Add internal URI handling API #19073

kocsismate · 2025-07-08T21:13:52Z

No description provided.

TimWolla

Some first remarks. Did not yet look at everything.

TimWolla · 2025-07-09T11:40:46Z

ext/standard/url.c

+static zend_string *parse_url_uri_to_string(void *uri, uri_recomposition_mode_t recomposition_mode, bool exclude_fragment)
+{
+	ZEND_UNREACHABLE();
+}


Might be better to simply NULL the pointer in the uri_handler_t struct instead.

I got the same comment from @DanielEScherzer in the original PR, and I wrote him that I would like to avoid making the handlers optional if possible, because this way the existence of the handlers don't have to be checked before their usage - it's advantageous both for maintainability and performance.

The parse_url based implementation is special because it's not directly exposed for userland - it's just an internal URI "backend" for BC, and these handlers aren't necessarily needed for now. We could of course expose the to_string handlers later for 3rd party extensions if we want to. Then the code should probably be changed to something else.

A function that triggers undefined behavior when called (this is what ZEND_UNREACHABLE implies for production builds) and not having a function (i.e. dereferencing a NULL pointer when trying to call the function) are functionally the same. In both cases the PHP binary will do something bad (ideally just crash).

Thus it seems to be preferable to clearly indicate that the handler is not available by using NULL rather than pretending there is a handler when calling it is unsafe.

OK, I see what you mean now. I can't comment on what method is preferable, but intentionally passing a NULL value instead of a handler function, while callers of the handlers never expect NULL also seems wrong. Normally, static analyzers would emit an error in this case (in PHP for sure, and I don't know about C), that's why I didn't even think about this solution.

TBH the code which uses ZEND_UNREACHABLE() is unreachable indeed if one uses the internal API: currently, no function is exposed that would make use of the relevant handlers.

TimWolla · 2025-07-09T11:41:07Z

ext/standard/url.c

+static void *parse_url_clone_uri(void *uri)
+{
+	ZEND_UNREACHABLE();
+}


ext/standard/url.c

TimWolla · 2025-07-09T11:43:21Z

ext/uri/php_uri.c

+	if (uri_handler_name == NULL) {
+		return uri_handler_by_name("parse_url", sizeof("parse_url") - 1);
+	}
+


Defaulting to parse_url in a new API is probably not a good idea. Instead the “legacy” users should just pass "parse_url" explicitly.

Defaulting to parse_url here works because that's the default indeed where php_uri_get_handler() is called, the other "backends" can only be used if the config is explicitly passed (not null).

The other reason why I opted for this approach is that it would be inconvenient to create and free a new zend_string when the legacy implementation is needed, and I wanted to avoid adding a known string just for this purpose, or exposing the C string based uri_handler_by_name function instead.

ext/uri/php_uri.c

TimWolla

I've looked at this again and I must say that I'm having trouble meaningfully reviewing this. It adds a large amount of code with unclear purpose and confusing (to me) naming.

TimWolla · 2025-07-16T11:54:07Z

ext/uri/php_uri.c

+PHPAPI zend_result php_uri_get_scheme(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_SCHEME, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_username(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_USERNAME, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_password(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PASSWORD, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_host(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_HOST, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_port(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PORT, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_path(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PATH, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_query(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_QUERY, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_fragment(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_FRAGMENT, read_mode, zv);
+}


The addition of these new helpers is not clear to me. It feels like just another layer of indirection by moving the enum into the function name. There's also already uri_property_handler_from_internal_uri(), why doesn't it work here?

These functions come from the time when the property was passed as a zend_string, so having separate methods used to make sense. You are right, these are not really needed anymore, so I'm fine with removing them.

Although the alternative code is quite much longer, and a bit more difficult to use:

- zend_result result = php_uri_get_host(internal_uri, URI_COMPONENT_READ_RAW, &host_zv); + zend_result result = php_uri_property_handler_from_internal_uri(internal_uri, URI_PROPERTY_NAME_USERNAME)->read_func(internal_uri, URI_COMPONENT_READ_RAW, &host_zv);

+it makes the handlers directly available for usage, which I wanted to avoid for now (because write handlers are not always available)

I don't really mind the extra helpers

I had already implemented the suggestion when I realized that the helpers really simplify usage, so I got rid of my changes after all.

nielsdos

Preliminary review round

nielsdos · 2025-07-21T13:29:24Z

ext/uri/php_uri.c

+		uri->scheme = zend_string_copy(Z_STR(tmp));
+		zval_ptr_dtor(&tmp);


