From fb28a78331016bb5850b63a4935b6285176cc21a Mon Sep 17 00:00:00 2001
From: Andrew Nacin <nacin@git.wordpress.org>
Date: Thu, 20 Nov 2014 14:27:08 +0000
Subject: [PATCH] Anchor texturize to shortcodes to improve regex efficiency.

props miqrogroove.
see #29557 for segfault issues.


git-svn-id: https://develop.svn.wordpress.org/trunk@30449 602fd350-edb4-49c9-b593-d223f7449a82
---
 src/wp-includes/formatting.php                | 22 +++++++++++++------
 .../phpunit/tests/formatting/WPTexturize.php  | 16 +++++++-------
 2 files changed, 23 insertions(+), 15 deletions(-)

diff --git a/src/wp-includes/formatting.php b/src/wp-includes/formatting.php
index 57cfcb1eac..5209124c4d 100644
--- a/src/wp-includes/formatting.php
+++ b/src/wp-includes/formatting.php
@@ -28,7 +28,7 @@
  * @return string The string replaced with html entities
  */
 function wptexturize($text, $reset = false) {
-	global $wp_cockneyreplace;
+	global $wp_cockneyreplace, $shortcode_tags;
 	static $static_characters, $static_replacements, $dynamic_characters, $dynamic_replacements,
 		$default_no_texturize_tags, $default_no_texturize_shortcodes, $run_texturize = true;
 
@@ -205,6 +205,10 @@ function wptexturize($text, $reset = false) {
 
 	// Look for shortcodes and HTML elements.
 
+	$tagnames = array_keys( $shortcode_tags );
+	$tagregexp = join( '|', array_map( 'preg_quote', $tagnames ) );
+	$tagregexp = "(?:$tagregexp)(?![\\w-])"; // Excerpt of get_shortcode_regex().
+
 	$comment_regex =
 		  '!'           // Start of comment, after the <.
 		. '(?:'         // Unroll the loop: Consume everything until --> is found.
@@ -214,12 +218,16 @@ function wptexturize($text, $reset = false) {
 		. '(?:-->)?';   // End of comment. If not found, match all input.
 
 	$shortcode_regex =
-		  '\['          // Find start of shortcode.
-		. '[\/\[]?'     // Shortcodes may begin with [/ or [[
-		. '[^\s\/\[\]]' // No whitespace before name.
-		. '[^\[\]]*+'   // Shortcodes do not contain other shortcodes. Possessive critical.
-		. '\]'          // Find end of shortcode.
-		. '\]?';        // Shortcodes may end with ]]
+		  '\['              // Find start of shortcode.
+		. '[\/\[]?'         // Shortcodes may begin with [/ or [[
+		. $tagregexp        // Only match registered shortcodes, because performance.
+		. '(?:'
+		.     '[^\[\]<>]+'  // Shortcodes do not contain other shortcodes. Quantifier critical.
+		. '|'
+		.     '<[^\[\]>]*>' // HTML elements permitted. Prevents matching ] before >.
+		. ')*+'             // Possessive critical.
+		. '\]'              // Find end of shortcode.
+		. '\]?';            // Shortcodes may end with ]]
 
 	$regex =
 		  '/('                   // Capture the entire match.
diff --git a/tests/phpunit/tests/formatting/WPTexturize.php b/tests/phpunit/tests/formatting/WPTexturize.php
index 1bc197bfef..1187f1eb52 100644
--- a/tests/phpunit/tests/formatting/WPTexturize.php
+++ b/tests/phpunit/tests/formatting/WPTexturize.php
@@ -1196,12 +1196,12 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
 				'[ is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]',
 			),
 			array(
-				'[is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]', // HTML corruption is a known bug.  See tickets #12690 and #29557.
-				'[is it wise to <a title="allow user content ] here? hmm&#8221;> maybe </a> ]',
+				'[is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]',
+				'[is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]',
 			),
 			array(
 				'[caption - is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]',
-				'[caption - is it wise to <a title="allow user content ] here? hmm&#8221;> maybe </a> ]',
+				'[caption &#8211; is it wise to <a title="allow user content ] here? hmm"> maybe </a> ]',
 			),
 			array(
 				'[ photos by <a href="http://example.com/?a[]=1&a[]=2"> this guy </a> ]',
@@ -1767,23 +1767,23 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
 			),
 			array(
 				'[code ...]...[/code]', // code is not a registered shortcode.
-				'[code ...]...[/code]',
+				'[code &#8230;]&#8230;[/code]',
 			),
 			array(
 				'[hello ...]...[/hello]', // hello is not a registered shortcode.
-				'[hello ...]&#8230;[/hello]',
+				'[hello &#8230;]&#8230;[/hello]',
 			),
 			array(
 				'[...]...[/...]', // These are potentially usable shortcodes.
-				'[...]&#8230;[/...]',
+				'[&#8230;]&#8230;[/&#8230;]',
 			),
 			array(
 				'[gal>ery ...]',
-				'[gal>ery ...]',
+				'[gal>ery &#8230;]',
 			),
 			array(
 				'[randomthing param="test"]',
-				'[randomthing param="test"]',
+				'[randomthing param=&#8221;test&#8221;]',
 			),
 			array(
 				'[[audio]...[/audio]...', // These are potentially usable shortcodes.  Unfortunately, the meaning of [[audio] is ambiguous unless we run the entire shortcode regexp.