`wp_extract_urls()` needs to also grab URLs with query strings and hashes.

Updates unit test.

Props hauvong, dlh.
Fixes #29314.


git-svn-id: https://develop.svn.wordpress.org/trunk@29756 602fd350-edb4-49c9-b593-d223f7449a82

src/wp-includes/functions.php

@@ -457,23 +457,23 @@ function xmlrpc_removepostdata( $content ) {
  */
 function wp_extract_urls( $content ) {
 	preg_match_all(
-		"#("
+		"#([\"']?)("
 			. "(?:([\w-]+:)?//?)"
 			. "[^\s()<>]+"
 			. "[.]"
 			. "(?:"
 				. "\([\w\d]+\)|"
 				. "(?:"
-					. "[^`!()\[\]{};:'\".,<>?«»“”‘’\s]|"
+					. "[^`!()\[\]{};:'\".,<>«»“”‘’\s]|"
 					. "(?:[:]\d+)?/?"
 				. ")+"
 			. ")"
-		. ")#",
+		. ")\\1#",
 		$content,
 		$post_links
 	);
 
-	$post_links = array_unique( array_map( 'html_entity_decode', $post_links[0] ) );
+	$post_links = array_unique( array_map( 'html_entity_decode', $post_links[2] ) );
 
 	return array_values( $post_links );
 }

tests/phpunit/tests/functions.php

@@ -421,7 +421,9 @@ class Tests_Functions extends WP_UnitTestCase {
 			'http://türkiye.icom.museum',
 			'http://украина.icom.museum',
 			'http://việtnam.icom.museum',
-			'ftp://127.0.0.1/'
+			'ftp://127.0.0.1/',
+			'http://www.woo.com/video?v=exvUH2qKLTU',
+			'http://taco.com?burrito=enchilada#guac'
 		);
 
 		$blob ="
@@ -481,6 +483,9 @@ class Tests_Functions extends WP_UnitTestCase {
 			http://украина.icom.museum
 			http://việtnam.icom.museum
 			ftp://127.0.0.1/
+			http://www.woo.com/video?v=exvUH2qKLTU
+
+			http://taco.com?burrito=enchilada#guac
 		";
 
 		$urls = wp_extract_urls( $blob );