fix: improve lazyload and replacement on json strings

Codeinwp · Dec 14, 2019 · b7f67fd · b7f67fd
1 parent 02df077
commit b7f67fd
Show file tree

Hide file tree

Showing 7 changed files with 171 additions and 214 deletions.
diff --git a/inc/lazyload_replacer.php b/inc/lazyload_replacer.php
@@ -183,7 +183,7 @@ public function lazyload_tag_replace( $new_tag, $original_url, $new_url, $optml_
 		if ( ! self::$is_lazyload_placeholder && ! $should_ignore_rescale ) {
 			$optml_args['quality'] = 'eco';
 			$optml_args['resize']  = [];
-			$low_url               = apply_filters( 'optml_content_url', $is_slashed ? stripslashes( $original_url ) : $original_url, $optml_args );
+			$low_url               = apply_filters( 'optml_content_url', $original_url, $optml_args );
 			$low_url               = $is_slashed ? addcslashes( $low_url, '/' ) : $low_url;
 		} else {
 			$low_url = $this->get_svg_for(
@@ -221,12 +221,12 @@ public function lazyload_tag_replace( $new_tag, $original_url, $new_url, $optml_
 		);
 		$new_tag       = preg_replace(
 			[
-				'/( src(?>=|"|\'|\s|\\\\)*)' . preg_quote( $original_url, '/' ) . '/m',
-				'/ src=/m',
+				'/((?:\s|\'|"){1,}src(?>=|"|\'|\s|\\\\)*)' . preg_quote( $original_url, '/' ) . '/m',
+				'/<img/im',
 			],
 			[
 				"$1$low_url",
-				$opt_src . ' src=',
+				'<img' . $opt_src,
 			],
 			$new_tag,
 			1

diff --git a/inc/manager.php b/inc/manager.php
@@ -260,107 +260,13 @@ public function replace_meta( $metadata, $object_id, $meta_key, $single ) {
 				return $metadata;
 			}
 
-			return $this->process_urls_from_json( $current_meta );
+			return $this->replace_content( $current_meta );
 		}
 
 		// Return original if the check does not pass
 		return $metadata;
 	}
 
-	/**
-	 * Process json string.
-	 *
-	 * @param string $json Json string.
-	 *
-	 * @return string Processed string.
-	 */
-	public function process_urls_from_json( $json ) {
-
-		$extracted_urls = $this->extract_urls_from_json( $json );
-
-		return $this->do_url_replacement( $json, $extracted_urls );
-	}
-
-	/**
-	 * Extract urls used as values in json string, i.e not prefixed by =("|') char.
-	 *
-	 * @param string $content Raw json string.
-	 *
-	 * @return array array of urls.
-	 */
-	public function extract_urls_from_json( $content ) {
-		$regex = '/(?<!(=|\\\\)(?:"|\'|"))(?:http(?:s?):)(?:[\/\\\\|.|\w|\s|@|%|-])*\.(?:' . implode( '|', array_keys( Optml_Config::$extensions ) ) . ')(?:\??[\w|=|&|\-|\.|:]*)/';
-		preg_match_all(
-			$regex,
-			$content,
-			$urls
-		);
-
-		return $this->normalize_urls( $urls[0] );
-	}
-
-	/**
-	 * Normalize extracted urls.
-	 *
-	 * @param array $urls Raw urls extracted.
-	 *
-	 * @return array Normalized array.
-	 */
-	private function normalize_urls( $urls ) {
-
-		$urls = array_map(
-			function ( $value ) {
-				$value = str_replace( '&quot;', '', $value );
-
-				return rtrim( $value, '\\";\'' );
-			},
-			$urls
-		);
-		$urls = array_unique( $urls );
-
-		return array_values( $urls );
-	}
-
-	/**
-	 * Process string content and replace possible urls.
-	 *
-	 * @param string $html String content.
-	 * @param array  $extracted_urls Urls to check.
-	 *
-	 * @return string Processed html.
-	 */
-	private function do_url_replacement( $html, $extracted_urls ) {
-		$extracted_urls = apply_filters( 'optml_extracted_urls', $extracted_urls );
-
-		if ( empty( $extracted_urls ) ) {
-			return $html;
-		}
-
-		$upload_resource = $this->tag_replacer->get_upload_resource();
-		$urls            = array_combine( $extracted_urls, $extracted_urls );
-		$urls            = array_map(
-			function ( $url ) use ( $upload_resource ) {
-				$is_relative = strpos( $url, $upload_resource['content_path'] ) === 0;
-				if ( $is_relative ) {
-					$url = $upload_resource['content_host'] . $url;
-				}
-
-				$is_slashed = strpos( $url, '\/' ) !== false;
-				$url        = html_entity_decode( $url );
-				$new_url    = apply_filters( 'optml_content_url', $url );
-
-				return $is_slashed ? addcslashes( $new_url, '/' ) : $new_url;
-			},
-			$urls
-		);
-
-		foreach ( $urls as $origin => $replace ) {
-			$html = preg_replace( '/(?<![\/|:|\\w])' . preg_quote( $origin, '/' ) . '/m', $replace, $html );
-		}
-
-		return $html;
-	}
-
 	/**
 	 * Filter raw HTML content for urls.
 	 *
@@ -444,7 +350,7 @@ public static function parse_images_from_html( $content ) {
 			$header_end   = $header_start + strlen( $matches[0][0] );
 		}
 
