From 25c8fe4eeb1e9c1a3055bb27984f1274f99fc1d7 Mon Sep 17 00:00:00 2001
From: Aleksander Machniak <alec@alec.pl>
Date: Wed, 17 Jun 2015 08:03:13 -0400
Subject: [PATCH] Fix handling of non-break spaces in html to text conversion (#1490436)
---
program/lib/Roundcube/rcube_html2text.php | 20 ++++++++++++++++----
1 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/program/lib/Roundcube/rcube_html2text.php b/program/lib/Roundcube/rcube_html2text.php
index 01362e6..a2f6288 100644
--- a/program/lib/Roundcube/rcube_html2text.php
+++ b/program/lib/Roundcube/rcube_html2text.php
@@ -142,7 +142,7 @@
'/<script[^>]*>.*?<\/script>/i', // <script>s -- which strip_tags supposedly has problems with
'/<style[^>]*>.*?<\/style>/i', // <style>s -- which strip_tags supposedly has problems with
'/<p[^>]*>/i', // <P>
- '/<br[^>]*>/i', // <br>
+ '/<br[^>]*>\s*/i', // <br>
'/<i[^>]*>(.*?)<\/i>/i', // <i>
'/<em[^>]*>(.*?)<\/em>/i', // <em>
'/(<ul[^>]*>|<\/ul>)/i', // <ul> and </ul>
@@ -216,7 +216,7 @@
* @see $ent_search
*/
protected $ent_replace = array(
- ' ', // Non-breaking space
+ "\xC2\xA0", // Non-breaking space
'"', // Double quotes
"'", // Single quotes
'>',
@@ -423,7 +423,7 @@
// Variables used for building the link list
$this->_link_list = array();
- $text = trim(stripslashes($this->html));
+ $text = $this->html;
// Convert HTML to TXT
$this->_converter($text);
@@ -473,6 +473,9 @@
// Replace known html entities
$text = html_entity_decode($text, ENT_QUOTES, $this->charset);
+ // Replace unicode nbsp to regular spaces
+ $text = preg_replace('/\xC2\xA0/', ' ', $text);
+
// Remove unknown/unhandled entities (this cannot be done in search-and-replace block)
$text = preg_replace('/&([a-zA-Z0-9]{2,6}|#[0-9]{2,4});/', '', $text);
@@ -506,7 +509,7 @@
* @param string $link URL of the link
* @param string $display Part of the text to associate number with
*/
- protected function _build_link_list( $link, $display )
+ protected function _build_link_list($link, $display)
{
if (!$this->_do_links || empty($link)) {
return $display;
@@ -514,6 +517,11 @@
// Ignored link types
if (preg_match('!^(javascript:|mailto:|#)!i', $link)) {
+ return $display;
+ }
+
+ // skip links with href == content (#1490434)
+ if ($link === $display) {
return $display;
}
@@ -616,6 +624,10 @@
break;
}
+ // abort on invalid tag structure (e.g. no closing tag found)
+ else {
+ break;
+ }
}
while ($end || $next);
}
--
Gitblit v1.9.1