Some cleanups

php · Nov 18, 2024 · 8e21e67 · 8e21e67
1 parent 2ac633f
commit 8e21e67
Show file tree

Hide file tree

Showing 32 changed files with 415 additions and 168 deletions.
diff --git a/ext/filter/tests/062.phpt b/ext/filter/tests/062.phpt
@@ -5,7 +5,7 @@ filter
 --FILE--
 <?php
 
-function validateUrls()
+function validateUrls(string $parserName)
 {
     $values = [
         'http://example.com/index.html',
@@ -53,27 +53,25 @@ function validateUrls()
     ];
 
     foreach ($values as $value) {
-        var_dump(filter_var($value, FILTER_VALIDATE_URL));
+        var_dump(filter_var($value, FILTER_VALIDATE_URL, ["parser" => $parserName]));
     }
 
-    var_dump(filter_var("qwe", FILTER_VALIDATE_URL));
-    var_dump(filter_var("http://qwe", FILTER_VALIDATE_URL));
-    var_dump(filter_var("http://", FILTER_VALIDATE_URL));
-    var_dump(filter_var("/tmp/test", FILTER_VALIDATE_URL));
-    var_dump(filter_var("http://www.example.com", FILTER_VALIDATE_URL));
-    var_dump(filter_var("http://www.example.com", FILTER_VALIDATE_URL, FILTER_FLAG_PATH_REQUIRED));
-    var_dump(filter_var("http://www.example.com/path/at/the/server/", FILTER_VALIDATE_URL, FILTER_FLAG_PATH_REQUIRED));
-    var_dump(filter_var("http://www.example.com/index.html", FILTER_VALIDATE_URL, FILTER_FLAG_QUERY_REQUIRED));
-    var_dump(filter_var("http://www.example.com/index.php?a=b&c=d", FILTER_VALIDATE_URL, FILTER_FLAG_QUERY_REQUIRED));
+    var_dump(filter_var("qwe", FILTER_VALIDATE_URL, ["parser" => $parserName]));
+    var_dump(filter_var("http://qwe", FILTER_VALIDATE_URL, ["parser" => $parserName]));
+    var_dump(filter_var("http://", FILTER_VALIDATE_URL, ["parser" => $parserName]));
+    var_dump(filter_var("/tmp/test", FILTER_VALIDATE_URL, ["parser" => $parserName]));
+    var_dump(filter_var("http://www.example.com", FILTER_VALIDATE_URL, ["parser" => $parserName]));
+    var_dump(filter_var("http://www.example.com", FILTER_VALIDATE_URL, ["parser" => $parserName, "flags" => FILTER_FLAG_PATH_REQUIRED]));
+    var_dump(filter_var("http://www.example.com/path/at/the/server/", FILTER_VALIDATE_URL, ["parser" => $parserName, "flags" => FILTER_FLAG_PATH_REQUIRED]));
+    var_dump(filter_var("http://www.example.com/index.html", FILTER_VALIDATE_URL, ["parser" => $parserName, "flags" => FILTER_FLAG_QUERY_REQUIRED]));
+    var_dump(filter_var("http://www.example.com/index.php?a=b&c=d", FILTER_VALIDATE_URL, ["parser" => $parserName, "flags" => FILTER_FLAG_QUERY_REQUIRED]));
 }
 
 echo "RFC3986:\n";
-ini_set("uri.default_handler", "rfc3986");
-validateUrls();
+validateUrls(Uri\URI_PARSER_RFC3986);
 
 echo "\nWHATWG:\n";
-ini_set("uri.default_handler", "whatwg");
-validateUrls();
+validateUrls(Uri\URI_PARSER_WHATWG);
 
 echo "Done\n";
 ?>

diff --git a/ext/uri/php_lexbor.c b/ext/uri/php_lexbor.c
@@ -27,7 +27,6 @@ static zend_result lexbor_init_parser(void);
 static void *lexbor_parse_uri(const zend_string *url_str, const zend_string *base_url_str, zval *errors);
 static zend_class_entry *lexbor_get_uri_ce(void);
 static void *lexbor_clone_uri(void *uri);
-static zend_result lexbor_normalize_uri(void *uri);
 static zend_string *lexbor_uri_to_string(void *uri, bool exclude_fragment);
 static void lexbor_free_uri(void *uri);
 static zend_result lexbor_destroy_parser(void);
@@ -43,7 +42,7 @@ const uri_handler_t lexbor_uri_handler = {
 	lexbor_parse_uri,
 	lexbor_get_uri_ce,
 	lexbor_clone_uri,
-	lexbor_normalize_uri,
+	NULL,
 	lexbor_uri_to_string,
 	lexbor_free_uri,
 	lexbor_destroy_parser,
@@ -383,11 +382,6 @@ static void *lexbor_clone_uri(void *uri)
 	return lxb_url_clone(lexbor_parser->mraw, lexbor_uri);
 }
 
