xref: /dokuwiki/inc/search.php (revision 44d638d049d687159666813023108e0fa9f32181) 
1<?php
2/**
3 * DokuWiki search functions
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Andreas Gohr <andi@splitbrain.org>
7 */
8
9if(!defined('DOKU_INC')) die('meh.');
10
11/**
12 * recurse direcory
13 *
14 * This function recurses into a given base directory
15 * and calls the supplied function for each file and directory
16 *
17 * @param   array ref $data The results of the search are stored here
18 * @param   string    $base Where to start the search
19 * @param   callback  $func Callback (function name or arayy with object,method)
20 * @param   string    $dir  Current directory beyond $base
21 * @param   int       $lvl  Recursion Level
22 * @author  Andreas Gohr <andi@splitbrain.org>
23 */
24function search(&$data,$base,$func,$opts,$dir='',$lvl=1,$sort=false){
25    $dirs   = array();
26    $files  = array();
27    $filepaths = array();
28
29    //read in directories and files
30    $dh = @opendir($base.'/'.$dir);
31    if(!$dh) return;
32    while(($file = readdir($dh)) !== false){
33        if(preg_match('/^[\._]/',$file)) continue; //skip hidden files and upper dirs
34        if(is_dir($base.'/'.$dir.'/'.$file)){
35            $dirs[] = $dir.'/'.$file;
36            continue;
37        }
38        $files[] = $dir.'/'.$file;
39        $filepaths[] = $base.'/'.$dir.'/'.$file;
40    }
41    closedir($dh);
42    if ($sort == 'date') {
43        @array_multisort(array_map('filemtime', $filepaths), SORT_NUMERIC, SORT_DESC, $files);
44    } else {
45        sort($files);
46    }
47    sort($dirs);
48
49    //give directories to userfunction then recurse
50    foreach($dirs as $dir){
51        if (call_user_func_array($func, array(&$data,$base,$dir,'d',$lvl,$opts))){
52            search($data,$base,$func,$opts,$dir,$lvl+1);
53        }
54    }
55    //now handle the files
56    foreach($files as $file){
57        call_user_func_array($func, array(&$data,$base,$file,'f',$lvl,$opts));
58    }
59}
60
61/**
62 * Wrapper around call_user_func_array.
63 *
64 * @deprecated
65 */
66function search_callback($func,&$data,$base,$file,$type,$lvl,$opts){
67    return call_user_func_array($func, array(&$data,$base,$file,$type,$lvl,$opts));
68}
69
70/**
71 * The following functions are userfunctions to use with the search
72 * function above. This function is called for every found file or
73 * directory. When a directory is given to the function it has to
74 * decide if this directory should be traversed (true) or not (false)
75 * The function has to accept the following parameters:
76 *
77 * &$data - Reference to the result data structure
78 * $base  - Base usually $conf['datadir']
79 * $file  - current file or directory relative to $base
80 * $type  - Type either 'd' for directory or 'f' for file
81 * $lvl   - Current recursion depht
82 * $opts  - option array as given to search()
83 *
84 * return values for files are ignored
85 *
86 * All functions should check the ACL for document READ rights
87 * namespaces (directories) are NOT checked as this would break
88 * the recursion (You can have an nonreadable dir over a readable
89 * one deeper nested) also make sure to check the file type (for example
90 * in case of lockfiles).
91 */
92
93/**
94 * Searches for pages beginning with the given query
95 *
96 * @author Andreas Gohr <andi@splitbrain.org>
97 */
98function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
99    $opts = array(
100            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
101            'listfiles' => true,
102            'pagesonly' => true,
103            );
104    return search_universal($data,$base,$file,$type,$lvl,$opts);
105}
106
107/**
108 * Build the browsable index of pages
109 *
110 * $opts['ns'] is the current namespace
111 *
112 * @author  Andreas Gohr <andi@splitbrain.org>
113 */
114function search_index(&$data,$base,$file,$type,$lvl,$opts){
115    global $conf;
116    $return = true;
117
118    $item = array();
119
120    if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
121        //add but don't recurse
122        $return = false;
123    }elseif($type == 'f' && ($opts['nofiles'] || substr($file,-4) != '.txt')){
124        //don't add
125        return false;
126    }
127
128    $id = pathID($file,($type == 'd'));
129
130    if($type=='d' && $conf['sneaky_index'] && auth_quickaclcheck($id.':') < AUTH_READ){
131        return false;
132    }
133
134    //check hidden
135    if(isHiddenPage($id)){
136        return false;
137    }
138
139    //check ACL
140    if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
141        return false;
142    }
143
144    $data[]=array( 'id'    => $id,
145            'type'  => $type,
146            'level' => $lvl,
147            'open'  => $return );
148    return $return;
149}
150
151/**
152 * List all namespaces
153 *
154 * @author  Andreas Gohr <andi@splitbrain.org>
155 */
156function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
157    $opts = array(
158            'listdirs' => true,
159            );
160    return search_universal($data,$base,$file,$type,$lvl,$opts);
161}
162
163/**
164 * List all mediafiles in a namespace
165 *
166 * @author  Andreas Gohr <andi@splitbrain.org>
