Merge pull request #995 from paulschreiber/fix/parse-url

Consistently use wp_parse_url(); deprecate AMP_WP_Utils

includes/embeds/class-amp-dailymotion-embed.php

@@ -150,7 +150,7 @@ public function render( $args ) {
 	 * @return integer Video ID.
 	 */
 	private function get_video_id_from_url( $url ) {
-		$parsed_url = AMP_WP_Utils::parse_url( $url );
+		$parsed_url = wp_parse_url( $url );
 		parse_str( $parsed_url['path'], $path );
 		$tok      = explode( '/', $parsed_url['path'] );
 		$tok      = explode( '_', $tok[2] );

includes/embeds/class-amp-soundcloud-embed.php

@@ -178,7 +178,7 @@ private function render_embed_fallback( $url ) {
 	 * @return string Track ID or empty string if none matched.
 	 */
 	private function get_track_id_from_url( $url ) {
-		$parsed_url = AMP_WP_Utils::parse_url( $url );
+		$parsed_url = wp_parse_url( $url );
 		if ( ! preg_match( '#tracks/(?P<track_id>[^/]+)#', $parsed_url['path'], $matches ) ) {
 			return '';
 		}

includes/embeds/class-amp-vimeo-embed.php

@@ -156,7 +156,7 @@ public function render( $args ) {
 	 * @return integer Video ID.
 	 */
 	private function get_video_id_from_url( $url ) {
-		$parsed_url = parse_url( $url );
+		$parsed_url = wp_parse_url( $url );
 		parse_str( $parsed_url['path'], $path );
 
 		$video_id = '';

includes/embeds/class-amp-youtube-embed.php

@@ -149,7 +149,7 @@ public function render( $args ) {
 	 */
 	private function get_video_id_from_url( $url ) {
 		$video_id   = false;
-		$parsed_url = AMP_WP_Utils::parse_url( $url );
+		$parsed_url = wp_parse_url( $url );
 
 		if ( self::SHORT_URL_HOST === substr( $parsed_url['host'], -strlen( self::SHORT_URL_HOST ) ) ) {
 			/* youtu.be/{id} */

includes/sanitizers/class-amp-img-sanitizer.php

@@ -227,7 +227,7 @@ private function adjust_and_replace_node( $node ) {
 	 */
 	private function is_gif_url( $url ) {
 		$ext  = self::$anim_extension;
-		$path = AMP_WP_Utils::parse_url( $url, PHP_URL_PATH );
+		$path = wp_parse_url( $url, PHP_URL_PATH );
 		return substr( $path, -strlen( $ext ) ) === $ext;
 	}
 }

includes/sanitizers/class-amp-tag-and-attribute-sanitizer.php

@@ -1216,7 +1216,7 @@ private function check_attr_spec_rule_allowed_protocol( $node, $attr_name, $attr
 					 * This seems to be an acceptable check since the AMP validator
 					 * will allow a URL with no protocol to pass validation.
 					 */
-					$url_scheme = AMP_WP_Utils::parse_url( $url, PHP_URL_SCHEME );
+					$url_scheme = wp_parse_url( $url, PHP_URL_SCHEME );
 					if ( $url_scheme ) {
 						if ( ! in_array( strtolower( $url_scheme ), $attr_spec_rule[ AMP_Rule_Spec::VALUE_URL ][ AMP_Rule_Spec::ALLOWED_PROTOCOL ], true ) ) {
 							return AMP_Rule_Spec::FAIL;
@@ -1233,7 +1233,7 @@ private function check_attr_spec_rule_allowed_protocol( $node, $attr_name, $attr
 							 * This seems to be an acceptable check since the AMP validator
 							 *  will allow a URL with no protocol to pass validation.
 							 */
-							$url_scheme = AMP_WP_Utils::parse_url( $url, PHP_URL_SCHEME );
+							$url_scheme = wp_parse_url( $url, PHP_URL_SCHEME );
 							if ( $url_scheme ) {
 								if ( ! in_array( strtolower( $url_scheme ), $attr_spec_rule[ AMP_Rule_Spec::VALUE_URL ][ AMP_Rule_Spec::ALLOWED_PROTOCOL ], true ) ) {
 									return AMP_Rule_Spec::FAIL;
@@ -1266,7 +1266,7 @@ private function check_attr_spec_rule_disallowed_relative( $node, $attr_name, $a
 			if ( $node->hasAttribute( $attr_name ) ) {
 				$urls_to_test = preg_split( '/\s*,\s*/', $node->getAttribute( $attr_name ) );
 				foreach ( $urls_to_test as $url ) {
-					$parsed_url = AMP_WP_Utils::parse_url( $url );
+					$parsed_url = wp_parse_url( $url );
 
 					/*
 					 *  The JS AMP validator seems to consider 'relative' to mean
@@ -1285,7 +1285,7 @@ private function check_attr_spec_rule_disallowed_relative( $node, $attr_name, $a
 					if ( $node->hasAttribute( $alternative_name ) ) {
 						$urls_to_test = preg_split( '/\s*,\s*/', $node->getAttribute( $alternative_name ) );
 						foreach ( $urls_to_test as $url ) {
-							$parsed_url = AMP_WP_Utils::parse_url( $url );
+							$parsed_url = wp_parse_url( $url );
 							if ( empty( $parsed_url['scheme'] ) ) {
 								return AMP_Rule_Spec::FAIL;
 							}
@@ -1338,7 +1338,7 @@ private function check_attr_spec_rule_disallowed_empty( $node, $attr_name, $attr
 	private function check_attr_spec_rule_disallowed_domain( $node, $attr_name, $attr_spec_rule ) {
 		if ( isset( $attr_spec_rule[ AMP_Rule_Spec::DISALLOWED_DOMAIN ] ) && $node->hasAttribute( $attr_name ) ) {
 			$attr_value = $node->getAttribute( $attr_name );
-			$url_domain = AMP_WP_Utils::parse_url( $attr_value, PHP_URL_HOST );
+			$url_domain = wp_parse_url( $attr_value, PHP_URL_HOST );
 			if ( ! empty( $url_domain ) ) {
 				foreach ( $attr_spec_rule[ AMP_Rule_Spec::DISALLOWED_DOMAIN ] as $disallowed_domain ) {
 					if ( strtolower( $url_domain ) === strtolower( $disallowed_domain ) ) {