You can just "transfer the lifetime" instead of copy+dtor

Suggested change

uri->scheme = zend_string_copy(Z_STR(tmp));

zval_ptr_dtor(&tmp);

uri->scheme = Z_STR(tmp);

This occurs a couple of times in this function.

nielsdos · 2025-07-21T13:31:03Z

ext/uri/php_uri.c

+PHPAPI zend_result php_uri_get_scheme(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_SCHEME, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_username(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_USERNAME, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_password(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PASSWORD, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_host(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_HOST, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_port(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PORT, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_path(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_PATH, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_query(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_QUERY, read_mode, zv);
+}
+
+PHPAPI zend_result php_uri_get_fragment(const uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *zv)
+{
+	return php_uri_get_property(internal_uri, URI_PROPERTY_NAME_FRAGMENT, read_mode, zv);
+}


I don't really mind the extra helpers

nielsdos · 2025-07-21T13:32:19Z

ext/uri/php_uri.c

+
+	result = php_uri_get_scheme(uri_internal, read_mode, &tmp);
+	if (result == FAILURE) {
+		php_uri_free(uri_internal);


Is it possible to avoid the repeated error blocks by using a single "error" label that you goto to?
I'd say one of the only proper use-cases for goto in C is for these kinds of repeated error handlings.

Yes, agreed! This is also a very good idea :)

nielsdos · 2025-07-21T13:34:30Z

ext/standard/http_fopen_wrapper.c

+	if (uri_handler == NULL) {
+		return NULL;
+	}
+	zend_string *tmp_uri = zend_string_init(path, strlen(path), false);


Bah the fact that we need to make a copy is sad :-(

Yes, it's a pity... Should I then add C string based helper functions for parsing? 🤔

If the parse_uri handlers would take a const char*, size_t pair instead of a zend_string*, we wouldn't have this problem. So perhaps the internal API should be changed. FWIW I don't think that we should have one for both a zend_string* and a ptr-length pair, just one should be enough. The pair one is more flexible and since we don't take copies of the input string it should not make any other case worse.

I've just implemented this.

ext/standard/ftp_fopen_wrapper.c

nielsdos · 2025-07-21T13:37:42Z

ext/openssl/xp_ssl.c

+	zend_string *resource = zend_string_init(resourcename, resourcenamelen, false);
+	uri_internal_t *internal_uri = php_uri_parse(uri_handler, resource, true);
+	if (internal_uri == NULL) {
+		zend_string_release(resource);


Can be zend_string_efree even

nielsdos · 2025-07-21T13:38:23Z

ext/soap/php_http.h

-                           char        *soapaction,
-                           int          soap_version,
-                           zval        *response);
+int make_http_soap_request(zval              *this_ptr,


Indentation is a bit weird here

ext/soap/php_http.c

nielsdos · 2025-07-21T13:43:38Z

ext/soap/php_http.c

@@ -1143,39 +1152,48 @@ int make_http_soap_request(zval        *this_ptr,
 		char *loc;

 		if ((loc = get_http_header_value(ZSTR_VAL(http_headers), "Location:")) != NULL) {
-			php_url *new_url  = php_url_parse(loc);
+			uri_handler_t *uri_handler = php_uri_get_handler(uri_parser_class);
+			if (uri_handler == NULL) {


nielsdos · 2025-07-21T13:43:55Z

ext/soap/php_http.c

+
+			zend_string *loc_str = zend_string_init(loc, strlen(loc), false);
+			php_uri *new_uri = php_uri_parse_to_struct(uri_handler, loc_str, URI_COMPONENT_READ_NORMALIZED_ASCII, true);
+			zend_string_release(loc_str);


Could be zend_string_release_ex with persistent=false

nielsdos

The switch from zend_string to the pointer-length pair seems to have been a good idea

nielsdos · 2025-07-22T10:56:14Z

main/streams/php_stream_context.h

@@ -14,6 +14,8 @@
   +----------------------------------------------------------------------+
 */

+#include "ext/uri/php_uri.h"


I wonder if this should be a forward declaration instead, as to not complicate the headers further

nielsdos · 2025-07-22T10:56:44Z

ext/uri/php_uri.h

@@ -22,6 +22,37 @@
 extern zend_module_entry uri_module_entry;
 #define phpext_uri_ptr &uri_module_entry

-PHPAPI void php_uri_implementation_set_object_handlers(zend_class_entry *ce, zend_object_handlers *object_handlers);
+typedef struct {


Suggested change

typedef struct {

typedef struct php_uri {

The reason being that it's more convenient for tools like pahole

nielsdos · 2025-07-22T10:57:37Z

ext/uri/php_uri.c

+
+	php_uri *uri = ecalloc(1, sizeof(*uri));
+	zval tmp;
+	ZVAL_UNDEF(&tmp);


