From 3fd5e2edfc72cea56ede0512f081a67d547ed61f Mon Sep 17 00:00:00 2001
From: alecpl <alec@alec.pl>
Date: Wed, 16 Nov 2011 03:57:32 -0500
Subject: [PATCH] - Apply fixes from trunk up to r5433
---
program/lib/washtml.php | 63 +++++++++++++++++++++++--------
1 files changed, 46 insertions(+), 17 deletions(-)
diff --git a/program/lib/washtml.php b/program/lib/washtml.php
index 923c97e..9292aa1 100644
--- a/program/lib/washtml.php
+++ b/program/lib/washtml.php
@@ -75,22 +75,41 @@
* - changed $ignore_elements behaviour
* - added RFC2397 support
* - base URL support
+ * - invalid HTML comments removal before parsing
*/
class washtml
{
/* Allowed HTML elements (default) */
- static $html_elements = array('a', 'abbr', 'acronym', 'address', 'area', 'b', 'basefont', 'bdo', 'big', 'blockquote', 'br', 'caption', 'center', 'cite', 'code', 'col', 'colgroup', 'dd', 'del', 'dfn', 'dir', 'div', 'dl', 'dt', 'em', 'fieldset', 'font', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr', 'i', 'ins', 'label', 'legend', 'li', 'map', 'menu', 'nobr', 'ol', 'p', 'pre', 'q', 's', 'samp', 'small', 'span', 'strike', 'strong', 'sub', 'sup', 'table', 'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'tt', 'u', 'ul', 'var', 'wbr', 'img');
-
+ static $html_elements = array('a', 'abbr', 'acronym', 'address', 'area', 'b',
+ 'basefont', 'bdo', 'big', 'blockquote', 'br', 'caption', 'center',
+ 'cite', 'code', 'col', 'colgroup', 'dd', 'del', 'dfn', 'dir', 'div', 'dl',
+ 'dt', 'em', 'fieldset', 'font', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr', 'i',
+ 'ins', 'label', 'legend', 'li', 'map', 'menu', 'nobr', 'ol', 'p', 'pre', 'q',
+ 's', 'samp', 'small', 'span', 'strike', 'strong', 'sub', 'sup', 'table',
+ 'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'tt', 'u', 'ul', 'var', 'wbr', 'img',
+ // form elements
+ 'button', 'input', 'textarea', 'select', 'option', 'optgroup'
+ );
+
/* Ignore these HTML tags and their content */
static $ignore_elements = array('script', 'applet', 'embed', 'object', 'style');
-
+
/* Allowed HTML attributes */
- static $html_attribs = array('name', 'class', 'title', 'alt', 'width', 'height', 'align', 'nowrap', 'col', 'row', 'id', 'rowspan', 'colspan', 'cellspacing', 'cellpadding', 'valign', 'bgcolor', 'color', 'border', 'bordercolorlight', 'bordercolordark', 'face', 'marginwidth', 'marginheight', 'axis', 'border', 'abbr', 'char', 'charoff', 'clear', 'compact', 'coords', 'vspace', 'hspace', 'cellborder', 'size', 'lang', 'dir');
+ static $html_attribs = array('name', 'class', 'title', 'alt', 'width', 'height',
+ 'align', 'nowrap', 'col', 'row', 'id', 'rowspan', 'colspan', 'cellspacing',
+ 'cellpadding', 'valign', 'bgcolor', 'color', 'border', 'bordercolorlight',
+ 'bordercolordark', 'face', 'marginwidth', 'marginheight', 'axis', 'border',
+ 'abbr', 'char', 'charoff', 'clear', 'compact', 'coords', 'vspace', 'hspace',
+ 'cellborder', 'size', 'lang', 'dir',
+ // attributes of form elements
+ 'type', 'rows', 'cols', 'disabled', 'readonly', 'checked', 'multiple', 'value'
+ );
/* Block elements which could be empty but cannot be returned in short form (<tag />) */
- static $block_elements = array('div', 'p', 'pre', 'blockquote', 'a', 'font');
-
+ static $block_elements = array('div', 'p', 'pre', 'blockquote', 'a', 'font', 'center',
+ 'table', 'ul', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'dl', 'strong', 'i', 'b');
+
/* State for linked objects in HTML */
public $extlinks = false;
@@ -99,7 +118,7 @@
/* Registered callback functions for tags */
private $handlers = array();
-
+
/* Allowed HTML elements */
private $_html_elements = array();
@@ -111,7 +130,7 @@
/* Allowed HTML attributes */
private $_html_attribs = array();
-
+
/* Constructor */
public function __construct($p = array()) {
@@ -122,13 +141,13 @@
unset($p['html_elements'], $p['html_attribs'], $p['ignore_elements'], $p['block_elements']);
$this->config = $p + array('show_washed'=>true, 'allow_remote'=>false, 'cid_map'=>array());
}
-
+
/* Register a callback function for a certain tag */
public function add_callback($tagName, $callback)
{
$this->handlers[$tagName] = $callback;
}
-
+
/* Check CSS style */
private function wash_style($style) {
$s = '';
@@ -142,7 +161,7 @@
preg_match('/^(url\(\s*[\'"]?([^\'"\)]*)[\'"]?\s*\)'./*1,2*/
'|rgb\(\s*[0-9]+\s*,\s*[0-9]+\s*,\s*[0-9]+\s*\)'.