167 */
168function search_media(&$data,$base,$file,$type,$lvl,$opts){
169
170    //we do nothing with directories
171    if($type == 'd') {
172        if(!$opts['depth']) return true; // recurse forever
173        $depth = substr_count($file,'/');
174        if($depth >= $opts['depth']) return false; // depth reached
175        return true;
176    }
177
178    $info         = array();
179    $info['id']   = pathID($file,true);
180    if($info['id'] != cleanID($info['id'])){
181        if($opts['showmsg'])
182            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
183        return false; // skip non-valid files
184    }
185
186    //check ACL for namespace (we have no ACL for mediafiles)
187    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
188    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
189        return false;
190    }
191
192    //check pattern filter
193    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
194        return false;
195    }
196
197    $info['file']     = basename($file);
198    $info['size']     = filesize($base.'/'.$file);
199    $info['mtime']    = filemtime($base.'/'.$file);
200    $info['writable'] = is_writable($base.'/'.$file);
201    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
202        $info['isimg'] = true;
203        $info['meta']  = new JpegMeta($base.'/'.$file);
204    }else{
205        $info['isimg'] = false;
206    }
207    if($opts['hash']){
208        $info['hash'] = md5(io_readFile(mediaFN($info['id']),false));
209    }
210
211    $data[] = $info;
212
213    return false;
214}
215
216/**
217 * This function just lists documents (for RSS namespace export)
218 *
219 * @author  Andreas Gohr <andi@splitbrain.org>
220 */
221function search_list(&$data,$base,$file,$type,$lvl,$opts){
222    //we do nothing with directories
223    if($type == 'd') return false;
224    //only search txt files
225    if(substr($file,-4) == '.txt'){
226        //check ACL
227        $id = pathID($file);
228        if(auth_quickaclcheck($id) < AUTH_READ){
229            return false;
230        }
231        $data[]['id'] = $id;
232    }
233    return false;
234}
235
236/**
237 * Quicksearch for searching matching pagenames
238 *
239 * $opts['query'] is the search query
240 *
241 * @author  Andreas Gohr <andi@splitbrain.org>
242 */
243function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
244    //we do nothing with directories
245    if($type == 'd') return true;
246    //only search txt files
247    if(substr($file,-4) != '.txt') return true;
248
249    //simple stringmatching
250    if (!empty($opts['query'])){
251        if(strpos($file,$opts['query']) !== false){
252            //check ACL
253            $id = pathID($file);
254            if(auth_quickaclcheck($id) < AUTH_READ){
255                return false;
256            }
257            $data[]['id'] = $id;
258        }
259    }
260    return true;
261}
262
263/**
264 * Just lists all documents
265 *
266 * $opts['depth']   recursion level, 0 for all
267 * $opts['hash']    do md5 sum of content?
268 * $opts['skipacl'] list everything regardless of ACL
269 *
270 * @author  Andreas Gohr <andi@splitbrain.org>
271 */
272function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
273    //we do nothing with directories
274    if($type == 'd'){
275        if(!$opts['depth']) return true; // recurse forever
276        $parts = explode('/',ltrim($file,'/'));
277        if(count($parts) == $opts['depth']) return false; // depth reached
278        return true;
279    }
280
281    //only search txt files
282    if(substr($file,-4) != '.txt') return true;
283
284    $item['id']   = pathID($file);
285    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
286        return false;
287    }
288
289    $item['rev']   = filemtime($base.'/'.$file);
290    $item['mtime'] = $item['rev'];
291    $item['size']  = filesize($base.'/'.$file);
292    if($opts['hash']){
293        $item['hash'] = md5(trim(rawWiki($item['id'])));
294    }
295
296    $data[] = $item;
297    return true;
298}
299
300/**
301 * Search for backlinks to a given page
302 *
303 * $opts['ns']    namespace of the page
304 * $opts['name']  name of the page without namespace
305 *
306 * @author  Andreas Gohr <andi@splitbrain.org>
307 * @deprecated Replaced by ft_backlinks()
308 */
309function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
310    //we do nothing with directories
311    if($type == 'd') return true;
312    //only search txt files
313    if(substr($file,-4) != '.txt') return true;
314
315    //absolute search id
316    $sid = cleanID($opts['ns'].':'.$opts['name']);
317
318    //current id and namespace
319    $cid = pathID($file);
320    $cns = getNS($cid);
321
322    //check ACL
323    if(auth_quickaclcheck($cid) < AUTH_READ){
324        return false;
325    }
326
327    //fetch instructions
328    $instructions = p_cached_instructions($base.$file,true);
329    if(is_null($instructions)) return false;
330
331    global $conf;
332    //check all links for match
333    foreach($instructions as $ins){
334        if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
335            $mid = $ins[1][0];
336            resolve_pageid($cns,$mid,$exists); //exists is not used