-static zend_result lexbor_normalize_uri(void *uri)
-{
-	return SUCCESS;
-}
-
 static lxb_status_t lexbor_serialize_callback(const lxb_char_t *data, size_t length, void *ctx)
 {
 	smart_str *uri_str = (smart_str *) ctx;

diff --git a/ext/uri/php_uri.c b/ext/uri/php_uri.c
@@ -224,7 +224,7 @@ PHPAPI void php_uri_instantiate_uri(
 	ZEND_ASSERT(Z_TYPE(errors) == IS_UNDEF);
 
 	if (!is_constructor) {
-		object_init_ex(return_value, handler->get_uri_ce());
+		object_init_ex(return_value, Z_CE_P(ZEND_THIS));
 	}
 
 	uri_object_t *uri_object = Z_URI_OBJECT_P(is_constructor ? ZEND_THIS : return_value);
@@ -255,7 +255,7 @@ static void create_rfc3986_uri(INTERNAL_FUNCTION_PARAMETERS, bool is_constructor
 	php_uri_instantiate_uri(INTERNAL_FUNCTION_PARAM_PASSTHRU, &uriparser_uri_handler, uri_str, base_url_str, is_constructor, false);
 }
 
-PHP_METHOD(Uri_Rfc3986Uri, create)
+PHP_METHOD(Uri_Rfc3986Uri, parse)
 {
 	create_rfc3986_uri(INTERNAL_FUNCTION_PARAM_PASSTHRU, false);
 }
@@ -288,7 +288,7 @@ static void create_whatwg_uri(INTERNAL_FUNCTION_PARAMETERS, bool is_constructor)
 	php_uri_instantiate_uri(INTERNAL_FUNCTION_PARAM_PASSTHRU, &lexbor_uri_handler, uri_str, base_url_str, is_constructor, true);
 }
 
-PHP_METHOD(Uri_WhatWgUri, create)
+PHP_METHOD(Uri_WhatWgUri, parse)
 {
 	create_whatwg_uri(INTERNAL_FUNCTION_PARAM_PASSTHRU, false);
 }
@@ -403,13 +403,57 @@ PHP_METHOD(Uri_Rfc3986Uri, equalsTo)
 		RETURN_FALSE;
 	}
 
-	zend_string *this_str = this_internal_uri->handler->uri_to_string(this_internal_uri->uri, exclude_fragment);
+	void *this_uri, *that_uri;
+
+	if (this_internal_uri->handler->normalize_uri != NULL) {
+		this_uri = this_internal_uri->handler->clone_uri(this_internal_uri->uri);
+		if (UNEXPECTED(this_uri == NULL)) {
+			zend_throw_error(uri_operation_exception_ce, "Failed to normalize %s", ZSTR_VAL(this_object->ce->name));
+			RETURN_THROWS();
+		}
+
+		zend_result result = this_internal_uri->handler->normalize_uri(this_uri);
+		if (UNEXPECTED(result == FAILURE)) {
+			zend_throw_error(uri_operation_exception_ce, "Failed to normalize %s", ZSTR_VAL(this_object->ce->name));
+			this_internal_uri->handler->free_uri(this_uri);
+			RETURN_THROWS();
+		}
+	} else {
+		this_uri = this_internal_uri->uri;
+	}
+
+	if (that_internal_uri->handler->normalize_uri != NULL) {
+		that_uri = that_internal_uri->handler->clone_uri(that_internal_uri->uri);
+		if (UNEXPECTED(that_uri == NULL)) {
+			zend_throw_error(uri_operation_exception_ce, "Failed to normalize %s", ZSTR_VAL(that_object->ce->name));
+			RETURN_THROWS();
+		}
+
+		zend_result result = that_internal_uri->handler->normalize_uri(that_uri);
+		if (UNEXPECTED(result == FAILURE)) {
+			zend_throw_error(uri_operation_exception_ce, "Failed to normalize %s", ZSTR_VAL(that_object->ce->name));
+			that_internal_uri->handler->free_uri(that_uri);
+			RETURN_THROWS();
+		}
+	} else {
+		that_uri = that_internal_uri->uri;
+	}
+
+	zend_string *this_str = this_internal_uri->handler->uri_to_string(this_internal_uri->uri, exclude_fragment); // TODO what happens if the __clone() method is overridden?
 	zend_string *that_str = that_internal_uri->handler->uri_to_string(that_internal_uri->uri, exclude_fragment);
 
