prevent access of wiki files

This prevents the crawler to descent into the wiki directory or data
directory, even when for some reason a higher up directory was made
accessible. This should prevent the circumvention of ACLs and prevent
access to sensitive data like user password hashes etc.

Crawler.php

@@ -57,6 +57,9 @@ public function crawl($root, $local, $pattern, $recursive, $titlefile)
     {
         $path = $root . $local;
 
+        // do not descent into wiki or data directories
+        if(Path::isWikiControlled($path)) return [];
+
         if (($dir = opendir($path)) === false) return [];
         $result = [];
         while (($file = readdir($dir)) !== false) {

Path.php

@@ -149,4 +149,26 @@ public static function realpath($path)
         }
         return implode('/', $output);
     }
+
+    /**
+     * Check if the given path is within the data or dokuwiki dir
+     *
+     * This whould prevent accidental or deliberate circumvention of the ACLs
+     *
+     * @param string $path and already cleaned path
+     * @return bool
+     */
+    public static function isWikiControlled($path)
+    {
+        global $conf;
+        $dataPath = self::cleanPath($conf['savedir']);
+        if (str_starts_with($path, $dataPath)) {
+            return true;
+        }
+        $wikiDir = self::cleanPath(DOKU_INC);
+        if (str_starts_with($path, $wikiDir)) {
+            return true;
+        }
+        return false;
+    }
 }

file.php

@@ -19,6 +19,10 @@
 
 try {
     $pathInfo = $pathUtil->getPathInfo($path, false);
+    if ($pathUtil::isWikiControlled($pathInfo['path'])) {
+        throw new Exception('Access to wiki files is not allowed');
+    }
+
     if (!is_readable($pathInfo['path'])) {
         header('Content-Type: text/plain');
         http_status(404);