Merge remote-tracking branch 'my-fork/master' into indexer_improvements

inc/fulltext.php

@@ -130,7 +130,6 @@ function ft_backlinks($id){
     $result = array();
 
     $result = idx_get_indexer()->lookupKey('relation_references', $id);
-    $result = $result[$id];
 
     if(!count($result)) return $result;
 
@@ -234,8 +233,12 @@ function _ft_pageLookup(&$data){
                     $pages[$p_id] = p_get_first_heading($p_id, false);
             }
         }
-        //if ($in_title)
-        //    $titles = $Indexer->lookupKey('title', "*$id*");
+        if ($in_title) {
+            foreach ($Indexer->lookupKey('title', "*$id*") as $p_id) {
+                if (!isset($pages[$p_id]))
+                    $pages[$p_id] = p_get_first_heading($p_id, false);
+            }
+        }
     }
     if (isset($ns)) {
         foreach ($page_idx as $p_id) {

inc/indexer.php

@@ -320,6 +320,46 @@ class Doku_Indexer {
      * @author Tom N Harris <tnharris@whoopdedo.org>
      */
     public function deletePage($page) {
+        if (!$this->_lock())
+            return "locked";
+
+        // load known documents
+        $page_idx = $this->_getIndexKey('page', '', $page);
+        if ($page_idx === false) {
+            $this->_unlock();
+            return false;
+        }
+
+        // Remove obsolete index entries
+        $pageword_idx = $this->_getIndexKey('pageword', '', $pid);
+        if ($pageword_idx !== '') {
+            $delwords = explode(':',$pageword_idx);
+            $upwords = array();
+            foreach ($delwords as $word) {
+                if ($word != '') {
+                    list($wlen,$wid) = explode('*', $word);
+                    $wid = (int)$wid;
+                    $upwords[$wlen][] = $wid;
+                }
+            }
+            foreach ($upwords as $wlen => $widx) {
+                $index = $this->_getIndex('i', $wlen);
+                foreach ($widx as $wid) {
+                    $index[$wid] = $this->_updateTuple($index[$wid], $pid, 0);
+                }
+                $this->_saveIndex('i', $wlen, $index);
+            }
+        }
+        // Save the reverse index
+        if (!$this->_saveIndexKey('pageword', '', $pid, "")) {
+            $this->_unlock();
+            return false;
+        }
+
+        // XXX TODO: delete meta keys
+
+        $this->_unlock();
+        return true;
     }
 
     /**
@@ -423,7 +463,7 @@ class Doku_Indexer {
      * @param string    $key    name of the metadata key to look for
      * @param string    $value  search term to look for
      * @param callback  $func   comparison function
-     * @return array            lists with page names, keys are query values
+     * @return array            lists with page names, keys are query values if $key is array
      * @author Tom N Harris <tnharris@whoopdedo.org>
      * @author Michael Hamann <michael@content-space.de>
      */
@@ -436,15 +476,38 @@ class Doku_Indexer {
         // the matching ids for the provided value(s)
         $value_ids = array();
 
-        if (!is_array($value)) $value = array($value);
+        if (!is_array($value))
+            $value_array = array($value);
+        else
+            $value_array =& $value;
 