337            if($mid == $sid){
338                //we have a match - finish
339                $data[]['id'] = $cid;
340                break;
341            }
342        }
343    }
344
345    return false;
346}
347
348/**
349 * Fulltextsearch
350 *
351 * $opts['query'] is the search query
352 *
353 * @author  Andreas Gohr <andi@splitbrain.org>
354 * @deprecated - fulltext indexer is used instead
355 */
356function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
357    //we do nothing with directories
358    if($type == 'd') return true;
359    //only search txt files
360    if(substr($file,-4) != '.txt') return true;
361
362    //check ACL
363    $id = pathID($file);
364    if(auth_quickaclcheck($id) < AUTH_READ){
365        return false;
366    }
367
368    //create regexp from queries
369    $poswords = array();
370    $negwords = array();
371    $qpreg = preg_split('/\s+/',$opts['query']);
372
373    foreach($qpreg as $word){
374        switch(substr($word,0,1)){
375            case '-':
376                if(strlen($word) > 1){  // catch single '-'
377                    array_push($negwords,preg_quote(substr($word,1),'#'));
378                }
379                break;
380            case '+':
381                if(strlen($word) > 1){  // catch single '+'
382                    array_push($poswords,preg_quote(substr($word,1),'#'));
383                }
384                break;
385            default:
386                array_push($poswords,preg_quote($word,'#'));
387                break;
388        }
389    }
390
391    // a search without any posword is useless
392    if (!count($poswords)) return true;
393
394    $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
395            $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
396            search_regex($data,$base,$file,$reg,$poswords);
397            return true;
398            }
399
400            /**
401             * Reference search
402             * This fuction searches for existing references to a given media file
403             * and returns an array with the found pages. It doesn't pay any
404             * attention to ACL permissions to find every reference. The caller
405             * must check if the user has the appropriate rights to see the found
406             * page and eventually have to prevent the result from displaying.
407             *
408             * @param array  $data Reference to the result data structure
409             * @param string $base Base usually $conf['datadir']
410             * @param string $file current file or directory relative to $base
411             * @param char   $type Type either 'd' for directory or 'f' for file
412             * @param int    $lvl  Current recursion depht
413             * @param mixed  $opts option array as given to search()
414             *
415             * $opts['query'] is the demanded media file name
416             *
417             * @author  Andreas Gohr <andi@splitbrain.org>
418             * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
419             */
420function search_reference(&$data,$base,$file,$type,$lvl,$opts){
421    global $conf;
422
423    //we do nothing with directories
424    if($type == 'd') return true;
425
426    //only search txt files
427    if(substr($file,-4) != '.txt') return true;
428
429    //we finish after 'cnt' references found. The return value
430    //'false' will skip subdirectories to speed search up.
431    $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
432    if(count($data) >= $cnt) return false;
433
434    $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
435    search_regex($data,$base,$file,$reg,array($opts['query']));
436    return true;
437}
438
439/* ------------- helper functions below -------------- */
440
441/**
442 * fulltext search helper
443 * searches a text file with a given regular expression
444 * no ACL checks are performed. This have to be done by
445 * the caller if necessary.
446 *
447 * @param array  $data  reference to array for results
448 * @param string $base  base directory
449 * @param string $file  file name to search in
450 * @param string $reg   regular expression to search for
451 * @param array  $words words that should be marked in the results
452 *
453 * @author  Andreas Gohr <andi@splitbrain.org>
454 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
455 *
456 * @deprecated - fulltext indexer is used instead
457 */
458function search_regex(&$data,$base,$file,$reg,$words){
459
460    //get text
461    $text = io_readfile($base.'/'.$file);
462    //lowercase text (u modifier does not help with case)
463    $lctext = utf8_strtolower($text);
464
465    //do the fulltext search
466    $matches = array();
467    if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
468        //this is not the best way for snippet generation but the fastest I could find
469        $q = $words[0];  //use first word for snippet creation
470        $p = utf8_strpos($lctext,$q);
471        $f = $p - 100;
472        $l = utf8_strlen($q) + 200;
473        if($f < 0) $f = 0;
474        $snippet = '<span class="search_sep"> ... </span>'.
475            htmlspecialchars(utf8_substr($text,$f,$l)).