 	RETVAL_BOOL(zend_string_equals(this_str, that_str));
 
 	zend_string_release(this_str);
 	zend_string_release(that_str);
+
+	if (this_internal_uri->handler->normalize_uri != NULL) {
+		this_internal_uri->handler->free_uri(this_uri);
+	}
+
+	if (that_internal_uri->handler->normalize_uri != NULL) {
+		that_internal_uri->handler->free_uri(that_uri);
+	}
 }
 
 PHP_METHOD(Uri_Rfc3986Uri, normalize)
@@ -420,6 +464,10 @@ PHP_METHOD(Uri_Rfc3986Uri, normalize)
 	uri_internal_t *internal_uri = uri_internal_from_obj(this_object);
 	URI_CHECK_INITIALIZATION_RETURN_THROWS(internal_uri, this_object);
 
+	if (internal_uri->handler->normalize_uri == NULL) {
+		RETURN_COPY(ZEND_THIS);
+	}
+
 	zend_object *new_object = uri_clone_obj_handler(this_object);
 	if (UNEXPECTED(EG(exception) != NULL)) {
 		zend_object_release(new_object);
@@ -446,6 +494,10 @@ PHP_METHOD(Uri_Rfc3986Uri, toNormalizedString)
 	uri_internal_t *internal_uri = uri_internal_from_obj(object);
 	URI_CHECK_INITIALIZATION_RETURN_THROWS(internal_uri, object);
 
+	if (internal_uri->handler->normalize_uri == NULL) {
+		RETURN_STR(internal_uri->handler->uri_to_string(internal_uri->uri, false));
+	}
+
 	void *new_uri = internal_uri->handler->clone_uri(internal_uri->uri);
 	if (UNEXPECTED(new_uri == NULL)) {
 		zend_throw_error(uri_operation_exception_ce, "Failed to normalize %s", ZSTR_VAL(object->ce->name));
@@ -513,12 +565,13 @@ PHP_METHOD(Uri_Rfc3986Uri, __unserialize)
 	zend_object *object = Z_OBJ_P(ZEND_THIS);
 	uri_internal_t *internal_uri = uri_internal_from_obj(object);
 
-	zend_string *str = zend_string_init("https://example.com", sizeof("https://example.com") - 1, false);
+	zend_string *str = zend_string_init("https://example.com", sizeof("https://example.com") - 1, false); // TODO set URI components from ht
 
 	zval errors;
 	ZVAL_UNDEF(&errors);
 
 	internal_uri->handler = uri_handler_by_name(URI_PARSER_RFC3986, sizeof(URI_PARSER_RFC3986) - 1);
+	// TODO free current URI if any
 	internal_uri->uri = internal_uri->handler->parse_uri(str, NULL, &errors);
 	if (internal_uri->uri == NULL) {
 		throw_invalid_uri_exception(&errors);
@@ -543,12 +596,13 @@ PHP_METHOD(Uri_WhatWgUri, __unserialize)
 	zend_object *object = Z_OBJ_P(ZEND_THIS);
 	uri_internal_t *internal_uri = uri_internal_from_obj(object);
 
-	zend_string *str = zend_string_init("https://example.com", sizeof("https://example.com") - 1, false);
+	zend_string *str = zend_string_init("https://example.com", sizeof("https://example.com") - 1, false); // TODO set URI components from ht
 
 	zval errors;
 	ZVAL_UNDEF(&errors);
 
 	internal_uri->handler = uri_handler_by_name(URI_PARSER_WHATWG, sizeof(URI_PARSER_WHATWG) - 1);
+	// TODO free current URI if any
 	internal_uri->uri = internal_uri->handler->parse_uri(str, NULL, &errors);
 	if (internal_uri->uri == NULL) {
 		throw_invalid_uri_exception(&errors);
@@ -708,7 +762,7 @@ static zend_object *uri_clone_obj_handler(zend_object *object)
 
 	new_uri_object->internal.handler = internal_uri->handler;
 
