Merge pull request #491 from ArthurHoaro/markdown-escape2

Markdown: don't escape content + sanitize sensible tags

application/Utils.php

@@ -62,13 +62,29 @@ function endsWith($haystack, $needle, $case=true)
 }
 
 /**
- * htmlspecialchars wrapper
+ * Htmlspecialchars wrapper
+ *
+ * @param string $str the string to escape.
+ *
+ * @return string escaped.
  */
 function escape($str)
 {
     return htmlspecialchars($str, ENT_COMPAT, 'UTF-8', false);
 }
 
+/**
+ * Reverse the escape function.
+ *
+ * @param string $str the string to unescape.
+ *
+ * @return string unescaped string.
+ */
+function unescape($str)
+{
+    return htmlspecialchars_decode($str);
+}
+
 /**
  * Link sanitization before templating
  */

plugins/markdown/markdown.php

@@ -117,23 +117,43 @@ function reverse_space2nbsp($description)
 }
 
 /**
- * Remove '>' at start of line auto generated by Shaarli core system
+ * Remove dangerous HTML tags (tags, iframe, etc.).
- * to allow markdown blockquotes.
+ * Doesn't affect <code> content (already escaped by Parsedown).
  *
  * @param string $description input description text.
  *
- * @return string $description without HTML links.
+ * @return string given string escaped.
  */
-function reset_quote_tags($description)
+function sanitize_html($description)
 {
-    return preg_replace('/^( *)&gt; /m', '$1> ', $description);
+    $escapeTags = array(
+        'script',
+        'style',
+        'link',
+        'iframe',
+        'frameset',
+        'frame',
+    );
+    foreach ($escapeTags as $tag) {
+        $description = preg_replace_callback(
+            '#<\s*'. $tag .'[^>]*>(.*</\s*'. $tag .'[^>]*>)?#is',
+            function ($match) { return escape($match[0]); },
+            $description);
+    }
+    $description = preg_replace(
+        '#(<[^>]+)on[a-z]*="[^"]*"#is',
+        '$1',
+        $description);
+    return $description;
 }
 
 /**
  * Render shaare contents through Markdown parser.
  *   1. Remove HTML generated by Shaarli core.
- *   2. Generate markdown descriptions.
+ *   2. Reverse the escape function.
- *   3. Wrap description in 'markdown' CSS class.
+ *   3. Generate markdown descriptions.
+ *   4. Sanitize sensible HTML tags for security.
+ *   5. Wrap description in 'markdown' CSS class.
  *
  * @param string $description input description text.
  *
@@ -147,11 +167,12 @@ function process_markdown($description)
     $processedDescription = reverse_text2clickable($processedDescription);
     $processedDescription = reverse_nl2br($processedDescription);
     $processedDescription = reverse_space2nbsp($processedDescription);
-    $processedDescription = reset_quote_tags($processedDescription);
+    $processedDescription = unescape($processedDescription);
     $processedDescription = $parsedown
         ->setMarkupEscaped(false)
         ->setBreaksEnabled(true)
         ->text($processedDescription);
+    $processedDescription = sanitize_html($processedDescription);
     $processedDescription = '<div class="markdown">'. $processedDescription . '</div>';
 
     return $processedDescription;

tests/plugins/PluginMarkdownTest.php

@@ -100,13 +100,18 @@ function testReverseSpace2nbsp()
     }
 
     /**
-     * Test reset_quote_tags()
+     * Test sanitize_html().
      */
-    function testResetQuoteTags()
+    function testSanitizeHtml() {
-    {
+        $input = '< script src="js.js"/>';
-        $text = '> quote1'. PHP_EOL . ' > quote2 ' . PHP_EOL . 'noquote';
+        $input .= '< script attr>alert(\'xss\');</script>';
-        $processedText = escape($text);
+        $input .= '<style> * { display: none }</style>';
-        $reversedText = reset_quote_tags($processedText);
+        $output = escape($input);
-        $this->assertEquals($text, $reversedText);
+        $input .= '<a href="#" onmouseHover="alert(\'xss\');" attr="tt">link</a>';
+        $output .= '<a href="#"  attr="tt">link</a>';
+        $this->assertEquals($output, sanitize_html($input));
+        // Do not touch escaped HTML.
+        $input = escape($input);
+        $this->assertEquals($input, sanitize_html($input));
     }
 }