xref: /dokuwiki/bin/indexer.php (revision 981f048a597be073d01aefa348e6eecb2c52f100)
124c3e0d2SAndreas Gohr#!/usr/bin/php
224c3e0d2SAndreas Gohr<?php
378ba70a7Schrisif ('cli' != php_sapi_name()) die();
424c3e0d2SAndreas Gohr
524c3e0d2SAndreas Gohrif(!defined('DOKU_INC')) define('DOKU_INC',realpath(dirname(__FILE__).'/../').'/');
624c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/init.php');
724c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/common.php');
824c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/pageutils.php');
924c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/search.php');
1024c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/indexer.php');
1124c3e0d2SAndreas Gohrrequire_once(DOKU_INC.'inc/cliopts.php');
1224c3e0d2SAndreas Gohrsession_write_close();
1324c3e0d2SAndreas Gohr
1466b95544STom N Harris// Version tag used to force rebuild on upgrade
1566b95544STom N Harris// Need to keep in sync with lib/exe/indexer.php
1666b95544STom N Harrisif(!defined('INDEXER_VERSION')) define('INDEXER_VERSION', 1);
1766b95544STom N Harris
1824c3e0d2SAndreas Gohr// handle options
19*981f048aSGabriel Birke$short_opts = 'hcuq';
20*981f048aSGabriel Birke$long_opts  = array('help', 'clear', 'update', 'quiet');
2124c3e0d2SAndreas Gohr$OPTS = Doku_Cli_Opts::getOptions(__FILE__,$short_opts,$long_opts);
2224c3e0d2SAndreas Gohrif ( $OPTS->isError() ) {
2324c3e0d2SAndreas Gohr    fwrite( STDERR, $OPTS->getMessage() . "\n");
2424c3e0d2SAndreas Gohr    _usage();
2524c3e0d2SAndreas Gohr    exit(1);
2624c3e0d2SAndreas Gohr}
2724c3e0d2SAndreas Gohr$CLEAR = false;
28*981f048aSGabriel Birke$QUIET = false;
2924c3e0d2SAndreas Gohrforeach ($OPTS->options as $key => $val) {
3024c3e0d2SAndreas Gohr    switch ($key) {
3124c3e0d2SAndreas Gohr        case 'h':
3224c3e0d2SAndreas Gohr        case 'help':
3324c3e0d2SAndreas Gohr            _usage();
3424c3e0d2SAndreas Gohr            exit;
3524c3e0d2SAndreas Gohr        case 'c':
3624c3e0d2SAndreas Gohr        case 'clear':
3724c3e0d2SAndreas Gohr            $CLEAR = true;
3824c3e0d2SAndreas Gohr            break;
39*981f048aSGabriel Birke        case 'q':
40*981f048aSGabriel Birke        case 'quiet':
41*981f048aSGabriel Birke            $QUIET = true;
42*981f048aSGabriel Birke            break;
4324c3e0d2SAndreas Gohr    }
4424c3e0d2SAndreas Gohr}
4524c3e0d2SAndreas Gohr
4624c3e0d2SAndreas Gohr#------------------------------------------------------------------------------
4724c3e0d2SAndreas Gohr# Action
4824c3e0d2SAndreas Gohr
4924c3e0d2SAndreas Gohrif($CLEAR) _clearindex();
5024c3e0d2SAndreas Gohr_update();
5124c3e0d2SAndreas Gohr
5224c3e0d2SAndreas Gohr
5324c3e0d2SAndreas Gohr
5424c3e0d2SAndreas Gohr#------------------------------------------------------------------------------
5524c3e0d2SAndreas Gohr
5624c3e0d2SAndreas Gohrfunction _usage() {
5724c3e0d2SAndreas Gohr    print "Usage: indexer.php <options>
5824c3e0d2SAndreas Gohr
5924c3e0d2SAndreas Gohr    Updates the searchindex by indexing all new or changed pages
6024c3e0d2SAndreas Gohr    when the -c option is given the index is cleared first.