476            '<span class="search_sep"> ... </span>';
477        $mark    = '('.join('|', $words).')';
478        $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
479
480        $data[] = array(
481                'id'       => pathID($file),
482                'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
483                'poswords' => join(' ',$words),
484                'snippet'  => $snippet,
485                );
486    }
487
488    return true;
489}
490
491
492/**
493 * fulltext sort
494 *
495 * Callback sort function for use with usort to sort the data
496 * structure created by search_fulltext. Sorts descending by count
497 *
498 * @author  Andreas Gohr <andi@splitbrain.org>
499 */
500function sort_search_fulltext($a,$b){
501    if($a['count'] > $b['count']){
502        return -1;
503    }elseif($a['count'] < $b['count']){
504        return 1;
505    }else{
506        return strcmp($a['id'],$b['id']);
507    }
508}
509
510/**
511 * translates a document path to an ID
512 *
513 * @author  Andreas Gohr <andi@splitbrain.org>
514 * @todo    move to pageutils
515 */
516function pathID($path,$keeptxt=false){
517    $id = utf8_decodeFN($path);
518    $id = str_replace('/',':',$id);
519    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
520    $id = trim($id, ':');
521    return $id;
522}
523
524
525/**
526 * This is a very universal callback for the search() function, replacing
527 * many of the former individual functions at the cost of a more complex
528 * setup.
529 *
530 * How the function behaves, depends on the options passed in the $opts
531 * array, where the following settings can be used.
532 *
533 * depth      int     recursion depth. 0 for unlimited
534 * keeptxt    bool    keep .txt extension for IDs
535 * listfiles  bool    include files in listing
536 * listdirs   bool    include namespaces in listing
537 * pagesonly  bool    restrict files to pages
538 * skipacl    bool    do not check for READ permission
539 * sneakyacl  bool    don't recurse into nonreadable dirs
540 * hash       bool    create MD5 hash for files
541 * meta       bool    return file metadata
542 * filematch  string  match files against this regexp
543 * idmatch    string  match full ID against this regexp
544 * dirmatch   string  match directory against this regexp when adding
545 * nsmatch    string  match namespace against this regexp when adding
546 * recmatch   string  match directory against this regexp when recursing
547 * showmsg    bool    warn about non-ID files
548 * showhidden bool    show hidden files too
549 * firsthead  bool    return first heading for pages
550 *
551 * @author Andreas Gohr <gohr@cosmocode.de>
552 */
553function search_universal(&$data,$base,$file,$type,$lvl,$opts){
554    $item   = array();
555    $return = true;
556
557    // get ID and check if it is a valid one
558    $item['id'] = pathID($file,($type == 'd' || $opts['keeptxt']));
559    if($item['id'] != cleanID($item['id'])){
560        if($opts['showmsg'])
561            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
562        return false; // skip non-valid files
563    }
564    $item['ns']  = getNS($item['id']);
565
566    if($type == 'd') {
567        // decide if to recursion into this directory is wanted
568        if(!$opts['depth']){
569            $return = true; // recurse forever
570        }else{
571            $depth = substr_count($file,'/');
572            if($depth >= $opts['depth']){
573                $return = false; // depth reached
574            }else{
575                $return = true;
576            }
577        }
578        if($return && !preg_match('/'.$opts['recmatch'].'/',$file)){
579            $return = false; // doesn't match
580        }
581    }
582
583    // check ACL
584    if(!$opts['skipacl']){
585        if($type == 'd'){
586            $item['perm'] = auth_quickaclcheck($item['id'].':*');
587        }else{
588            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
589        }
590    }else{
591        $item['perm'] = AUTH_DELETE;
592    }
593
594    // are we done here maybe?
595    if($type == 'd'){
596        if(!$opts['listdirs']) return $return;
597        if(!$opts['skipacl'] && $opts['sneakyacl'] && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
598        if($opts['dirmatch'] && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
599        if($opts['nsmatch'] && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
600    }else{
601        if(!$opts['listfiles']) return $return;
602        if(!$opts['skipacl'] && $item['perm'] < AUTH_READ) return $return;
603        if($opts['pagesonly'] && (substr($file,-4) != '.txt')) return $return;
604        if(!$opts['showhidden'] && isHiddenPage($item['id'])) return $return;
605        if($opts['filematch'] && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
606        if($opts['idmatch'] && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
607    }
608
609    // still here? prepare the item
610    $item['type']  = $type;
611    $item['level'] = $lvl;
612    $item['open']  = $return;
613
614    if($opts['meta']){
615        $item['file']       = basename($file);
616        $item['size']       = filesize($base.'/'.$file);
617        $item['mtime']      = filemtime($base.'/'.$file);
618        $item['rev']        = $item['mtime'];
619        $item['writable']   = is_writable($base.'/'.$file);
620        $item['executable'] = is_executable($base.'/'.$file);
621    }
622
623    if($type == 'f'){
624        if($opts['hash']) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
625        if($opts['firsthead']) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
626    }
627
628    // finally add the item
629    $data[] = $item;
630    return $return;
631}
632
633//Setup VIM: ex: et ts=4 :
634