Ticket #36356: 36356-wp_parse_url.patch

src/wp-includes/http.php

@@ -630 +630 @@
  * when URL parsing failed.
  *
  * @since 4.4.0
- *
- * @param string $url The URL to parse.
- * @return bool|array False on failure; Array of URL components on success;
- *                    See parse_url()'s return values.
+ * @since 4.6.0 The $component parameter was added for function completeness.
+ *
+ * @param string $url       The URL to parse.
+ * @param int    $component The specific component to retrieve. Use one of the PHP
+ *                          predefined constants to specify which one.
+ *                          Defaults to -1 (= return all parts as an array).
+ *                          @see http://php.net/manual/en/function.parse-url.php
+ * @return mixed False on failure; Array of URL components on success;
+ *               When a specific component has been requested: null if the component doesn't
+ *               exist in the given URL; a sting or - in the case of PHP_URL_PORT - integer
+ *               when it does;
+ *               See parse_url()'s return values.
  */
-function wp_parse_url( $url ) {
-        $parts = @parse_url( $url );
-        if ( ! $parts ) {
+function wp_parse_url( $url, $component = -1 ) {
+        $parts = @parse_url( $url, $component );
+
+        if ( version_compare( PHP_VERSION, '5.4.7', '>=' ) ) {
+                return $parts;
+        }
+
+        if ( false === $parts ) {
                 // < PHP 5.4.7 compat, trouble with relative paths including a scheme break in the path
                 if ( '/' == $url[0] && false !== strpos( $url, '://' ) ) {
                         // Since we know it's a relative path, prefix with a scheme/host placeholder and try again
-                        if ( ! $parts = @parse_url( 'placeholder://placeholder' . $url ) ) {
+                        if ( ! $parts = @parse_url( 'placeholder://placeholder' . $url, $component ) ) {
                                 return $parts;
                         }
                         // Remove the placeholder values
@@ -652 +665 @@
         }
 
         // < PHP 5.4.7 compat, doesn't detect schemeless URL's host field
-        if ( '//' == substr( $url, 0, 2 ) && ! isset( $parts['host'] ) ) {
-                $path_parts = explode( '/', substr( $parts['path'], 2 ), 2 );
-                $parts['host'] = $path_parts[0];
-                if ( isset( $path_parts[1] ) ) {
-                        $parts['path'] = '/' . $path_parts[1];
-                } else {
-                        unset( $parts['path'] );
+        if ( '//' == substr( $url, 0, 2 ) ) {
+                if ( -1 === $component && ! isset( $parts['host'] ) ) {
+                        $path_parts = explode( '/', substr( $parts['path'], 2 ), 2 );
+                        $parts['host'] = $path_parts[0];
+                        if ( isset( $path_parts[1] ) ) {
+                                $parts['path'] = '/' . $path_parts[1];
+                        } else {
+                                unset( $parts['path'] );
+                        }
+                } elseif ( PHP_URL_HOST === $component || PHP_URL_PATH === $component ) {
+                        $all_parts = @parse_url( $url );
+                        if ( ! isset( $all_parts['host'] ) ) {
+                                $path_parts = explode( '/', substr( $all_parts['path'], 2 ), 2 );
+                                if ( PHP_URL_PATH === $component ) {
+                                        if ( isset( $path_parts[1] ) ) {
+                                                $parts = '/' . $path_parts[1];
+                                        } else {
+                                                $parts = null;
+                                        }
+                                } elseif ( PHP_URL_HOST === $component ) {
+                                        $parts = $path_parts[0];
+                                }
+                        }
                 }
         }
 

tests/phpunit/tests/http/http.php

@@ -104 +104 @@
         }
 
         /**
+         * @ticket 36356
+         *
+         * @dataProvider parse_url_component_testcases
+         */
+        function test_wp_parse_url_with_component( $url, $component, $expected ) {
+                $actual = wp_parse_url( $url, $component );
+                $this->assertEquals( $expected, $actual );
+        }
+
+        function parse_url_component_testcases() {
+                $test_url = 'http://username:password@hostname:9090/path?arg=value#anchor';
+
+                // 0: The URL, 1: The requested component, 2: The expected resulting structure.
+                return array(
+                        array( $test_url, -1, array( 'scheme' => 'http', 'host' => 'hostname', 'port' => 9090, 'user' => 'username', 'pass' => 'password', 'path' => '/path', 'query' => 'arg=value', 'fragment' => 'anchor' ) ),
+                        array( $test_url, PHP_URL_SCHEME, 'http' ),
+                        array( $test_url, PHP_URL_USER, 'username' ),
+                        array( $test_url, PHP_URL_PASS, 'password' ),
+                        array( $test_url, PHP_URL_HOST, 'hostname' ),
+                        array( $test_url, PHP_URL_PORT, 9090 ),
+                        array( $test_url, PHP_URL_PATH, '/path' ),
+                        array( $test_url, PHP_URL_QUERY, 'arg=value' ),
+                        array( $test_url, PHP_URL_FRAGMENT, 'anchor' ),
+
+                        // < PHP 5.4.7: Schemeless URL
+                        array( '//example.com/path/', PHP_URL_HOST, 'example.com' ),
+                        array( '//example.com/path/', PHP_URL_PATH, '/path/' ),
+                        array( '//example.com/', PHP_URL_HOST, 'example.com' ),
+                        array( '//example.com/', PHP_URL_PATH, '/' ),
+                        array( 'http://example.com//path/', PHP_URL_HOST, 'example.com' ),
+                        array( 'http://example.com//path/', PHP_URL_PATH, '//path/' ),
+
+                        // < PHP 5.4.7: Scheme seperator in the URL
+                        array( 'http://example.com/http://example.net/', PHP_URL_HOST, 'example.com' ),
+                        array( 'http://example.com/http://example.net/', PHP_URL_PATH, '/http://example.net/' ),
+                        array( '/path/http://example.net/', PHP_URL_HOST, null ),
+                        array( '/path/http://example.net/', PHP_URL_PATH, '/path/http://example.net/' ),
+
+                        // < PHP 5.4.7: IPv6 literals in schemeless URLs are handled incorrectly.
+                        array( '//[::FFFF::127.0.0.1]/', PHP_URL_HOST, '[::FFFF::127.0.0.1]' ),
+                        array( '//[::FFFF::127.0.0.1]/', PHP_URL_PATH, '/' ),
+
+                        // PHP's parse_url() calls this an invalid url, we handle it as a path
+                        array( '/://example.com/', PHP_URL_PATH, '/://example.com/' ),
+
+                );
+                /*
+                Untestable edge cases in various PHP:
+                  - ///example.com - Fails in PHP >= 5.4.7, assumed path in <5.4.7
+                  - ://example.com - assumed path in PHP >= 5.4.7, fails in <5.4.7
+                */
+        }
+
+        /**
          * @ticket 35426
          */
         public function test_http_response_code_constants() {