diff options
Diffstat (limited to 'inc/indexer.php')
-rw-r--r-- | inc/indexer.php | 40 |
1 files changed, 17 insertions, 23 deletions
diff --git a/inc/indexer.php b/inc/indexer.php index 32fbf4a1a..d4432026e 100644 --- a/inc/indexer.php +++ b/inc/indexer.php @@ -86,9 +86,7 @@ function idx_saveIndex($pre, $wlen, &$idx){ $fn = $conf['indexdir'].'/'.$pre.$wlen; $fh = @fopen($fn.'.tmp','w'); if(!$fh) return false; - foreach ($idx as $line) { - fwrite($fh,$line); - } + fwrite($fh,join('', $idx)); fclose($fh); if(isset($conf['fperm'])) chmod($fn.'.tmp', $conf['fperm']); io_rename($fn.'.tmp', $fn.'.idx'); @@ -154,7 +152,7 @@ function idx_saveIndexLine($pre, $wlen, $idx, $line){ $ih = @fopen($fn.'.idx','r'); if ($ih) { $ln = -1; - while (($curline = _freadline($ih)) !== false) { + while (($curline = fgets($ih)) !== false) { if (++$ln == $idx) { fwrite($fh, $line); } else { @@ -186,7 +184,7 @@ function idx_getIndexLine($pre, $wlen, $idx){ $fh = @fopen($fn,'r'); if(!$fh) return ''; $ln = -1; - while (($line = _freadline($fh)) !== false) { + while (($line = fgets($fh)) !== false) { if (++$ln == $idx) break; } fclose($fh); @@ -252,6 +250,8 @@ function idx_getPageWords($page){ // arrive here with $words = array(wordlen => array(word => frequency)) + $word_idx_modified = false; + $index = array(); //resulting index foreach (array_keys($words) as $wlen){ $word_idx = idx_getIndex('w',$wlen); @@ -260,6 +260,7 @@ function idx_getPageWords($page){ if(!is_int($wid)){ $wid = count($word_idx); $word_idx[] = "$word\n"; + $word_idx_modified = true; } if(!isset($index[$wlen])) $index[$wlen] = array(); @@ -267,7 +268,7 @@ function idx_getPageWords($page){ } // save back word index - if(!idx_saveIndex('w',$wlen,$word_idx)){ + if($word_idx_modified && !idx_saveIndex('w',$wlen,$word_idx)){ trigger_error("Failed to write word index", E_USER_ERROR); return false; } @@ -388,26 +389,19 @@ function idx_writeIndexLine($fh,$line,$pid,$count){ * @author Andreas Gohr <andi@splitbrain.org> */ function idx_updateIndexLine($line,$pid,$count){ - $line = trim($line); - $updated = array(); - if($line != ''){ - $parts = explode(':',$line); - // remove doc from given line - foreach($parts as $part){ - if($part == '') continue; - list($doc,$cnt) = explode('*',$part); - if($doc != $pid){ - $updated[] = $part; - } - } + if ($line == ''){ + $newLine = "\n"; + }else{ + $newLine = preg_replace('/(^|:)'.preg_quote($pid, '/').'\*\d*/', '', $line); } - - // add doc if ($count){ - $updated[] = "$pid*$count"; + if (strlen($newLine) > 1){ + return "$pid*$count:".$newLine; + }else{ + return "$pid*$count".$newLine; + } } - - return join(':',$updated)."\n"; + return $newLine; } /** |