6124c3e0d2SAndreas Gohr
6224c3e0d2SAndreas Gohr    OPTIONS
6324c3e0d2SAndreas Gohr        -h, --help     show this help and exit
6424c3e0d2SAndreas Gohr        -c, --clear    clear the index before updating
65*981f048aSGabriel Birke        -q, --quiet    don't produce any output
6624c3e0d2SAndreas Gohr";
6724c3e0d2SAndreas Gohr}
6824c3e0d2SAndreas Gohr
6924c3e0d2SAndreas Gohrfunction _update(){
7024c3e0d2SAndreas Gohr    global $conf;
7124c3e0d2SAndreas Gohr    $data = array();
72*981f048aSGabriel Birke    _quietecho("Searching pages... ");
7324c3e0d2SAndreas Gohr    search($data,$conf['datadir'],'search_allpages',array());
74*981f048aSGabriel Birke    _quietecho(count($data)." pages found.\n");
7524c3e0d2SAndreas Gohr
7624c3e0d2SAndreas Gohr    foreach($data as $val){
7724c3e0d2SAndreas Gohr        _index($val['id']);
7824c3e0d2SAndreas Gohr    }
7924c3e0d2SAndreas Gohr}
8024c3e0d2SAndreas Gohr
8124c3e0d2SAndreas Gohrfunction _index($id){
8224c3e0d2SAndreas Gohr    global $CLEAR;
8324c3e0d2SAndreas Gohr
8424c3e0d2SAndreas Gohr    // if not cleared only update changed and new files
8524c3e0d2SAndreas Gohr    if(!$CLEAR){
8666b95544STom N Harris        $idxtag = metaFN($id,'.indexed');
8766b95544STom N Harris        if(@file_exists($idxtag)){
8866b95544STom N Harris            if(io_readFile($idxtag) >= INDEXER_VERSION){
8924c3e0d2SAndreas Gohr                $last = @filemtime(metaFN($id,'.indexed'));
9024c3e0d2SAndreas Gohr                if($last > @filemtime(wikiFN($id))) return;
9124c3e0d2SAndreas Gohr            }
9266b95544STom N Harris        }
9366b95544STom N Harris    }
9424c3e0d2SAndreas Gohr
9524c3e0d2SAndreas Gohr    _lock();
96*981f048aSGabriel Birke    _quietecho("$id... ");
9724c3e0d2SAndreas Gohr    idx_addPage($id);
9866b95544STom N Harris    io_saveFile(metaFN($id,'.indexed'),INDEXER_VERSION);
99*981f048aSGabriel Birke    _quietecho("done.\n");
10024c3e0d2SAndreas Gohr    _unlock();
10124c3e0d2SAndreas Gohr}
10224c3e0d2SAndreas Gohr
10324c3e0d2SAndreas Gohr/**
10424c3e0d2SAndreas Gohr * lock the indexer system
10524c3e0d2SAndreas Gohr */
10624c3e0d2SAndreas Gohrfunction _lock(){
10724c3e0d2SAndreas Gohr    global $conf;
10824c3e0d2SAndreas Gohr    $lock = $conf['lockdir'].'/_indexer.lock';
10924c3e0d2SAndreas Gohr    $said = false;
11044881d27STroels Liebe Bentsen    while(!@mkdir($lock, $conf['dmode'])){
11124c3e0d2SAndreas Gohr        if(time()-@filemtime($lock) > 60*5){
11224c3e0d2SAndreas Gohr            // looks like a stale lock - remove it
11324c3e0d2SAndreas Gohr            @rmdir($lock);
11424c3e0d2SAndreas Gohr        }else{
11524c3e0d2SAndreas Gohr            if($said){
116*981f048aSGabriel Birke                _quietecho(".");
11724c3e0d2SAndreas Gohr            }else{
118*981f048aSGabriel Birke                _quietecho("Waiting for lockfile (max. 5 min)");
11924c3e0d2SAndreas Gohr                $said = true;
12024c3e0d2SAndreas Gohr            }
12124c3e0d2SAndreas Gohr            sleep(15);
12224c3e0d2SAndreas Gohr        }
12324c3e0d2SAndreas Gohr    }
1241ca31cfeSAndreas Gohr    if($conf['dperm']) chmod($lock, $conf['dperm']);
125*981f048aSGabriel Birke    if($said) _quietecho("\n");
12624c3e0d2SAndreas Gohr}
12724c3e0d2SAndreas Gohr
12824c3e0d2SAndreas Gohr/**
12924c3e0d2SAndreas Gohr * unlock the indexer sytem
13024c3e0d2SAndreas Gohr */
13124c3e0d2SAndreas Gohrfunction _unlock(){
13224c3e0d2SAndreas Gohr    global $conf;
13324c3e0d2SAndreas Gohr    $lock = $conf['lockdir'].'/_indexer.lock';
13424c3e0d2SAndreas Gohr    @rmdir($lock);
13524c3e0d2SAndreas Gohr}
13624c3e0d2SAndreas Gohr
13724c3e0d2SAndreas Gohr/**
13824c3e0d2SAndreas Gohr * Clear all index files
13924c3e0d2SAndreas Gohr */
14024c3e0d2SAndreas Gohrfunction _clearindex(){
14124c3e0d2SAndreas Gohr    global $conf;
14224c3e0d2SAndreas Gohr    _lock();
143*981f048aSGabriel Birke    _quietecho("Clearing index... ");
14466b95544STom N Harris    io_saveFile($conf['indexdir'].'/page.idx','');
14566b95544STom N Harris    $dir = @opendir($conf['indexdir']);
14666b95544STom N Harris    if($dir!==false){
14766b95544STom N Harris        while(($f = readdir($dir)) !== false){
14866b95544STom N Harris            if(substr($f,-4)=='.idx' &&
14966b95544STom N Harris               (substr($f,0,1)=='i' || substr($f,0,1)=='w'))
15066b95544STom N Harris                @unlink($conf['indexdir']."/$f");
15166b95544STom N Harris        }
15266b95544STom N Harris    }
153*981f048aSGabriel Birke    _quietecho("done.\n");
15424c3e0d2SAndreas Gohr    _unlock();
15524c3e0d2SAndreas Gohr}
15624c3e0d2SAndreas Gohr
157*981f048aSGabriel Birkefunction _quietecho($msg) {
158*981f048aSGabriel Birke    global $QUIET;
159*981f048aSGabriel Birke    if(!$QUIET) echo $msg;
160*981f048aSGabriel Birke}
161*981f048aSGabriel Birke
16224c3e0d2SAndreas Gohr//Setup VIM: ex: et ts=2 enc=utf-8 :
163