Changeset b6f0405 in github

Timestamp:

May 27, 2010 9:34:58 AM (3 years ago)

Author:

alecpl <alec@…>

Branches:

master, HEAD, courier-fix, dev-browser-capabilities, pdo, release-0.6, release-0.7, release-0.8

Children:

e559232

Parents:

a3c9bd1

Message:

• support base URL for inline images

File:

• program/lib/washtml.php (modified) (5 diffs)

program/lib/washtml.php

@@ -75 +75 @@
  * - changed $ignore_elements behaviour
  * - added RFC2397 support
+ * - base URL support
  */
 
@@ -145 +146 @@
                  ')\s*/i', $str, $match)) {
           if ($match[2]) {
-            if ($src = $this->config['cid_map'][$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], $url)) {
               if ($this->config['allow_remote'])
@@ -181 +184 @@
         $t .= ' style="' . $style . '"';
       else if ($key == 'background' || ($key == 'src' && strtolower($node->tagName) == 'img')) { //check tagName anyway
-        if ($src = $this->config['cid_map'][$value]) {
+        if (($src = $this->config['cid_map'][$value])
+            || ($src = $this->config['cid_map'][$this->config['base_url'].$value])) {
           $t .= ' ' . $key . '="' . htmlspecialchars($src, ENT_QUOTES) . '"';
         }
@@ -216 +220 @@
       case XML_ELEMENT_NODE: //Check element
         $tagName = strtolower($node->tagName);
-        if($callback = $this->handlers[$tagName]) {
+        if ($callback = $this->handlers[$tagName]) {
           $dump .= call_user_func($callback, $tagName, $this->wash_attribs($node), $this->dumpHtml($node));
-        } else if(isset($this->_html_elements[$tagName])) {
+        }
+        else if (isset($this->_html_elements[$tagName])) {
           $content = $this->dumpHtml($node);
           $dump .= '<' . $tagName . $this->wash_attribs($node) .
             ($content != '' || isset($this->_block_elements[$tagName]) ? ">$content</$tagName>" : ' />');
-        } else if(isset($this->_ignore_elements[$tagName])) {
+        }
+        else if (isset($this->_ignore_elements[$tagName])) {
           $dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' not allowed -->';
-        } else {
+        }
+        else {
           $dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' ignored -->';
           $dump .= $this->dumpHtml($node); // ignore tags not its content
-        }
+        }
         break;
       case XML_CDATA_SECTION_NODE:
@@ -250 +257 @@
   /* Main function, give it untrusted HTML, tell it if you allow loading
    * remote images and give it a map to convert "cid:" urls. */
-  public function wash($html) {
-    //Charset seems to be ignored (probably if defined in the HTML document)
+  public function wash($html)
+  {
+    // Charset seems to be ignored (probably if defined in the HTML document)
     $node = new DOMDocument('1.0', $this->config['charset']);
     $this->extlinks = false;
+
+    // Find base URL for images
+    if (preg_match('/<base\s+href=[\'"]*([^\'"]+)/is', $html, $matches))
+      $this->config['base_url'] = $matches[1];
+    else
+      $this->config['base_url'] = '';
+
     @$node->loadHTML($html);
     return $this->dumpHtml($node);