-        foreach ($value as $val) {
-            if (is_null($func)) {
-                if (($i = array_search($val, $words)) !== false)
-                    $value_ids[$i] = $val;
-            } else {
+        if (!is_null($func)) {
+            foreach ($value_array as $val) {
                 foreach ($words as $i => $word) {
-                    if (call_user_func_array($func, array($word, $value)))
+                    if (call_user_func_array($func, array($word, $val)))
+                        $value_ids[$i] = $val;
+                }
+            }
+        } else {
+            foreach ($value_array as $val) {
+                $xval = $val;
+                $caret = false;
+                $dollar = false;
+                // check for wildcards
+                if (substr($xval, 0, 1) == '*') {
+                    $xval = substr($xval, 1);
+                    $caret = '^';
+                }
+                if (substr($xval, -1, 1) == '*') {
+                    $xval = substr($xval, 0, -1);
+                    $dollar = '$';
+                }
+                if ($caret || $dollar) {
+                    $re = $caret.preg_quote($xval, '/').$dollar;
+                    foreach(array_keys(preg_grep('/'.$re.'/', $words)) as $i)
+                        $value_ids[$i] = $val;
+                } else {
+                    if (($i = array_search($val, $words)) !== false)
                         $value_ids[$i] = $val;
                 }
             }
@@ -462,6 +525,7 @@ class Doku_Indexer {
             // is an array with page_id => 1, page2_id => 1 etc. so take the keys only
             $result[$val] = array_keys($this->_parseTuples($page_idx, $lines[$value_id]));
         }
+        if (!is_array($value)) $result = $result[$value];
         return $result;
     }
 
@@ -492,12 +556,12 @@ class Doku_Indexer {
             // check for wildcards
             if (substr($xword, 0, 1) == '*') {
                 $xword = substr($xword, 1);
-                $caret = true;
+                $caret = '^';
                 $wlen -= 1;
             }
             if (substr($xword, -1, 1) == '*') {
                 $xword = substr($xword, 0, -1);
-                $dollar = true;
+                $dollar = '$';
                 $wlen -= 1;
             }
             if ($wlen < IDX_MINWORDLENGTH && !$caret && !$dollar && !is_numeric($xword))
@@ -505,9 +569,7 @@ class Doku_Indexer {
             if (!isset($tokens[$xword]))
                 $tokenlength[$wlen][] = $xword;
             if ($caret || $dollar) {
-                $re = preg_quote($xword, '/');
-                if ($caret) $re = '^'.$re;
-                if ($dollar) $re = $re.'$';
+                $re = $caret.preg_quote($xword, '/').$dollar;
                 $tokens[$xword][] = array($word, '/'.$re.'/');
                 if (!isset($tokenwild[$xword]))
                     $tokenwild[$xword] = $wlen;
@@ -926,6 +988,36 @@ function idx_addPage($page, $verbose=false) {
         }
     }
 
+    if (!page_exists($page)) {
+        if (!@file_exists($idxtag)) {
+            if ($verbose) print("Indexer: $page does not exist, ignoring".DOKU_LF);
+            return false;
+        }
+        $Indexer = idx_get_indexer();
+        $result = $Indexer->deletePage($page);
+        if ($result === "locked") {
+            if ($verbose) print("Indexer: locked".DOKU_LF);
+            return false;
+        }
+        @unlink($idxtag);
+        return $result;
+    }
+    $indexenabled = p_get_metadata($page, 'internal index', false);
+    if ($indexenabled === false) {
+        $result = false;
+        if (@file_exists($idxtag)) {
+            $Indexer = idx_get_indexer();
+            $result = $Indexer->deletePage($page);
+            if ($result === "locked") {
+                if ($verbose) print("Indexer: locked".DOKU_LF);
+                return false;
+            }
+            @unlink($idxtag);
+        }
+        if ($verbose) print("Indexer: index disabled for $page".DOKU_LF);
+        return $result;
+    }
+
     $body = '';
     $data = array($page, $body);
     $evt = new Doku_Event('INDEXER_PAGE_ADD', $data);
@@ -944,7 +1036,8 @@ function idx_addPage($page, $verbose=false) {
     if ($result) {
         $data = array('page' => $page, 'metadata' => array());
 
-        if (($references = p_get_metadata($page, 'relation references')) !== null)
+        $data['metadata']['title'] = p_get_metadata($page, 'title', false);
+        if (($references = p_get_metadata($page, 'relation references', false)) !== null)
             $data['metadata']['relation_references'] = array_keys($references);
 
         $evt = new Doku_Event('INDEXER_METADATA_INDEX', $data);