-	void *uri = internal_uri->handler->clone_uri(internal_uri->uri);
+	void *uri = internal_uri->handler->clone_uri(internal_uri->uri);  // TODO what happens if the __clone() method is overridden?
 	if (UNEXPECTED(uri == NULL)) {
 		zend_throw_error(uri_operation_exception_ce, "Failed to clone %s", ZSTR_VAL(object->ce->name));
 		return &new_uri_object->std;
@@ -773,9 +827,8 @@ zend_result uri_handler_register(const uri_handler_t *uri_handler)
 	ZEND_ASSERT(uri_handler->name != NULL);
 	ZEND_ASSERT(uri_handler->init_parser != NULL);
 	ZEND_ASSERT(uri_handler->parse_uri != NULL);
-	ZEND_ASSERT(uri_handler->get_uri_ce != NULL);
+	ZEND_ASSERT(uri_handler->get_uri_ce != NULL); // TODO unused handler, maybe remove?
 	ZEND_ASSERT(uri_handler->clone_uri != NULL);
-	ZEND_ASSERT(uri_handler->normalize_uri != NULL);
 	ZEND_ASSERT(uri_handler->uri_to_string != NULL);
 	ZEND_ASSERT(uri_handler->free_uri != NULL);
 	ZEND_ASSERT(uri_handler->destroy_parser != NULL);

diff --git a/ext/uri/php_uri.stub.php b/ext/uri/php_uri.stub.php
@@ -38,7 +38,7 @@ class InvalidUriException extends \Uri\UriException
 }
 
 /** @strict-properties */
-final readonly class WhatWgError
+readonly class WhatWgError
 {
     /** @cvalue LXB_URL_ERROR_TYPE_DOMAIN_TO_ASCII */
     public const int ERROR_TYPE_DOMAIN_TO_ASCII = UNKNOWN;
@@ -168,7 +168,7 @@ public function __toString(): string {}
     /** @virtual */
     private ?string $fragment;
 
-    public static function create(string $uri, ?string $baseUrl = null): ?static {}
+    public static function parse(string $uri, ?string $baseUrl = null): ?static {}
 
     public function __construct(string $uri, ?string $baseUrl = null) {}
 
@@ -237,7 +237,7 @@ public function __unserialize(array $data): void;
     /** @virtual */
     private ?string $fragment;
 
-    public static function create(string $uri, ?string $baseUrl = null): static|array {}
+    public static function parse(string $uri, ?string $baseUrl = null): static|array {}
 
     public function __construct(string $uri, ?string $baseUrl = null) {}
 

diff --git a/ext/uri/php_uri_arginfo.h b/ext/uri/php_uri_arginfo.h
diff --git a/ext/uri/php_uriparser.c b/ext/uri/php_uriparser.c
@@ -341,7 +341,7 @@ static zend_result uriparser_write_port(uri_internal_t *internal_uri, zval *valu
 	uriparser_append_authority(internal_uri, &uri_str);
 	uriparser_append_host(internal_uri, &uri_str);
 
-	if (Z_TYPE_P(value) == IS_LONG && Z_LVAL_P(value) != 0) {
+	if (Z_TYPE_P(value) == IS_LONG) {
 		smart_str_appendc(&uri_str, ':');
 		smart_str_append_long(&uri_str, Z_LVAL_P(value));
 	}

diff --git a/ext/uri/tests/001.phpt b/ext/uri/tests/001.phpt
@@ -7,13 +7,13 @@ uri
 
 $t1 = hrtime(true);
 for ($i = 0; $i < 1000; $i++) {
-   Uri\Rfc3986Uri::create("https://example.com");
+   Uri\Rfc3986Uri::parse("https://example.com");
 }
 $t2 = hrtime(true);
 
 $t3 = hrtime(true);
 for ($i = 0; $i < 1000; $i++) {
-    Uri\WhatWgUri::create("https://example.com/");
+    Uri\WhatWgUri::parse("https://example.com/");
 }
 $t4 = hrtime(true);
 
@@ -25,12 +25,12 @@ $t6 = hrtime(true);
 
 echo "RFC 3986 parser: https://example.com\n";
 var_dump(($t2 - $t1) / 1_000_000_000);
-var_dump(Uri\Rfc3986Uri::create("https://example.com"));
+var_dump(Uri\Rfc3986Uri::parse("https://example.com"));
 echo "------------------------\n";
 
 echo "WHATWG parser: https://example.com\n";
 var_dump(($t4 - $t3) / 1_000_000_000);
-var_dump(Uri\WhatWgUri::create("https://example.com"));
+var_dump(Uri\WhatWgUri::parse("https://example.com"));
 echo "------------------------\n";
 
 echo "PHP parser: https://example.com\n";