-		if ( preg_match_all( '/(?:<a[^>]+?href=["|\'](?P<link_url>[^\s]+?)["|\'][^>]*?>\s*)?(?P<img_tag>(?:<noscript\s*>\s*)?<img[^>]*?\s+?(?:' . implode( '|', array_merge( [ 'src' ], Optml_Tag_Replacer::possible_src_attributes() ) ) . ')=\\\\?["|\'](?P<img_url>[^\s]+?)["|\'].*?>(?:\s*<\/noscript\s*>)?){1}(?:\s*<\/a>)?/ism', $content, $images, PREG_OFFSET_CAPTURE ) ) {
+		if ( preg_match_all( '/(?:<a[^>]+?href=["|\'](?P<link_url>[^\s]+?)["|\'][^>]*?>\s*)?(?P<img_tag>(?:<noscript\s*>\s*)?<img[^>]*?\s?(?:' . implode( '|', array_merge( [ 'src' ], Optml_Tag_Replacer::possible_src_attributes() ) ) . ')=\\\\?["|\'](?P<img_url>[^\s]+?)["|\'].*?>(?:\s*<\/noscript\s*>)?){1}(?:\s*<\/a>)?/ism', $content, $images, PREG_OFFSET_CAPTURE ) ) {
 
 			foreach ( $images as $key => $unused ) {
 				// Simplify the output as much as possible, mostly for confirming test results.
@@ -514,6 +420,79 @@ public function extract_image_urls_from_content( $content ) {
 		return $this->normalize_urls( $urls[1] );
 	}
 
+	/**
+	 * Normalize extracted urls.
+	 *
+	 * @param array $urls Raw urls extracted.
+	 *
+	 * @return array Normalized array.
+	 */
+	private function normalize_urls( $urls ) {
+
+		$urls = array_map(
+			function ( $value ) {
+				$value = str_replace( '&quot;', '', $value );
+
+				return rtrim( $value, '\\";\'' );
+			},
+			$urls
+		);
+		$urls = array_unique( $urls );
+
+		return array_values( $urls );
+	}
+
+	/**
+	 * Process string content and replace possible urls.
+	 *
+	 * @param string $html String content.
+	 * @param array  $extracted_urls Urls to check.
+	 *
+	 * @return string Processed html.
+	 */
+	private function do_url_replacement( $html, $extracted_urls ) {
+		$extracted_urls = apply_filters( 'optml_extracted_urls', $extracted_urls );
+
+		if ( empty( $extracted_urls ) ) {
+			return $html;
+		}
+		$slashed_config = addcslashes( Optml_Config::$service_url, '/' );
+
+		$extracted_urls  = array_filter(
+			$extracted_urls,
+			function ( $value ) use ( $slashed_config ) {
+				return strpos( $value, Optml_Config::$service_url ) === false && strpos( $value, $slashed_config ) === false;
+			}
+		);
+		$upload_resource = $this->tag_replacer->get_upload_resource();
+		$urls            = array_combine( $extracted_urls, $extracted_urls );
+
+		$urls = array_map(
+			function ( $url ) use ( $upload_resource ) {
+				$is_slashed  = strpos( $url, '\/' ) !== false;
+				$is_relative = strpos(
+					$url,
+					$is_slashed ?
+							addcslashes( $upload_resource['content_path'], '/' ) :
+							$upload_resource['content_path']
+				) === 0;
+				if ( $is_relative ) {
+					$url = $upload_resource['content_host'] . $url;
+				}
+				$new_url = apply_filters( 'optml_content_url', $url );
+
+				return $new_url;
+			},
+			$urls
+		);
+
+		foreach ( $urls as $origin => $replace ) {
+			$html = preg_replace( '/(?<![\/|:|\\w])' . preg_quote( $origin, '/' ) . '/m', $replace, $html );
+		}
+
+		return $html;
+	}
+
 	/**
 	 * Init html replacer handler.
 	 */

diff --git a/inc/tag_replacer.php b/inc/tag_replacer.php
@@ -143,7 +143,7 @@ public function process_image_tags( $content, $images = array() ) {
 
