Remove the create_invalid_uri_exception URI handler

kocsismate · kocsismate · commit 22fef8b000a8 · 2025-06-07T23:22:33.000+02:00
diff --git a/ext/uri/php_lexbor.c b/ext/uri/php_lexbor.c
@@ -19,6 +19,7 @@
 #include "php_uri_common.h"
 #include "Zend/zend_enum.h"
 #include "Zend/zend_smart_str.h"
+#include "Zend/zend_exceptions.h"
 #ifdef HAVE_ARPA_INET_H
 #include <arpa/inet.h>
 #endif
@@ -225,6 +226,30 @@ static void fill_errors(zval *errors)
 	}
 }
 
+static void throw_invalid_url_exception(zval *errors)
+{
+	ZEND_ASSERT(errors != NULL && Z_TYPE_P(errors) == IS_ARRAY);
+
+	zval exception;
+
+	object_init_ex(&exception, uri_whatwg_invalid_url_exception_ce);
+
+	zval value;
+	ZVAL_STRING(&value, "URL parsing failed");
+	zend_update_property_ex(uri_whatwg_invalid_url_exception_ce, Z_OBJ(exception), ZSTR_KNOWN(ZEND_STR_MESSAGE), &value);
+	zval_ptr_dtor_str(&value);
+
+	zend_update_property(uri_whatwg_invalid_url_exception_ce, Z_OBJ(exception), ZEND_STRL("errors"), errors);
+
+	zend_throw_exception_object(&exception);
+}
+
+static void throw_invalid_url_exception_during_write(zval *errors)
+{
+	fill_errors(errors);
+	throw_invalid_url_exception(errors);
+}
+
 static lxb_status_t lexbor_serialize_callback(const lxb_char_t *data, size_t length, void *ctx)
 {
 	smart_str *uri_str = ctx;
@@ -255,7 +280,7 @@ static zend_result lexbor_write_scheme(struct uri_internal_t *internal_uri, zval
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_protocol_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -284,7 +309,7 @@ static zend_result lexbor_write_username(uri_internal_t *internal_uri, zval *val
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_username_set(lexbor_uri, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -313,7 +338,7 @@ static zend_result lexbor_write_password(struct uri_internal_t *internal_uri, zv
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_password_set(lexbor_uri, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -385,7 +410,7 @@ static zend_result lexbor_write_host(struct uri_internal_t *internal_uri, zval *
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_hostname_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -414,7 +439,7 @@ static zend_result lexbor_write_port(struct uri_internal_t *internal_uri, zval *
 	zval_long_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_port_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -443,7 +468,7 @@ static zend_result lexbor_write_path(struct uri_internal_t *internal_uri, zval *
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_pathname_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -472,7 +497,7 @@ static zend_result lexbor_write_query(struct uri_internal_t *internal_uri, zval
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_search_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -501,7 +526,7 @@ static zend_result lexbor_write_fragment(struct uri_internal_t *internal_uri, zv
 	zval_string_or_null_to_lexbor_str(value, &str);
 
 	if (lxb_url_api_hash_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) {
-		fill_errors(errors);
+		throw_invalid_url_exception_during_write(errors);
 
 		return FAILURE;
 	}
@@ -538,29 +563,24 @@ void lexbor_request_shutdown(void)
 	lexbor_urls = 0;
 }
 
-static void *lexbor_parse_uri(const zend_string *uri_str, const void *base_url, zval *errors)
+lxb_url_t *lexbor_parse_uri_ex(const zend_string *uri_str, const void *base_url, zval *errors, bool silent)
 {
 	lexbor_cleanup_parser();
 
 	const lxb_url_t *lexbor_base_url = base_url;
 	lxb_url_t *url = lxb_url_parse(&lexbor_parser, lexbor_base_url, (unsigned char *) ZSTR_VAL(uri_str), ZSTR_LEN(uri_str));
 	fill_errors(errors);
 
