Limiting use of readdir in the idx_indexLengths function (v2).

Each searches on the wiki use this function. Scanning the index directory eachtime is time consuming with a constant series of disk access.
Switching a normal search to use file_exists 1 or more times, and not readdir all the directory.
Switching a wildcard search to use a lengths.idx file containing all the word lengths used in the wiki, file generated if a new configuration parameter $conf[readdircache] is not 0 and fixed to a time in second. Creation of a new function idx_listIndexLengths to do this part.

5 files changed, 189 insertions, 20 deletions

diff --git a/_test/cases/inc/indexer_idx_indexlengths.test.php b/_test/cases/inc/indexer_idx_indexlengths.test.php
new file mode 100644
index 000000000..d1339a111
--- /dev/null
+++ b/_test/cases/inc/indexer_idx_indexlengths.test.php
@@ -0,0 +1,121 @@
+<?php
+
+require_once DOKU_INC.'inc/indexer.php';
+
+class indexer_idx_indexlengths_test extends UnitTestCase {
+
+    /**
+     * Test the function with an array of one value
+     */
+    function test_oneWord(){
+        global $conf;
+        $filter[8] = array('dokuwiki');
+        // one word should return the index
+        $ref[] = 8;
+        sort($ref);
+        $result = idx_indexLengths(&$filter);
+        sort($result);
+        $this->assertIdentical($result, $ref);
+    }
+
+    /**
+     * Test the function with an array of values
+     */
+    function test_moreWords() {
+        global $conf;
+        $filter = array( 4 => array('test'), 8 => array('dokuwiki'), 7 => array('powered'));
+        // more words should return the indexes
+        $ref = array(4, 7, 8);
+        sort($ref);
+        $result = idx_indexLengths(&$filter);
+        sort($result);
+        $this->assertIdentical($result, $ref);
+    }
+
+    /**
+     * Test a minimal value in case of wildcard search
+     */
+    function test_minValue() {
+        global $conf;
+        $filter = 5;
+        // construction of the list of the index to compare
+        $dir = @opendir($conf['indexdir']);
+        $ref = array();
+        while (($f = readdir($dir)) !== false) {
+            if (substr($f,0,1) == 'i' && substr($f,-4) == '.idx'){
+                $i = substr($f,1,-4);
+                if (is_numeric($i) && $i >= $filter)
+                $ref[] = (int)$i;
+            }
+        }
+        closedir($dir);
+        sort($ref);
+        $result = idx_indexLengths(&$filter);
+        sort($result);
+        $this->assertIdentical($result, $ref);
+    }
+}
+
+class indexer_idx_indexlengths_time extends UnitTestCase {
+
+    /**
+     * Test the time improvments of the new function
+     * Time reference for 10000 call oneWords: 4,6s
+     * It's 90% faster
+     */
+    function test_oneWord(){
+        global $conf;
+        $filter[8] = array('dokuwiki');
+        $start = microtime(true);
+        for ($i = 0; $i < 10000; $i++) {
+            $result = idx_indexLengths(&$filter);
+        }
+        $end = microtime(true);
+        $time = $end - $start;
+        $timeref = 4.6*0.10; // actual execution time of 4,6s for 10000 calls
+        echo "1) 10% ref : $timeref -> $time \n";
+        $this->assertTrue($time < $timeref);
+    }
+
+    /**
+     * Test the time improvments of the new function
+     * Time reference for 10000 call moreWords: 4,6s
+     * It's 90% faster
+     */
+    function test_moreWords() {
+        global $conf;
+        $filter = array( 4 => array('test'), 8 => array('dokuwiki'), 7 => array('powered'));
+        // more words should return the indexes
+        $start = microtime(true);
+        for ($i = 0; $i < 10000; $i++) {
+            $result = idx_indexLengths(&$filter);
+        }
+        $end = microtime(true);
+        $time = $end - $start;
+        $timeref = 4.6*0.10; // actual execution time of 4,6s for 10000 calls
+        echo "2) 10% ref : $timeref -> $time \n";
+        $this->assertTrue($time < $timeref);
+    }
+
+    /**
+     * Test the time improvments of the new function
+     * Time reference for 10000 call on minValue: 4,9s
+     * Sould be at least 65% faster
+     * Test fail with no cache
+     */
+    function test_minValue() {
+        global $conf;
+        $filter = 5;
+        $start = microtime(true);
+        for ($i = 0; $i < 10000; $i++) {
+            $result = idx_indexLengths(&$filter);
+        }
+        $end = microtime(true);
+        $time = $end - $start;
+        $timeref = 4.9 * 0.35; // actual execution time of 4,9s for 10000 calls
+        echo "3) 35% ref : $timeref -> $time \n";
+        $this->assertTrue($time < $timeref);
+    }
+}
+
+//Setup VIM: ex: et ts=4 enc=utf-8 :
diff --git a/conf/dokuwiki.php b/conf/dokuwiki.php
index 19397861c..d3823eb94 100644
--- a/conf/dokuwiki.php
+++ b/conf/dokuwiki.php
@@ -156,3 +156,4 @@ $conf['ftp']['user'] = 'user';
 $conf['ftp']['pass'] = 'password';
 $conf['ftp']['root'] = '/home/user/htdocs';
 