'|-?[0-9.]+\s*(em|ex|px|cm|mm|in|pt|pc|deg|rad|grad|ms|s|hz|khz|%)?'.
- '|#[0-9a-f]{3,6}|[a-z0-9\-]+'.
+ '|#[0-9a-f]{3,6}|[a-z0-9", -]+'.
')\s*/i', $str, $match)) {
if ($match[2]) {
if (($src = $this->config['cid_map'][$match[2]])
@@ -155,11 +174,13 @@
else
$this->extlinks = true;
}
- else if (preg_match('/^data:.+/i', $url)) { // RFC2397
- $value .= ' url('.htmlspecialchars($url, ENT_QUOTES).')';
+ else if (preg_match('/^data:.+/i', $match[2])) { // RFC2397
+ $value .= ' url('.htmlspecialchars($match[2], ENT_QUOTES).')';
}
- } else if ($match[0] != 'url' && $match[0] != 'rbg') //whitelist ?
+ }
+ else if ($match[0] != 'url' && $match[0] != 'rgb') //whitelist ?
$value .= ' ' . $match[0];
+
$str = substr($str, strlen($match[0]));
}
if ($value)
@@ -180,8 +201,10 @@
if (isset($this->_html_attribs[$key]) ||
($key == 'href' && preg_match('/^(http:|https:|ftp:|mailto:|#).+/i', $value)))
$t .= ' ' . $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
- else if ($key == 'style' && ($style = $this->wash_style($value)))
- $t .= ' style="' . $style . '"';
+ else if ($key == 'style' && ($style = $this->wash_style($value))) {
+ $quot = strpos($style, '"') !== false ? "'" : '"';
+ $t .= ' style=' . $quot . $style . $quot;
+ }
else if ($key == 'background' || ($key == 'src' && strtolower($node->tagName) == 'img')) { //check tagName anyway
if (($src = $this->config['cid_map'][$value])
|| ($src = $this->config['cid_map'][$this->config['base_url'].$value])) {
@@ -225,7 +248,9 @@
else if (isset($this->_html_elements[$tagName])) {
$content = $this->dumpHtml($node);
$dump .= '<' . $tagName . $this->wash_attribs($node) .
- ($content != '' || isset($this->_block_elements[$tagName]) ? ">$content</$tagName>" : ' />');
+ // create closing tag for block elements, but also for elements
+ // with content or with some attributes (eg. style, class) (#1486812)
+ ($content != '' || $node->hasAttributes() || isset($this->_block_elements[$tagName]) ? ">$content</$tagName>" : ' />');
}
else if (isset($this->_ignore_elements[$tagName])) {
$dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' not allowed -->';
@@ -268,6 +293,10 @@
else
$this->config['base_url'] = '';
+ // Remove invalid HTML comments (#1487759)
+ // Don't remove valid conditional comments
+ $html = preg_replace('/<!--[^->[\n]*>/', '', $html);
+
@$node->loadHTML($html);
return $this->dumpHtml($node);
}
--
Gitblit v1.9.1