soted indexer is now default

darcs-hash:20070226175529-7ad00-4d3d984da1edbf2ded546cfbd7374f97f032d032.gz

4 files changed, 6 insertions, 67 deletions

diff --git a/conf/dokuwiki.php b/conf/dokuwiki.php
index d442f0e93..a6ba7d6ce 100644
--- a/conf/dokuwiki.php
+++ b/conf/dokuwiki.php
@@ -134,5 +134,3 @@ $conf['ftp']['user'] = 'user';
 $conf['ftp']['pass'] = 'password';
 $conf['ftp']['root'] = '/home/user/htdocs';
 
-/* FIXME: delete when no longer needed */
-$conf['test_indexer'] = 0;
diff --git a/inc/indexer.php b/inc/indexer.php
index 09b43a1b5..5a88ac439 100644
--- a/inc/indexer.php
+++ b/inc/indexer.php
@@ -253,8 +253,8 @@ function idx_writeIndexLine($fh,$line,$pid,$count){
 /**
  * Modify an index line with new information
  *
- * This returns a line of the index. It removes the 
- * given document from the line and readds it if 
+ * This returns a line of the index. It removes the
+ * given document from the line and readds it if
  * $count is >0.
  *
  * @author Tom N Harris <tnharris@whoopdedo.org>
@@ -324,72 +324,19 @@ function idx_indexLengths(&$filter){
 /**
  * Find the the index number of each search term.
  *
- * There are two variation: Simple and Sorted.
- * The simple version just takes the words one at a time.
- * The sorted version will group together words that appear in the same index.
+ * This will group together words that appear in the same index.
  * So it should perform better, because it only opens each index once.
  * Actually, it's not that great. (in my experience) Probably because of the disk cache.
  * And the sorted function does more work, making it slightly slower in some cases.
  *
- * For now, you can choose to use the sorted version by setting $conf['test_indexer'] = 1
- * Eventually, the more worthy will be chosen and the loser cast into the deepest depths.
- *
  * @param array    $words   The query terms. Words should only contain valid characters,
  *                          with a '*' at either the beginning or end of the word (or both)
- * @param arrayref $result  Set to word => array("length*id" ...), use this to merge the 
+ * @param arrayref $result  Set to word => array("length*id" ...), use this to merge the
  *                          index locations with the appropriate query term.
  * @return array            Set to length => array(id ...)
  *
  * @author Tom N Harris <tnharris@whoopdedo.org>
  */
-function idx_getIndexWordsSimple($words, &$result){
-    // get word IDs
-    $wids = array();
-    foreach($words as $word){
-        $result[$word] = array();
-        $wild = 0;
-        $xword = $word;
-        $wlen = wordlen($word);
-
-        // check for wildcards
-        if(substr($xword,0,1) == '*'){
-            $xword = substr($xword,1);
-            $wild |= 1;
-            $wlen -= 1;
-        }
-        if(substr($xword,-1,1) == '*'){
-            $xword = substr($xword,0,-1);
-            $wild |= 2;
-            $wlen -= 1;
-        }
-        if ($wlen < 3 && $wild == 0 && !is_numeric($xword)) continue;
-
-        // look for the ID(s) for the given word
-        if($wild){  // handle wildcard search
-            $ptn = preg_quote($xword,'/');
-            if(($wild&1) == 0) $ptn = '^'.$ptn;
-            if(($wild&2) == 0) $ptn = $ptn.'$';
-            $ptn = '/'.$ptn.'/';
-            foreach (idx_indexLengths($wlen) as $ixlen){
-                $word_idx = idx_getIndex('w',$ixlen);
-                foreach(array_keys(preg_grep($ptn,$word_idx)) as $wid){
-                    $wids[$ixlen][] = $wid;
-                    $result[$word][] = "$ixlen*$wid";
-                }
-            }
-        }else{     // handle exact search
-            $word_idx = idx_getIndex('w',$wlen);
-            $wid = array_search("$word\n",$word_idx);
-            if(is_int($wid)){
-                $wids[$wlen][] = $wid;
-                $result[$word][] = "$wlen*$wid";
-            }else{
-                $result[$word] = array();
-            }
-        }
-    }
-    return $wids;
-}
 function idx_getIndexWordsSorted($words,&$result){
     // parse and sort tokens
     $tokens = array();
@@ -480,15 +427,12 @@ function idx_lookup($words){
 
     $result = array();
 
-    if(isset($conf['test_indexer']) && ($conf['test_indexer']&1))
-        $wids = idx_getIndexWordsSorted($words, $result);
-    else
-        $wids = idx_getIndexWordsSimple($words, $result);
+    $wids = idx_getIndexWordsSorted($words, $result);
     if(empty($wids)) return array();
 
     // load known words and documents
     $page_idx = idx_getIndex('page','');
-    
+
     $docs = array();                          // hold docs found
     foreach(array_keys($wids) as $wlen){
         $wids[$wlen] = array_unique($wids[$wlen]);
diff --git a/lib/plugins/config/lang/en/lang.php b/lib/plugins/config/lang/en/lang.php
index 825f5cf85..41f688d0a 100644
--- a/lib/plugins/config/lang/en/lang.php
+++ b/lib/plugins/config/lang/en/lang.php
@@ -184,5 +184,3 @@ $lang['compression_o_0']   = 'none';
 $lang['compression_o_gz']  = 'gzip';
 $lang['compression_o_bz2'] = 'bz2';
 
-/* FIXME: delete when no longer needed */
-$lang['test_indexer'] = 'New Indexer testing bitfield: 0x1 -> sorted searching';
diff --git a/lib/plugins/config/settings/config.metadata.php b/lib/plugins/config/settings/config.metadata.php
index ee8db882f..20a62f72d 100644
--- a/lib/plugins/config/settings/config.metadata.php
+++ b/lib/plugins/config/settings/config.metadata.php
@@ -167,7 +167,6 @@ $meta['rss_update']  = array('numeric');
 $meta['recent_days'] = array('numeric');
 $meta['rss_show_summary'] = array('onoff');
 $meta['broken_iua']  = array('onoff');
-$meta['test_indexer'] = array('numeric'); // FIXME: delete when no longer needed
 
 $meta['_network']    = array('fieldset');
 $meta['proxy____host'] = array('string','_pattern' => '#^[a-z0-9\-\.+]+?#i');