+$conf['readdircache'] = 0;               //time cache in second for the readdir opération, 0 to deactivate.
diff --git a/inc/indexer.php b/inc/indexer.php
index 07b67c014..54277a88c 100644
--- a/inc/indexer.php
+++ b/inc/indexer.php
@@ -411,40 +411,85 @@ function idx_updateIndexLine($line,$pid,$count){
 }
 
 /**
+ * Get the list of lenghts indexed in the wiki
+ *
+ * Read the index directory or a cache file and returns
+ * a sorted array of lengths of the words used in the wiki.
+ *
+ * @author YoBoY <yoboy.leguesh@gmail.com>
+ */
+function idx_listIndexLengths() {
+    global $conf;
+    // testing what we have to do, create a cache file or not.
+    if ($conf['readdircache'] == 0) {
+        $docache = false;
+    } else {
+        clearstatcache();
+        if (@file_exists($conf['indexdir'].'/lengths.idx') and (time() < @filemtime($conf['indexdir'].'/lengths.idx') + $conf['readdircache'])) {
+            if (($lengths = @file($conf['indexdir'].'/lengths.idx', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES) ) !== false) {
+                $idx = array();
+                foreach ( $lengths as $length) {
+                    $idx[] = (int)$length;
+                }
+                return $idx;
+            }
+        }
+        $docache = true;
+    }
+
+    if ($conf['readdircache'] == 0 or $docache ) {
+        $dir = @opendir($conf['indexdir']);
+        if($dir===false)
+            return array();
+        $idx[] = array();
+        while (($f = readdir($dir)) !== false) {
+            if (substr($f,0,1) == 'i' && substr($f,-4) == '.idx'){
+                $i = substr($f,1,-4);
+                if (is_numeric($i))
+                    $idx[] = (int)$i;
+            }
+        }
+        closedir($dir);
+        sort($idx);
+        // we save this in a file.
+        if ($docache === true) {
+            $handle = @fopen($conf['indexdir'].'/lengths.idx','w');
+            @fwrite($handle, implode("\n",$idx));
+            @fclose($handle);
+        }
+        return $idx;
+    }
+
+    return array();
+}
+
+/**
  * Get the word lengths that have been indexed.
  *
  * Reads the index directory and returns an array of lengths
  * that there are indices for.
  *
- * @author Tom N Harris <tnharris@whoopdedo.org>
+ * @author YoBoY <yoboy.leguesh@gmail.com>
  */
 function idx_indexLengths(&$filter){
     global $conf;
-    $dir = @opendir($conf['indexdir']);
-    if($dir===false)
-        return array();
     $idx = array();
-    if(is_array($filter)){
-        while (($f = readdir($dir)) !== false) {
-            if (substr($f,0,1) == 'i' && substr($f,-4) == '.idx'){
-                $i = substr($f,1,-4);
-                if (is_numeric($i) && isset($filter[(int)$i]))
-                    $idx[] = (int)$i;
+    if (is_array($filter)){
+        // testing if index files exists only
+        foreach ($filter as $key => $value) {
+            if (@file_exists($conf['indexdir']."/i$key.idx")) {
+                $idx[] = $key;
             }
         }
-    }else{
-        // Exact match first.
-        if(@file_exists($conf['indexdir']."/i$filter.idx"))
-            $idx[] = $filter;
-        while (($f = readdir($dir)) !== false) {
-            if (substr($f,0,1) == 'i' && substr($f,-4) == '.idx'){
-                $i = substr($f,1,-4);
-                if (is_numeric($i) && $i > $filter)
-                    $idx[] = (int)$i;
+    } else {
+        $lengths = idx_listIndexLengths();
+        foreach ( $lengths as $key => $length) {
+            // we keep all the values equal or superior 
+            if ((int)$length >= (int)$filter) {
+                $idx[] = $length;
             }
         }
     }
-    closedir($dir);
     return $idx;
 }
 
diff --git a/lib/plugins/config/lang/en/lang.php b/lib/plugins/config/lang/en/lang.php
index 7198ca36d..2bcd17c12 100644
--- a/lib/plugins/config/lang/en/lang.php
+++ b/lib/plugins/config/lang/en/lang.php
@@ -234,3 +234,4 @@ $lang['useheading_o_navigation'] = 'Navigation Only';
 $lang['useheading_o_content'] = 'Wiki Content Only';
 $lang['useheading_o_1'] = 'Always';
 
+$lang['readdircache'] = 'Maximum age for readdir cache (sec)';
diff --git a/lib/plugins/config/settings/config.metadata.php b/lib/plugins/config/settings/config.metadata.php
index 2f91b0ad8..cea191f56 100644
--- a/lib/plugins/config/settings/config.metadata.php
+++ b/lib/plugins/config/settings/config.metadata.php
@@ -188,6 +188,7 @@ $meta['rss_show_summary'] = array('onoff');
 $meta['broken_iua']  = array('onoff');
 $meta['xsendfile']   = array('multichoice','_choices' => array(0,1,2,3));
 $meta['renderer_xhtml'] = array('renderer','_format' => 'xhtml','_choices' => array('xhtml'));
+$meta['readdircache'] = array('numeric');
 
 $meta['_network']    = array('fieldset');
 $meta['proxy____host'] = array('string','_pattern' => '#^(|[a-z0-9\-\.+]+)$#i');