From 96484fd3f918650227d0ededfbaed1c6528ad299 Mon Sep 17 00:00:00 2001
From: Scott Taylor <wonderboymusic@git.wordpress.org>
Date: Wed, 25 Jun 2014 17:48:20 +0000
Subject: [PATCH] Optimize the `wptexturize()` loop: * Take the ampersand
 pattern out of the loop for speed. * Fix old bugs in the ampersand pattern. *
 Refactor `_wptexturize_pushpop_element()` without PCRE for speed. * Update
 unit tests.

Props miqrogroove.
Fixes #28623.


git-svn-id: https://develop.svn.wordpress.org/trunk@28831 602fd350-edb4-49c9-b593-d223f7449a82
---
 src/wp-includes/formatting.php                | 86 +++++++++++--------
 .../phpunit/tests/formatting/WPTexturize.php  | 26 +++++-
 2 files changed, 72 insertions(+), 40 deletions(-)

diff --git a/src/wp-includes/formatting.php b/src/wp-includes/formatting.php
index a073a7cde9..5af533932f 100644
--- a/src/wp-includes/formatting.php
+++ b/src/wp-includes/formatting.php
@@ -178,7 +178,7 @@ function wptexturize($text, $reset = false) {
 	 *
 	 * @param array $default_no_texturize_tags An array of HTML element names.
 	 */
-	$no_texturize_tags = '(' . implode( '|', apply_filters( 'no_texturize_tags', $default_no_texturize_tags ) ) . ')';
+	$no_texturize_tags = apply_filters( 'no_texturize_tags', $default_no_texturize_tags );
 	/**
 	 * Filter the list of shortcodes not to texturize.
 	 *
@@ -186,7 +186,7 @@ function wptexturize($text, $reset = false) {
 	 *
 	 * @param array $default_no_texturize_shortcodes An array of shortcode names.
 	 */
-	$no_texturize_shortcodes = '(' . implode( '|', apply_filters( 'no_texturize_shortcodes', $default_no_texturize_shortcodes ) ) . ')';
+	$no_texturize_shortcodes = apply_filters( 'no_texturize_shortcodes', $default_no_texturize_shortcodes );
 
 	$no_texturize_tags_stack = array();
 	$no_texturize_shortcodes_stack = array();
@@ -206,7 +206,7 @@ function wptexturize($text, $reset = false) {
 		.	'(?:'
 		.		'[^\[\]<>]'	// Shortcodes do not contain other shortcodes.
 		.	'|'
-		.		'<.+?>' // HTML elements permitted. Prevents matching ] before >.
+		.		'<.+?>' 	// HTML elements permitted. Prevents matching ] before >.
 		.	')+'
 		.	'\]'		// Find end of shortcode.
 		.	'\]?'		// Shortcodes may end with ]]
@@ -221,13 +221,13 @@ function wptexturize($text, $reset = false) {
 			// This is an HTML delimeter.
 
 			if ( '<!--' !== substr( $curl, 0, 4 ) ) {
-				_wptexturize_pushpop_element( $curl, $no_texturize_tags_stack, $no_texturize_tags, '<', '>' );
+				_wptexturize_pushpop_element( $curl, $no_texturize_tags_stack, $no_texturize_tags );
 			}
 
 		} elseif ( '[' === $first && 1 === preg_match( '/^\[(?:[^\[\]<>]|<.+?>)+\]$/', $curl ) ) {
 			// This is a shortcode delimeter.
 
-			_wptexturize_pushpop_element( $curl, $no_texturize_shortcodes_stack, $no_texturize_shortcodes, '[', ']' );
+			_wptexturize_pushpop_element( $curl, $no_texturize_shortcodes_stack, $no_texturize_shortcodes );
 
 		} elseif ( '[' === $first && 1 === preg_match( '/^\[\[?(?:[^\[\]<>]|<.+?>)+\]\]?$/', $curl ) ) {
 			// This is an escaped shortcode delimeter.
@@ -235,11 +235,11 @@ function wptexturize($text, $reset = false) {
 			// Do not texturize.
 			// Do not push to the shortcodes stack.
 
-		} elseif ( empty($no_texturize_shortcodes_stack) && empty($no_texturize_tags_stack) ) {
+		} elseif ( empty( $no_texturize_shortcodes_stack ) && empty( $no_texturize_tags_stack ) ) {
 			// This is neither a delimeter, nor is this content inside of no_texturize pairs.  Do texturize.
 
-			$curl = str_replace($static_characters, $static_replacements, $curl);
-			$curl = preg_replace($dynamic_characters, $dynamic_replacements, $curl);
+			$curl = str_replace( $static_characters, $static_replacements, $curl );
+			$curl = preg_replace( $dynamic_characters, $dynamic_replacements, $curl );
 
 			// 9x9 (times), but never 0x9999
 			if ( 1 === preg_match( '/(?<=\d)x-?\d/', $curl ) ) {
@@ -247,31 +247,54 @@ function wptexturize($text, $reset = false) {
 				$curl = preg_replace( '/\b(\d(?(?<=0)[\d\.,]+|[\d\.,]*))x(-?\d[\d\.,]*)\b/', '$1&#215;$2', $curl );
 			}
 		}
-
-		// Replace each & with &#038; unless it already looks like an entity.
-		$curl = preg_replace('/&([^#])(?![a-zA-Z1-4]{1,8};)/', '&#038;$1', $curl);
 	}
-	return implode( '', $textarr );
+	$text = implode( '', $textarr );
+
+	// Replace each & with &#038; unless it already looks like an entity.
+	$text = preg_replace('/&(?!#(?:\d+|x[a-f0-9]+);|[a-z1-4]{1,8};)/i', '&#038;', $text);
+
+	return $text;
 }
 
