Replace the ancient `phpfreaks.com` RegEx to extract urls to ping with a more robust matcher. URLs with commas and things like `&` were not being pinged. The new matcher even works for most IDN URLs. Adds unit tests.

Fixes #9064.


Built from https://develop.svn.wordpress.org/trunk@25313


git-svn-id: http://core.svn.wordpress.org/trunk@25275 1a063a9b-81f0-0310-95a4-ce76da25c4cd

wp-includes/comment.php

@@ -1827,17 +1827,9 @@ function pingback($content, $post_ID) {
 
 	$pung = get_pung($post_ID);
 
-	// Variables
-	$ltrs = '\w';
-	$gunk = '/#~:.?+=&%@!\-';
-	$punc = '.:?\-';
-	$any = $ltrs . $gunk . $punc;
-
 	// Step 1
 	// Parsing the post, external links (if any) are stored in the $post_links array
-	// This regexp comes straight from phpfreaks.com
+	$post_links_temp = wp_extract_urls( $content );
-	// http://www.phpfreaks.com/quickcode/Extract_All_URLs_on_a_Page/15.php
-	preg_match_all("{\b http : [$any] +? (?= [$punc] * [^$any] | $)}x", $content, $post_links_temp);
 
 	// Step 2.
 	// Walking thru the links array
@@ -1848,7 +1840,7 @@ function pingback($content, $post_ID) {
 	// http://dummy-weblog.org/post.php
 	// We don't wanna ping first and second types, even if they have a valid <link/>
 
-	foreach ( (array) $post_links_temp[0] as $link_test ) :
+	foreach ( (array) $post_links_temp as $link_test ) :
 		if ( !in_array($link_test, $pung) && (url_to_postid($link_test) != $post_ID) // If we haven't pung it already and it isn't a link to itself
 				&& !is_local_attachment($link_test) ) : // Also, let's never ping local attachments.
 			if ( $test = @parse_url($link_test) ) {

wp-includes/functions.php

@@ -392,6 +392,26 @@ function xmlrpc_removepostdata( $content ) {
 	return $content;
 }
 
+/**
+ * Use RegEx to extract URLs from arbitrary content
+ *
+ * @since 3.7.0
+ *
+ * @param string $content
+ * @return array URLs found in passed string
+ */
+function wp_extract_urls( $content ) {
+	preg_match_all(
+		"#((?:[\w-]+://?|[\w\d]+[.])[^\s()<>]+[.](?:\([\w\d]+\)|(?:[^`!()\[\]{};:'\".,<>?«»“”‘’\s]|(?:[:]\d+)?/?)+))#",
+		$content,
+		$post_links
+	);
+
+	$post_links = array_unique( array_map( 'html_entity_decode', $post_links[0] ) );
+
+	return array_values( $post_links );
+}
+
 /**
  * Check content for video and audio links to add as enclosures.
  *
@@ -417,22 +437,17 @@ function do_enclose( $content, $post_ID ) {
 
 	$pung = get_enclosed( $post_ID );
 
-	$ltrs = '\w';
+	$post_links_temp = wp_extract_urls( $content );
-	$gunk = '/#~:.?+=&%@!\-';
-	$punc = '.:?\-';
-	$any = $ltrs . $gunk . $punc;
-
-	preg_match_all( "{\b https? : [$any] +? (?= [$punc] * [^$any] | $)}x", $content, $post_links_temp );
 
 	foreach ( $pung as $link_test ) {
-		if ( !in_array( $link_test, $post_links_temp[0] ) ) { // link no longer in post
+		if ( ! in_array( $link_test, $post_links_temp ) ) { // link no longer in post
 			$mids = $wpdb->get_col( $wpdb->prepare("SELECT meta_id FROM $wpdb->postmeta WHERE post_id = %d AND meta_key = 'enclosure' AND meta_value LIKE (%s)", $post_ID, like_escape( $link_test ) . '%') );
 			foreach ( $mids as $mid )
 				delete_metadata_by_mid( 'post', $mid );
 		}
 	}
 
-	foreach ( (array) $post_links_temp[0] as $link_test ) {
+	foreach ( (array) $post_links_temp as $link_test ) {
 		if ( !in_array( $link_test, $pung ) ) { // If we haven't pung it already
 			$test = @parse_url( $link_test );
 			if ( false === $test )