From 8e528fd7bf2e2bd1e20c7e7de5eefb26bf6b19d9 Mon Sep 17 00:00:00 2001
From: Aleksander Machniak <alec@alec.pl>
Date: Wed, 23 May 2012 14:09:57 -0400
Subject: [PATCH] hide_blockquote - a new plugin for hiding citation blocks

---
 program/lib/washtml.php |   51 +++++++++++++++++++++++++++++++--------------------
 1 files changed, 31 insertions(+), 20 deletions(-)

diff --git a/program/lib/washtml.php b/program/lib/washtml.php
index 1e6d599..4221abd 100644
--- a/program/lib/washtml.php
+++ b/program/lib/washtml.php
@@ -101,14 +101,14 @@
     'cellpadding', 'valign', 'bgcolor', 'color', 'border', 'bordercolorlight',
     'bordercolordark', 'face', 'marginwidth', 'marginheight', 'axis', 'border',
     'abbr', 'char', 'charoff', 'clear', 'compact', 'coords', 'vspace', 'hspace',
-    'cellborder', 'size', 'lang', 'dir',
+    'cellborder', 'size', 'lang', 'dir', 'usemap',
     // 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', 'center',
-    'table', 'ul', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'dl', 'strong', 'i', 'b');
+    'table', 'ul', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'dl', 'strong', 'i', 'b', 'u', 'span');
 
   /* State for linked objects in HTML */
   public $extlinks = false;
@@ -133,7 +133,8 @@
 
 
   /* Constructor */
-  public function __construct($p = array()) {
+  public function __construct($p = array())
+  {
     $this->_html_elements = array_flip((array)$p['html_elements']) + array_flip(self::$html_elements) ;
     $this->_html_attribs = array_flip((array)$p['html_attribs']) + array_flip(self::$html_attribs);
     $this->_ignore_elements = array_flip((array)$p['ignore_elements']) + array_flip(self::$ignore_elements);
@@ -141,15 +142,16 @@
     unset($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) {
+  private function wash_style($style)
+  {
     $s = '';
 
     foreach (explode(';', $style) as $declaration) {
@@ -161,14 +163,14 @@
           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]])
                 || ($src = $this->config['cid_map'][$this->config['base_url'].$match[2]])) {
               $value .= ' url('.htmlspecialchars($src, ENT_QUOTES) . ')';
             }
-            else if (preg_match('/^(http|https|ftp):.*$/i', $match[2], $url)) {
+            else if (preg_match('!^(https?:)?//[a-z0-9/._+-]+$!i', $match[2], $url)) {
               if ($this->config['allow_remote'])
                 $value .= ' url('.htmlspecialchars($url[0], ENT_QUOTES).')';
               else
@@ -178,8 +180,9 @@
               $value .= ' url('.htmlspecialchars($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)
@@ -190,7 +193,8 @@
   }
 
   /* Take a node and return allowed attributes and check values */
-  private function wash_attribs($node) {
+  private function wash_attribs($node)
+  {
     $t = '';
     $washed;
 
@@ -198,10 +202,12 @@
       $key = strtolower($key);
       $value = $node->getAttribute($key);
       if (isset($this->_html_attribs[$key]) ||
-         ($key == 'href' && preg_match('/^(http:|https:|ftp:|mailto:|#).+/i', $value)))
+         ($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'].$value])) {
@@ -228,7 +234,8 @@
   /* The main loop that recurse on a node tree.
    * It output only allowed tags with allowed attributes
    * and allowed inline styles */
-  private function dumpHtml($node) {
+  private function dumpHtml($node)
+  {
     if(!$node->hasChildNodes())
       return '';
 
@@ -240,14 +247,12 @@
       case XML_ELEMENT_NODE: //Check element
         $tagName = strtolower($node->tagName);
         if ($callback = $this->handlers[$tagName]) {
-          $dump .= call_user_func($callback, $tagName, $this->wash_attribs($node), $this->dumpHtml($node));
+          $dump .= call_user_func($callback, $tagName, $this->wash_attribs($node), $this->dumpHtml($node), $this);
         }
         else if (isset($this->_html_elements[$tagName])) {
           $content = $this->dumpHtml($node);
           $dump .= '<' . $tagName . $this->wash_attribs($node) .
-            // 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</$tagName>" : ' />');
+            ($content != '' || isset($this->_block_elements[$tagName]) ? ">$content</$tagName>" : ' />');
         }
         else if (isset($this->_ignore_elements[$tagName])) {
           $dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' not allowed -->';
@@ -298,6 +303,12 @@
     return $this->dumpHtml($node);
   }
 
-}
+  /**
+   * Getter for config parameters
+   */
+  public function get_config($prop)
+  {
+      return $this->config[$prop];
+  }
 
-?>
+}

--
Gitblit v1.9.1