+	if (url == NULL && !silent) {
+		throw_invalid_url_exception(errors);
+	}
+
 	return url;
 }
 
-static void lexbor_create_invalid_uri_exception(zval *exception_zv, zval *errors)
+static void *lexbor_parse_uri(const zend_string *uri_str, const void *base_url, zval *errors, bool silent)
 {
-	ZEND_ASSERT(Z_TYPE_P(errors) == IS_ARRAY && "Lexbor always returns an array of errors when URI parsing fails");
-
-	object_init_ex(exception_zv, uri_whatwg_invalid_url_exception_ce);
-
-	zval value;
-	ZVAL_STRING(&value, "URL parsing failed");
-	zend_update_property_ex(uri_whatwg_invalid_url_exception_ce, Z_OBJ_P(exception_zv), ZSTR_KNOWN(ZEND_STR_MESSAGE), &value);
-	zval_ptr_dtor_str(&value);
-
-	zend_update_property(uri_whatwg_invalid_url_exception_ce, Z_OBJ_P(exception_zv), ZEND_STRL("errors"), errors);
+	return lexbor_parse_uri_ex(uri_str, base_url, errors, silent);
 }
 
 static void *lexbor_clone_uri(void *uri)
@@ -603,7 +623,6 @@ static void lexbor_free_uri(void *uri)
 const uri_handler_t lexbor_uri_handler = {
 	.name = URI_PARSER_WHATWG,
 	.parse_uri = lexbor_parse_uri,
-	.create_invalid_uri_exception = lexbor_create_invalid_uri_exception,
 	.clone_uri = lexbor_clone_uri,
 	.uri_to_string = lexbor_uri_to_string,
 	.free_uri = lexbor_free_uri,
diff --git a/ext/uri/php_lexbor.h b/ext/uri/php_lexbor.h
@@ -22,6 +22,8 @@
 
 extern const uri_handler_t lexbor_uri_handler;
 
+lxb_url_t *lexbor_parse_uri_ex(const zend_string *uri_str, const void *base_url, zval *errors, bool silent);
+
 void lexbor_module_init(void);
 void lexbor_module_shutdown(void);
 
diff --git a/ext/uri/php_uri.c b/ext/uri/php_uri.c
@@ -209,11 +209,9 @@ PHPAPI void php_uri_instantiate_uri(
 		base_url = internal_base_url->uri;
 	}
 
-	void *uri = handler->parse_uri(uri_str, base_url, should_throw || errors_zv != NULL ? &errors : NULL);
+	void *uri = handler->parse_uri(uri_str, base_url, should_throw || errors_zv != NULL ? &errors : NULL, !should_throw);
 	if (UNEXPECTED(uri == NULL)) {
 		if (should_throw) {
-			throw_invalid_uri_exception(handler, &errors);
-			zval_ptr_dtor(&errors);
 			RETURN_THROWS();
 		} else {
 			if (pass_errors_by_ref_and_free(errors_zv, &errors) == FAILURE) {
@@ -352,7 +350,7 @@ static void uri_unserialize(INTERNAL_FUNCTION_PARAMETERS, const char *handler_na
 	if (internal_uri->uri != NULL) {
 		internal_uri->handler->free_uri(internal_uri->uri);
 	}
-	internal_uri->uri = internal_uri->handler->parse_uri(Z_STR_P(uri_zv), NULL, NULL);
+	internal_uri->uri = internal_uri->handler->parse_uri(Z_STR_P(uri_zv), NULL, NULL, true);
 	if (internal_uri->uri == NULL) {
 		zend_throw_exception_ex(NULL, 0, "Invalid serialization data for %s object", ZSTR_VAL(object->ce->name));
 		RETURN_THROWS();
@@ -617,7 +615,6 @@ zend_result uri_handler_register(const uri_handler_t *uri_handler)
 
