len = 0;

ext/standard/basic_functions.c

@@ -861,6 +861,13 @@ ZEND_END_ARG_INFO()
 ZEND_BEGIN_ARG_INFO(arginfo_sys_getloadavg, 0)
 ZEND_END_ARG_INFO()
 #endif
+
+ZEND_BEGIN_ARG_INFO(arginfo_request_set_encoding, 0)
+	ZEND_ARG_INFO(0, encoding)
+ZEND_END_ARG_INFO()
+
+ZEND_BEGIN_ARG_INFO(arginfo_request_had_errors, 0)
+ZEND_END_ARG_INFO()
 /* }}} */
 /* {{{ assert.c */
 ZEND_BEGIN_ARG_INFO(arginfo_assert, 0)
@@ -3364,6 +3371,9 @@ const zend_function_entry basic_functions[] = { /* {{{ */
 
 	PHP_FE(sys_get_temp_dir,												arginfo_sys_get_temp_dir)
 
+	PHP_FE(request_set_encoding,											arginfo_request_set_encoding)
+	PHP_FE(request_had_errors,												arginfo_request_had_errors)
+
 	{NULL, NULL, NULL}
 };
 /* }}} */
@@ -6213,6 +6223,55 @@ PHP_FUNCTION(sys_getloadavg)
 /* }}} */
 #endif
 
+PHP_FUNCTION(request_set_encoding)
+{
+	char *req_enc;
+	int req_enc_len;
+
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &req_enc, &req_enc_len) == FAILURE) {
+		return;
+	}
+
+	if (req_enc_len == 0 || strcmp(req_enc, "binary") == 0) {
+		/* converter is already unset, no need to re-arm, just return */
+		if (UG(request_encoding_conv) == NULL)
+			return;
+		ucnv_close(UG(request_encoding_conv));
+		UG(request_encoding_conv) = NULL;
+	} else {
+		/* if the converter is the same as the requested one, no need to re-arm, just
+		 * return */
+		if (UG(request_encoding_conv)) {
+			UErrorCode status = U_ZERO_ERROR;
+			const char *current = ucnv_getName(UG(request_encoding_conv), &status);
+			if (!ucnv_compareNames(current, req_enc)) {
+				return;
+			}
+		}
+		if (zend_set_converter_encoding(&UG(request_encoding_conv), req_enc) == FAILURE) {
+			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unrecognized encoding '%s'", req_enc);
+			RETURN_FALSE;
+		}
+		zend_set_converter_error_mode(UG(request_encoding_conv), ZEND_TO_UNICODE, ZEND_CONV_ERROR_STOP);
+	}
+
+	zend_auto_global_arm_by_name(ZEND_STRL("_GET") TSRMLS_CC);
+	zend_auto_global_arm_by_name(ZEND_STRL("_POST") TSRMLS_CC);
+	zend_auto_global_arm_by_name(ZEND_STRL("_FILES") TSRMLS_CC);
+	zend_auto_global_arm_by_name(ZEND_STRL("_REQUEST") TSRMLS_CC);
+
+	RETURN_TRUE;
+}
+
+PHP_FUNCTION(request_had_errors)
+{
+	if (zend_parse_parameters_none() == FAILURE) {
+		return;
+	}
+
+	RETURN_BOOL(PG(request_decoding_error));
+}
+
 /*
  * Local variables:
  * tab-width: 4

ext/standard/basic_functions.h

@@ -141,6 +141,10 @@ PHP_FUNCTION(stream_bucket_new);
 PHP_MINIT_FUNCTION(user_filters);
 PHP_RSHUTDOWN_FUNCTION(user_filters);
 
+/* Unicode-related */
+PHP_FUNCTION(request_set_encoding);
+PHP_FUNCTION(request_had_errors);
+
 PHPAPI int _php_error_log(int opt_err, char *message, char *opt, char *headers TSRMLS_DC);
 PHPAPI char *php_get_current_user(void);
 PHPAPI int php_prefix_varname(zval *result, zval *prefix, zstr var_name, int var_name_len, int var_name_type, zend_bool add_underscore TSRMLS_DC);

ext/standard/php_string.h

@@ -129,9 +129,9 @@ PHPAPI UChar *php_u_strtoupper(UChar *s, int *len, const char *locale);
 PHPAPI UChar *php_u_strtolower(UChar *s, int *len, const char *locale);
 PHPAPI char *php_strtr(char *str, int len, char *str_from, char *str_to, int trlen);
 PHPAPI UChar *php_u_addslashes(UChar *str, int length, int *new_length, int freeit TSRMLS_DC);
-PHPAPI UChar *php_u_addslashes_ex(UChar *str, int length, int *new_length, int freeit, int ignore_sybase TSRMLS_DC);
+PHPAPI UChar *php_u_addslashes_ex(UChar *str, int length, int *new_length, int freeit TSRMLS_DC);
 PHPAPI char *php_addslashes(char *str, int length, int *new_length, int freeit TSRMLS_DC);
-PHPAPI char *php_addslashes_ex(char *str, int length, int *new_length, int freeit, int ignore_sybase TSRMLS_DC);
+PHPAPI char *php_addslashes_ex(char *str, int length, int *new_length, int freeit TSRMLS_DC);
 PHPAPI char *php_addcslashes(char *str, int length, int *new_length, int freeit, char *what, int wlength TSRMLS_DC);
 PHPAPI void php_stripslashes(char *str, int *len TSRMLS_DC);
 PHPAPI void php_u_stripslashes(UChar *str, int *len TSRMLS_DC);