 /**
  * Search for disabled element tags. Push element to stack on tag open and pop
- * on tag close. Assumes first character of $text is tag opening.
+ * on tag close.
+ *
+ * Assumes first char of $text is tag opening and last char is tag closing.
+ * Assumes second char of $text is optionally '/' to indicate closing as in </html>.
  *
  * @since 2.9.0
  * @access private
  *
- * @param string $text Text to check. First character is assumed to be $opening
- * @param array $stack Array used as stack of opened tag elements
- * @param string $disabled_elements Tags to match against formatted as regexp sub-expression
- * @param string $opening Tag opening character, assumed to be 1 character long
- * @param string $closing Tag closing character
+ * @param string $text Text to check. Must be a tag like <html> or [shortcode].
+ * @param array $stack List of open tag elements.
+ * @param array $disabled_elements The tag names to match against. Spaces are not allowed in tag names.
  */
-function _wptexturize_pushpop_element($text, &$stack, $disabled_elements, $opening = '<', $closing = '>') {
-	// Check if it is a closing tag -- otherwise assume opening tag
-	if (strncmp($opening . '/', $text, 2)) {
-		// Opening? Check $text+1 against disabled elements
-		if (preg_match('/^' . $disabled_elements . '\b/', substr($text, 1), $matches)) {
+function _wptexturize_pushpop_element($text, &$stack, $disabled_elements) {
+	// Is it an opening tag or closing tag?
+	if ( '/' !== $text[1] ) {
+		$opening_tag = true;
+		$name_offset = 1;
+	} elseif ( 0 == count( $stack ) ) {
+		// Stack is empty. Just stop.
+		return;
+	} else {
+		$opening_tag = false;
+		$name_offset = 2;
+	}
+
+	// Parse out the tag name.
+	$space = strpos( $text, ' ' );
+	if ( FALSE === $space ) {
+		$space = -1;
+	} else {
+		$space -= $name_offset;
+	}
+	$tag = substr( $text, $name_offset, $space );
+
+	// Handle disabled tags.
+	if ( in_array( $tag, $disabled_elements ) ) {
+		if ( $opening_tag ) {
 			/*
 			 * This disables texturize until we find a closing tag of our type
 			 * (e.g. <pre>) even if there was invalid nesting before that
@@ -280,20 +303,9 @@ function _wptexturize_pushpop_element($text, &$stack, $disabled_elements, $openi
 			 *          "baba" won't be texturize
 			 */
 
-			array_push($stack, $matches[1]);
-		}
-	} elseif ( 0 == count( $stack ) ) {
-		// Stack is empty. Just stop.
-	} else {
-		// Closing? Check $text+2 against disabled elements
-		$c = preg_quote($closing, '/');
-		if (preg_match('/^' . $disabled_elements . $c . '/', substr($text, 2), $matches)) {
-			$last = array_pop($stack);
-
-			// Make sure it matches the opening tag
-			if ( $last != $matches[1] ) {
-				array_push( $stack, $last );
-			}
+			array_push( $stack, $tag );
+		} elseif ( end( $stack ) == $tag ) {
+			array_pop( $stack );
 		}
 	}
 }
diff --git a/tests/phpunit/tests/formatting/WPTexturize.php b/tests/phpunit/tests/formatting/WPTexturize.php
index 2196134939..ea22b62500 100644
--- a/tests/phpunit/tests/formatting/WPTexturize.php
+++ b/tests/phpunit/tests/formatting/WPTexturize.php
@@ -829,9 +829,17 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
 				"word &#038; word",
 				"word &#038; word",
 			),
+			array(
+				"word &#xabc; word",
+				"word &#xabc; word",
+			),
+			array(
+				"word &#X394; word",
+				"word &#X394; word",
+			),
 			array(
 				"word &# word",
-				"word &# word", // invalid output?
+				"word &#038;# word",
 			),
 			array(
 				"word &44; word",
@@ -839,11 +847,19 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
 			),
 			array(
 				"word &&amp; word",
-				"word &&amp; word",
+				"word &#038;&amp; word",
 			),
 			array(
 				"word &!amp; word",
-				"word &!amp; word",
+				"word &#038;!amp; word",
+			),
+			array(
+				"word &#",
+				"word &#038;#",
+			),
+			array(
+				"word &",
+				"word &#038;",
 			),
 		);
 	}
@@ -1285,6 +1301,10 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
 				'<ul><li>Hello.</li><!--<li>Goodbye.</li>--></ul>',
 				'<ul><li>Hello.</li><!--<li>Goodbye.</li>--></ul>',
 			),
+			array(
+				'word <img src="http://example.com/wp-content/uploads/2014/06/image-300x216.gif" /> word', // Ensure we are not corrupting image URLs.
+				'word <img src="http://example.com/wp-content/uploads/2014/06/image-300x216.gif" /> word',
+			),
 		);
 	}