 			$is_slashed = strpos( $images['img_url'][ $index ], '\/' ) !== false;
 
-			$src = $tmp = $is_slashed ? stripslashes( $images['img_url'][ $index ] ) : $images['img_url'][ $index ];
+			$src = $tmp = $is_slashed ? $this->strip_slashes( $images['img_url'][ $index ] ) : $images['img_url'][ $index ];
 
 			if ( strpos( $src, $this->upload_resource['content_path'] ) === 0 ) {
 				$src = $tmp = untrailingslashit( $this->upload_resource['content_host'] ) . $src;

diff --git a/inc/traits/normalizer.php b/inc/traits/normalizer.php
@@ -27,6 +27,16 @@ public function to_boolean( $value ) {
 		return boolval( $value );
 	}
 
+	/**
+	 * Strip slashes on unicode encoded strings.
+	 *
+	 * @param string $string Input string.
+	 *
+	 * @return string Decoded string.
+	 */
+	public function strip_slashes( $string ) {
+		return html_entity_decode( stripslashes( preg_replace( '/\\\u([\da-fA-F]{4})/', '&#x\1;', $string ) ) );
+	}
 	/**
 	 * Normalize value to an integer within bounds.
 	 *

diff --git a/inc/url_replacer.php b/inc/url_replacer.php
@@ -118,8 +118,7 @@ public function build_image_url(
 
 		// We do a little hack here, for json unicode chars we first replace them with html special chars,
 		// we then strip slashes to normalize the URL and last we convert html special chars back to get a clean URL
-		$url = $is_slashed ? html_entity_decode( stripslashes( preg_replace( '/\\\u([\da-fA-F]{4})/', '&#x\1;', $url ) ) ) : $url;
-
+		$url = $is_slashed ? html_entity_decode( stripslashes( preg_replace( '/\\\u([\da-fA-F]{4})/', '&#x\1;', $url ) ) ) : ( $url );
 		if ( strpos( $url, Optml_Config::$service_url ) !== false ) {
 			return $original_url;
 		}

diff --git a/tests/test-lazyload.php b/tests/test-lazyload.php
@@ -320,6 +320,28 @@ public function test_lazyload_json_data_disabled() {
 		$this->assertEquals( 2, substr_count( $replaced_content2, '/http:\/\/example.org' ) );
 	}
 
+	public function test_json_lazyload_replacement() {
+		$html = Test_Replacer::get_html_array();
+
+		$replaced_content = Optml_Manager::instance()->replace_content( json_encode( $html ) );
+		$this->assertContains( 'i.optimole.com', $replaced_content );
+		$this->assertEquals( ( 6 + ( 3 * 48 ) ), substr_count( $replaced_content, 'i.optimole.com' ) );
+
+		$this->assertTrue( is_array( json_decode( $replaced_content, true ) ) );
+		$this->assertNotContains( "\"https:\/\/www.example.org\/wp-content", $replaced_content );
+		$this->assertNotContains( "\"\/\/www.example.org\/wp-content", $replaced_content );
+		$this->assertNotContains( "\"\/wp-content", $replaced_content );
+		$count_unicode = 0;
+		$replaced_html = json_decode( $replaced_content, true );
+
+		foreach ( $replaced_html as $value ) {
+			$count_unicode += substr_count( $value, Test_Replacer::DECODED_UNICODE );
+
+		}
+
+		$this->assertEquals( $count_unicode, ( ( 24 * 3 ) + 3 ) );
+	}
+
 	public function test_should_replace_query_string_url() {
 		$content          = '<img src="https://example.org/photos/814499/pexels-photo-814499.jpeg?auto=compress&cs=tinysrgb&dpr=1&w=500" alt="">';
 		$replaced_content = Optml_Manager::instance()->replace_content( $content );