From 91dc7fabbc15c1e3dc932da327bbd052149ab8af Mon Sep 17 00:00:00 2001
From: Aleksander Machniak <alec@alec.pl>
Date: Thu, 15 Oct 2015 06:23:23 -0400
Subject: [PATCH] Small improvements in HTML to text conversion.
---
program/lib/Roundcube/rcube_html2text.php | 57 +++++++++++++++++++++++++++++++++++++++------------------
1 files changed, 39 insertions(+), 18 deletions(-)
diff --git a/program/lib/Roundcube/rcube_html2text.php b/program/lib/Roundcube/rcube_html2text.php
index 9b248a3..d20d7b7 100644
--- a/program/lib/Roundcube/rcube_html2text.php
+++ b/program/lib/Roundcube/rcube_html2text.php
@@ -136,13 +136,16 @@
* @see $replace
*/
protected $search = array(
- "/\r/", // Non-legal carriage return
- "/[\n\t]+/", // Newlines and tabs
+ '/\r/', // Non-legal carriage return
+ '/^.*<body[^>]*>\n*/i', // Anything before <body>
'/<head[^>]*>.*?<\/head>/i', // <head>
- '/<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>
+ '/<script[^>]*>.*?<\/script>/i', // <script>
+ '/<style[^>]*>.*?<\/style>/i', // <style>
+ '/[\n\t]+/', // Newlines and tabs
+ '/<p[^>]*>/i', // <p>
+ '/<\/p>[\s\n\t]*<div[^>]*>/i', // </p> before <div>
+ '/<br[^>]*>[\s\n\t]*<div[^>]*>/i', // <br> before <div>
+ '/<br[^>]*>\s*/i', // <br>
'/<i[^>]*>(.*?)<\/i>/i', // <i>
'/<em[^>]*>(.*?)<\/em>/i', // <em>
'/(<ul[^>]*>|<\/ul>)/i', // <ul> and </ul>
@@ -164,11 +167,14 @@
*/
protected $replace = array(
'', // Non-legal carriage return
- ' ', // Newlines and tabs
+ '', // Anything before <body>
'', // <head>
- '', // <script>s -- which strip_tags supposedly has problems with
- '', // <style>s -- which strip_tags supposedly has problems with
- "\n\n", // <P>
+ '', // <script>
+ '', // <style>
+ ' ', // Newlines and tabs
+ "\n\n", // <p>
+ "\n<div>", // </p> before <div>
+ '<div>', // <br> before <div>
"\n", // <br>
'_\\1_', // <i>
'_\\1_', // <em>
@@ -216,7 +222,7 @@
* @see $ent_search
*/
protected $ent_replace = array(
- ' ', // Non-breaking space
+ "\xC2\xA0", // Non-breaking space
'"', // Double quotes
"'", // Single quotes
'>',
@@ -423,7 +429,7 @@
// 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 +479,9 @@
// 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 +515,7 @@
* @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 +523,11 @@
// Ignored link types
if (preg_match('!^(javascript:|mailto:|#)!i', $link)) {
+ return $display;
+ }
+
+ // skip links with href == content (#1490434)
+ if ($link === $display) {
return $display;
}
@@ -608,24 +622,31 @@
$this->width = $p_width;
// Add citation markers and create <pre> block
- $body = preg_replace_callback('/((?:^|\n)>*)([^\n]*)/', array($this, 'blockquote_citation_ballback'), trim($body));
+ $body = preg_replace_callback('/((?:^|\n)>*)([^\n]*)/', array($this, 'blockquote_citation_callback'), trim($body));
$body = '<pre>' . htmlspecialchars($body) . '</pre>';
- $text = substr($text, 0, $start) . $body . "\n" . substr($text, $end + 13);
+ $text = substr_replace($text, $body . "\n", $start, $end + 13 - $start);
$offset = 0;
+
break;
}
- } while ($end || $next);
+ // abort on invalid tag structure (e.g. no closing tag found)
+ else {
+ break;
+ }
+ }
+ while ($end || $next);
}
}
/**
* Callback function to correctly add citation markers for blockquote contents
*/
- public function blockquote_citation_ballback($m)
+ public function blockquote_citation_callback($m)
{
- $line = ltrim($m[2]);
+ $line = ltrim($m[2]);
$space = $line[0] == '>' ? '' : ' ';
+
return $m[1] . '>' . $space . $line;
}
--
Gitblit v1.9.1