Index: includes/path.inc
===================================================================
RCS file: /cvs/drupal/drupal/includes/path.inc,v
retrieving revision 1.37
diff -u -F^f -r1.37 path.inc
--- includes/path.inc	26 May 2009 09:12:28 -0000	1.37
+++ includes/path.inc	31 May 2009 01:52:23 -0000
@@ -66,8 +66,16 @@ function drupal_lookup_path($action, $pa
     $count = NULL;
     $system_paths = array();
     $no_aliases = array();
+
+    $whitelist = drupal_path_alias_whitelist_rebuild();
+    variable_set('path_alias_whitelist', $whitelist);
   }
   elseif ($count > 0 && $path != '') {
+    // Retrieve the path alias whitelist
+    $whitelist = variable_get('path_alias_whitelist', array());
+    // And derive the top level component of the path
+    $top_level = strtok($path, '/');
+
     if ($action == 'alias') {
       // During the first call to drupal_lookup_path() per language, load the
       // expected system paths for the page from cache.
@@ -93,6 +101,12 @@ function drupal_lookup_path($action, $pa
       if (isset($map[$path_language][$path])) {
         return $map[$path_language][$path];
       }
+      elseif (!isset($whitelist[$top_level])) {
+        // Check the path whitelist, if the top_level part before the first /
+        // is not in the list, then there is no need to do anything further,
+        // it is not in the database
+        return FALSE;
+      }
       // For system paths which were not cached, query aliases individually.
       else if (!isset($no_aliases[$path_language][$path])) {
         // Get the most fitting result falling back with alias without language
@@ -347,3 +361,38 @@ function drupal_match_path($path, $patte
 function current_path() {
   return $_GET['q'];
 }
+
+/**
+ * Rebuild the path alias white list.
+ *
+ * This code may be hard to understand at first, but it will become clearer in a minute.
+ *
+ * The idea is to use indexed lookups and progressively exclude the top part
+ * that has been already looked up.
+ *
+ * So the initial query entering the look is like this:
+ *   SELECT src FROM url_alias LIMIT 1
+ * Suppose this returns 'node/nnnn', then the next query executed within the look will be:
+ *   SELECT src FROM url_alias WHERE src <> 'node' AND src NOT LIKE 'node/%' LIMIT 1
+ * This excludes the 'node' part, and suppose it returns 'user', the next query becomes
+ *   SELECT src FROM url_alias WHERE src <> 'node' AND src NOT LIKE 'node/%'
+ *     AND src <> 'user' AND src NOT LIKE 'user/%' LIMIT 1
+ * And so on. We keep going until we process all the top level parts, and store them in
+ * the whitelist array.
+ *
+ * @return
+ *   An array containing a white list of path aliases.
+ */
+function drupal_path_alias_whitelist_rebuild() {
+  $whitelist = array();
+  $query = db_select('url_alias', 'u');
+  $query->addField('u', 'src');
+  $query->range(0, 1);
+  while ($result = $query->execute()->fetchField()) {
+    $part = strtok($result, '/');
+    $whitelist[$part] = TRUE;
+    $query->condition('src', "$part/%", 'NOT LIKE');
+    $query->condition('src', $part, '<>');
+  }
+  return $whitelist;
+}