 	ZEND_ASSERT(uri_handler->name != NULL);
 	ZEND_ASSERT(uri_handler->parse_uri != NULL);
-	ZEND_ASSERT(uri_handler->create_invalid_uri_exception != NULL);
 	ZEND_ASSERT(uri_handler->clone_uri != NULL);
 	ZEND_ASSERT(uri_handler->uri_to_string != NULL);
 	ZEND_ASSERT(uri_handler->free_uri != NULL);
diff --git a/ext/uri/php_uri_common.c b/ext/uri/php_uri_common.c
@@ -65,15 +65,6 @@ static zend_string *get_known_string_by_property_name(uri_property_name_t proper
 	}
 }
 
-void throw_invalid_uri_exception(const uri_handler_t *uri_handler, zval *errors)
-{
-	zval exception_zv;
-
-	uri_handler->create_invalid_uri_exception(&exception_zv, errors);
-
-	zend_throw_exception_object(&exception_zv);
-}
-
 void uri_read_component(INTERNAL_FUNCTION_PARAMETERS, uri_property_name_t property_name, uri_component_read_mode_t component_read_mode)
 {
 	ZEND_PARSE_PARAMETERS_NONE();
@@ -117,7 +108,6 @@ static void uri_write_component_ex(INTERNAL_FUNCTION_PARAMETERS, uri_property_na
 	zval errors;
 	ZVAL_UNDEF(&errors);
 	if (property_handler->write_func(new_internal_uri, property_zv, &errors) == FAILURE) {
-		throw_invalid_uri_exception(new_internal_uri->handler, &errors);
 		zval_ptr_dtor(&errors);
 		zend_object_release(new_object);
 		RETURN_THROWS();
diff --git a/ext/uri/php_uri_common.h b/ext/uri/php_uri_common.h
@@ -83,19 +83,16 @@ typedef struct uri_handler_t {
 	 * returned.
 	 *
 	 * The errors by-ref parameter can contain errors that occurred during parsing.
-	 * If the input value is NULL, or there were no errors, errors should not be changed.
+	 * If the input value is NULL, or there were no errors, the errors parameter should
+	 * not be modified.
 	 *
 	 * If the URI string is valid and the base_url URI is not NULL, the URI object
 	 * is resolved against the base_url.
-	 */
-	void *(*parse_uri)(const zend_string *uri_str, const void *base_url, zval *errors);
-	/**
-	 * Create a Uri\InvalidUriException instance based on the errors parameter.
-	 * The errors parameter is either an array or an UNDEF zval.
 	 *
-	 * The exception object is passed by ref to the exception_zv parameter.
+	 * If the silent parameter is true, a Uri\InvalidUriException instance must be thrown.
+	 * If the parameter is false, the possible errors should be handled by the caller.
 	 */
-	void (*create_invalid_uri_exception)(zval *exception_zv, zval *errors);
+	void *(*parse_uri)(const zend_string *uri_str, const void *base_url, zval *errors, bool silent);
 	void *(*clone_uri)(void *uri);
 	zend_string *(*uri_to_string)(void *uri, uri_recomposition_mode_t recomposition_mode, bool exclude_fragment);
 	void (*free_uri)(void *uri);
@@ -129,7 +126,6 @@ static inline uri_internal_t *uri_internal_from_obj(const zend_object *object) {
 
 zend_result uri_handler_register(const uri_handler_t *uri_handler);
 const uri_property_handler_t *uri_property_handler_from_internal_uri(const uri_internal_t *internal_uri, uri_property_name_t property_name);
-void throw_invalid_uri_exception(const uri_handler_t *uri_handler, zval *errors);
 void uri_read_component(INTERNAL_FUNCTION_PARAMETERS, uri_property_name_t property_name, uri_component_read_mode_t component_read_mode);
 void uri_write_component_str(INTERNAL_FUNCTION_PARAMETERS, uri_property_name_t property_name);
 void uri_write_component_str_or_null(INTERNAL_FUNCTION_PARAMETERS, uri_property_name_t property_name);
