Remove obsolete words from search index

Creates another index file 'pagewords.idx' for the words in each page.
Words that are deleted from a page can then be removed from the word index.
The indexer version is incremented to force rebuilding of the index.
Also, a minor flaw in the regexp for asian words is fixed.

darcs-hash:20070919194244-6942e-2e08157dcf4fdf166b35b36a0faf8a3dfb415ad9.gz

1 files changed, 9 insertions, 1 deletions

diff --git a/bin/indexer.php b/bin/indexer.php
index d5f4c6446..c62f97882 100755
--- a/bin/indexer.php
+++ b/bin/indexer.php
@@ -13,7 +13,7 @@ session_write_close();
 
 // Version tag used to force rebuild on upgrade
 // Need to keep in sync with lib/exe/indexer.php
-if(!defined('INDEXER_VERSION')) define('INDEXER_VERSION', 1);
+if(!defined('INDEXER_VERSION')) define('INDEXER_VERSION', 2);
 
 // handle options
 $short_opts = 'hcuq';
@@ -68,6 +68,14 @@ function _usage() {
 
 function _update(){
     global $conf;
+
+    // upgrade to version 2
+    if (!@file_exists($conf['indexdir'].'/pageword.idx')){
+        _lock();
+        idx_upgradePageWords();
+        _unlock();
+    }
+
     $data = array();
     _quietecho("Searching pages... ");
     search($data,$conf['datadir'],'search_allpages',array());