security: replace vulnerable regex with parser (#1223)

* security: replace vulnerable regex with parser Problem: link regex was vulnerable Solution: dedicated parser Fixes: #1222
markedjs · Apr 17, 2018 · 5ab4ae3 · 5ab4ae3
1 parent fa15a37
commit 5ab4ae3
Showing 1 changed file with 62 additions and 3 deletions.
diff --git a/lib/marked.js b/lib/marked.js
@@ -554,9 +554,68 @@ inline.normal = merge({}, inline);
 inline.pedantic = merge({}, inline.normal, {
   strong: /^__(?=\S)([\s\S]*?\S)__(?!_)|^\*\*(?=\S)([\s\S]*?\S)\*\*(?!\*)/,
   em: /^_(?=\S)([\s\S]*?\S)_(?!_)|^\*(?=\S)([\s\S]*?\S)\*(?!\*)/,
-  link: edit(/^!?\[(label)\]\(\s*<?([\s\S]*?)>?(?:\s+(['"][\s\S]*?['"]))?\s*\)/)
-    .replace('label', inline._label)
-    .getRegex(),
+  /* Original link re: /^!?\[(label)\]\(\s*<?([\s\S]*?)>?(?:\s+(['"][\s\S]*?['"]))?\s*\)/
+   * This captures the spec reasonably well but is vulnerable to REDOS.
+   * Instead we use a custom parser that follows the RegExp.exec semantics. */
+  link: {
+    exec: function (s) {
+      // [TEXT](DESTINATION)
+      var generalLinkRe = edit(/^!?\[(label)\]\((.*?)\)/)
+        .replace('label', inline._label)
+        .getRegex();
+
+      // destination: DESTINATION from generalLinkRe
+      // returns [destination, title]: no angle-brackets on destination, no quotes on title
+      function splitIntoDestinationAndTitle (destination) {
+        function unwrapAngleBrackets (str) {
+          if (str.match(/^<.*>$/)) {
+            str = str.slice(1, -1);
+          }
+          return str;
+        }
+
+        // Valid DESTINATIONs, in decreasing specificity.
+        var destinationAndTitleRe = /^([^'"(]*[^\s])\s+(['"(].*['")])/;
+        var destinationRe = /^(<?[\s\S]*>?)/;
+        var parsingRegexes = [destinationAndTitleRe, destinationRe];
+
+        var match = false;
+        for (var i = 0; i < parsingRegexes.length; i++) {
+          match = parsingRegexes[i].exec(destination);
+          if (match) {
+            break;
+          }
+        }
+
+        if (!match) {
+          return null;
+        }
+
+        var dest = match[1];
+        var title = match[2] || ''; // Not all parsingRegexes have 2 groups.
+
+        // Format dest.
+        dest = dest.trim();
+        dest = unwrapAngleBrackets(dest);
+
+        return [dest, title];
+      }
+
+      var fullMatch = generalLinkRe.exec(s);
+      if (!fullMatch) {
+        return null;
+      }
+
+      var text = fullMatch[1];
+      var destination = fullMatch[2];
+
+      var destinationAndTitle = splitIntoDestinationAndTitle(destination);
+      if (!destinationAndTitle) {
+        return null;
+      }
+      return [fullMatch[0], text, destinationAndTitle[0], destinationAndTitle[1]];
+    }
+  },
   reflink: edit(/^!?\[(label)\]\s*\[([^\]]*)\]/)
     .replace('label', inline._label)
     .getRegex()