I wonder why this line is actually needed

nielsdos · 2025-07-22T10:58:15Z

ext/soap/php_http.h

-                           char        *soapaction,
-                           int          soap_version,
-                           zval        *response);
+						   zend_string *buf,


Indentation looks still weird on github

nielsdos · 2025-07-22T10:58:26Z

ext/soap/php_http.c

-                           char        *soapaction,
-                           int          soap_version,
-                           zval        *return_value)
+						   zend_string *buf,


Indentation looks still weird on github

nielsdos · 2025-07-22T10:59:58Z

ext/openssl/xp_ssl.c

+	zval host_zv;
+	zend_result result = php_uri_get_host(internal_uri, URI_COMPONENT_READ_RAW, &host_zv);
+	if (result == SUCCESS && Z_TYPE(host_zv) == IS_STRING) {
+		const char * host = Z_STRVAL(host_zv);
 		char * url_name = NULL;


If you move this declaration outside this if block, then you can remove lines 2655-2657 and change the return in line 2662 to return url_name;
The nice thing is that all the cleanup is then centralised

nielsdos · 2025-07-22T11:00:48Z

ext/filter/logical_filters.c

 		RETURN_VALIDATION_FAILED
 	}

-	if ((url->user != NULL && !is_userinfo_valid(url->user))
-		|| (url->pass != NULL && !is_userinfo_valid(url->pass))
+	if (strcmp(uri_handler->name, "parse_url") == 0 &&


Suggested change

if (strcmp(uri_handler->name, "parse_url") == 0 &&

if (strcmp(uri_handler->name, URI_PARSER_PHP) == 0 &&

nielsdos · 2025-07-22T12:26:31Z

ext/openssl/xp_ssl.c

-	if (url->host) {
-		const char * host = ZSTR_VAL(url->host);
+	zval host_zv;
+	zend_result result = php_uri_get_host(internal_uri, URI_COMPONENT_READ_RAW, &host_zv);


For all these property reads: why do you actually check the type? If reading was successful (i.e. returned SUCCESS) I would assume that the type is correct.

kocsismate requested a review from TimWolla July 8, 2025 21:13

github-actions bot added Extension: openssl Extension: filter Extension: soap Extension: standard ABI break Extension: uri labels Jul 8, 2025

TimWolla reviewed Jul 9, 2025

View reviewed changes

kocsismate force-pushed the ext-url7 branch from d2c1e8d to c8bceb8 Compare July 15, 2025 21:23

TimWolla reviewed Jul 16, 2025

View reviewed changes

kocsismate added 2 commits July 20, 2025 23:44

Add internal URI handling API

8d1c0cf

Some review comment fixes

b6f76b3

kocsismate force-pushed the ext-url7 branch from c8bceb8 to fcd17da Compare July 20, 2025 21:44

github-actions bot added the Extension: reflection label Jul 20, 2025

Fix tests

947bc64

kocsismate force-pushed the ext-url7 branch from fcd17da to 947bc64 Compare July 21, 2025 06:12

kocsismate marked this pull request as ready for review July 21, 2025 08:09

kocsismate requested review from bukka, nielsdos and DanielEScherzer as code owners July 21, 2025 08:09

nielsdos reviewed Jul 21, 2025

View reviewed changes

Implement review suggestions

2a85703

kocsismate mentioned this pull request Jul 21, 2025

[RFC] Add RFC 3986 and WHATWG compliant URL parsing support #14461

Open

Change the parse_uri handler to use const char *

a76e193

nielsdos reviewed Jul 22, 2025

View reviewed changes

		uri->scheme = zend_string_copy(Z_STR(tmp));
		zval_ptr_dtor(&tmp);

	uri->scheme = zend_string_copy(Z_STR(tmp));
	zval_ptr_dtor(&tmp);
	uri->scheme = Z_STR(tmp);

	if (strcmp(uri_handler->name, "parse_url") == 0 &&
	if (strcmp(uri_handler->name, URI_PARSER_PHP) == 0 &&

Add internal URI handling API #19073

Are you sure you want to change the base?

Add internal URI handling API #19073

Conversation

kocsismate commented Jul 8, 2025

Uh oh!

TimWolla left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

TimWolla left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kocsismate Jul 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kocsismate Jul 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nielsdos left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nielsdos left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

kocsismate Jul 18, 2025 •

edited

Loading

kocsismate Jul 18, 2025 •

edited

Loading

nielsdos left a comment •

edited

Loading