includes/utils/class-amp-image-dimension-extractor.php

@@ -51,7 +51,7 @@ public static function normalize_url( $url ) {
 			return set_url_scheme( $url, 'http' );
 		}
 
-		$parsed = AMP_WP_Utils::parse_url( $url );
+		$parsed = wp_parse_url( $url );
 		if ( ! isset( $parsed['host'] ) ) {
 			$path = '';
 			if ( isset( $parsed['path'] ) ) {

includes/utils/class-amp-wp-utils.php

@@ -1,15 +1,39 @@
 <?php
+/**
+ * Class AMP_WP_Utils
+ *
+ * @package AMP
+ */
 
+/**
+ * Class AMP_WP_Utils
+ *
+ * @since 0.5
+ *
+ * @deprecated 0.7 As WordPress 4.7 is our minimum supported version.
+ */
 class AMP_WP_Utils {
 	/**
-	 * wp_parse_url in < WordPress 4.7 does not respect the component arg, so we're adding this helper so we can use it.
+	 * The core function wp_parse_url in < WordPress 4.7 does not respect the component arg. This helper lets us use it.
 	 *
-	 * This can be removed once 4.8 is out and we bump up our min supported WP version.
+	 * Don't use.
+	 *
+	 * @deprecated 0.7 wp_parse_url() is now used instead.
+	 *
+	 * @param string $url       The raw URL. Can be false if the URL failed 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).
+	 * @return mixed False on parse 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 string or - in the case of
+	 *               PHP_URL_PORT - integer when it does. See parse_url()'s return values.
 	 */
 	public static function parse_url( $url, $component = -1 ) {
+		_deprecated_function( __METHOD__, '0.7', 'wp_parse_url' );
 		$parsed = wp_parse_url( $url, $component );
 
-		// Because < 4.7 always returned a full array regardless of component
+		// Because < 4.7 always returned a full array regardless of component.
 		if ( -1 !== $component && is_array( $parsed ) ) {
 			return self::_get_component_from_parsed_url_array( $parsed, $component );
 		}
@@ -21,6 +45,17 @@ public static function parse_url( $url, $component = -1 ) {
 	 * Included for 4.6 back-compat
 	 *
 	 * Copied from https://developer.wordpress.org/reference/functions/_get_component_from_parsed_url_array/
+	 *
+	 * @deprecated 0.7
+	 *
+	 * @param array|false $url_parts The parsed URL. Can be false if the URL failed 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).
+	 * @return mixed False on parse 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 string or - in the case of
+	 *               PHP_URL_PORT - integer when it does. See parse_url()'s return values.
 	 */
 	protected static function _get_component_from_parsed_url_array( $url_parts, $component = -1 ) {
 		if ( -1 === $component ) {
@@ -39,6 +74,12 @@ protected static function _get_component_from_parsed_url_array( $url_parts, $com
 	 * Included for 4.6 back-compat
 	 *
 	 * Copied from https://developer.wordpress.org/reference/functions/_wp_translate_php_url_constant_to_key/
+	 *
+	 * @param int $constant The specific component to retrieve. Use one of the PHP
+	 *                      predefined constants to specify which one.
+	 * @return mixed False if component not found. string or integer if found.
+	 *
+	 * @deprecated 0.7
 	 */
 	protected static function _wp_translate_php_url_constant_to_key( $constant ) {
 		$translation = array(

jetpack-helper.php

@@ -56,7 +56,7 @@ function jetpack_amp_build_stats_pixel_url() {
 		$blog = Jetpack_Options::get_option( 'id' );
 		$tz = get_option( 'gmt_offset' );
 		$v = 'ext';
-		$blog_url = AMP_WP_Utils::parse_url( site_url() );
+		$blog_url = wp_parse_url( site_url() );
 		$srv = $blog_url['host'];
 		$j = sprintf( '%s:%s', JETPACK__API_VERSION, JETPACK__VERSION );
 		$post = $wp_the_query->get_queried_object_id();

wpcom-helper.php

@@ -129,12 +129,12 @@ function wpcom_amp_extract_image_dimensions_from_querystring( $dimensions ) {
 			continue;
 		}
 
-		$host = parse_url( $url, PHP_URL_HOST );
+		$host = wp_parse_url( $url, PHP_URL_HOST );
 		if ( ! wp_endswith( $host, '.wp.com' ) || ! wp_endswith( $host, '.files.wordpress.com' ) ) {
 			continue;
 		}
 
-		parse_str( parse_url( $url, PHP_URL_QUERY ), $query );
+		parse_str( wp_parse_url( $url, PHP_URL_QUERY ), $query );
 		$w = isset( $query['w'] ) ? absint( $query['w'] ) : false;
 		$h = isset( $query['h'] ) ? absint( $query['h'] ) : false;