xref: /dokuwiki/inc/search.php (revision 5514a5a77e2d38cf763ed922b727ab305c8c8bb9) 
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 array with object,method)
20 * @param   string    $dir  Current directory beyond $base
21 * @param   int       $lvl  Recursion Level
22 * @param   mixed     $sort 'natural' to use natural order sorting (default); 'date' to sort by filemtime.
23 * @author  Andreas Gohr <andi@splitbrain.org>
24 */
25function search(&$data,$base,$func,$opts,$dir='',$lvl=1,$sort='natural'){
26    $dirs   = array();
27    $files  = array();
28    $filepaths = array();
29
30    //read in directories and files
31    $dh = @opendir($base.'/'.$dir);
32    if(!$dh) return;
33    while(($file = readdir($dh)) !== false){
34        if(preg_match('/^[\._]/',$file)) continue; //skip hidden files and upper dirs
35        if(is_dir($base.'/'.$dir.'/'.$file)){
36            $dirs[] = $dir.'/'.$file;
37            continue;
38        }
39        $files[] = $dir.'/'.$file;
40        $filepaths[] = $base.'/'.$dir.'/'.$file;
41    }
42    closedir($dh);
43    if ($sort == 'date') {
44        @array_multisort(array_map('filemtime', $filepaths), SORT_NUMERIC, SORT_DESC, $files);
45    } else {
46        sort($files);
47    }
48    sort($dirs);
49
50    //give directories to userfunction then recurse
51    foreach($dirs as $dir){
52        if (call_user_func_array($func, array(&$data,$base,$dir,'d',$lvl,$opts))){
53            search($data,$base,$func,$opts,$dir,$lvl+1,$sort);
54        }
55    }
56    //now handle the files
57    foreach($files as $file){
58        call_user_func_array($func, array(&$data,$base,$file,'f',$lvl,$opts));
59    }
60}
61
62/**
63 * Wrapper around call_user_func_array.
64 *
65 * @deprecated
66 */
67function search_callback($func,&$data,$base,$file,$type,$lvl,$opts){
68    return call_user_func_array($func, array(&$data,$base,$file,$type,$lvl,$opts));
69}
70
71/**
72 * The following functions are userfunctions to use with the search
73 * function above. This function is called for every found file or
74 * directory. When a directory is given to the function it has to
75 * decide if this directory should be traversed (true) or not (false)
76 * The function has to accept the following parameters:
77 *
78 * &$data - Reference to the result data structure
79 * $base  - Base usually $conf['datadir']
80 * $file  - current file or directory relative to $base
81 * $type  - Type either 'd' for directory or 'f' for file
82 * $lvl   - Current recursion depht
83 * $opts  - option array as given to search()
84 *
85 * return values for files are ignored
86 *
87 * All functions should check the ACL for document READ rights
88 * namespaces (directories) are NOT checked (when sneaky_index is 0) as this
89 * would break the recursion (You can have an nonreadable dir over a readable
90 * one deeper nested) also make sure to check the file type (for example
91 * in case of lockfiles).
92 */
93
94/**
95 * Searches for pages beginning with the given query
96 *
97 * @author Andreas Gohr <andi@splitbrain.org>
98 */
99function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
100    $opts = array(
101            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
102            'listfiles' => true,
103            'pagesonly' => true,
104            );
105    return search_universal($data,$base,$file,$type,$lvl,$opts);
106}
107
108/**
109 * Build the browsable index of pages
110 *
111 * $opts['ns'] is the currently viewed namespace
112 *
113 * @author  Andreas Gohr <andi@splitbrain.org>
114 */
115function search_index(&$data,$base,$file,$type,$lvl,$opts){
116    global $conf;
117    $opts = array(
118        'pagesonly' => true,
119        'listdirs' => true,
120        'listfiles' => !$opts['nofiles'],
121        'sneakyacl' => $conf['sneaky_index'],
122        // Hacky, should rather use recmatch
123        'depth' => preg_match('#^'.preg_quote($file, '#').'(/|$)#','/'.$opts['ns']) ? 0 : -1
124    );
125
126    return search_universal($data, $base, $file, $type, $lvl, $opts);
127}
128
129/**
130 * List all namespaces
131 *
132 * @author  Andreas Gohr <andi@splitbrain.org>
133 */
134function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
135    $opts = array(
136            'listdirs' => true,
137            );
138    return search_universal($data,$base,$file,$type,$lvl,$opts);
139}
140
141/**
142 * List all mediafiles in a namespace
143 *
144 * @author  Andreas Gohr <andi@splitbrain.org>
145 */
146function search_media(&$data,$base,$file,$type,$lvl,$opts){
147
148    //we do nothing with directories
149    if($type == 'd') {
150        if(!$opts['depth']) return true; // recurse forever
151        $depth = substr_count($file,'/');
152        if($depth >= $opts['depth']) return false; // depth reached
153        return true;
154    }
155
156    $info         = array();
157    $info['id']   = pathID($file,true);
158    if($info['id'] != cleanID($info['id'])){
159        if($opts['showmsg'])
160            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
161        return false; // skip non-valid files
162    }
163
164    //check ACL for namespace (we have no ACL for mediafiles)
165    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
166    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
167        return false;
168    }
169
170    //check pattern filter
171    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
172        return false;
173    }
174
175    $info['file']     = utf8_basename($file);
176    $info['size']     = filesize($base.'/'.$file);
177    $info['mtime']    = filemtime($base.'/'.$file);
178    $info['writable'] = is_writable($base.'/'.$file);
179    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
180        $info['isimg'] = true;
181        $info['meta']  = new JpegMeta($base.'/'.$file);
182    }else{
183        $info['isimg'] = false;
184    }
185    if($opts['hash']){
186        $info['hash'] = md5(io_readFile(mediaFN($info['id']),false));
187    }
188
189    $data[] = $info;
190
191    return false;
192}
193
194/**
195 * This function just lists documents (for RSS namespace export)
196 *
197 * @author  Andreas Gohr <andi@splitbrain.org>
198 */
199function search_list(&$data,$base,$file,$type,$lvl,$opts){
200    //we do nothing with directories
201    if($type == 'd') return false;
202    //only search txt files
203    if(substr($file,-4) == '.txt'){
204        //check ACL
205        $id = pathID($file);
206        if(auth_quickaclcheck($id) < AUTH_READ){
207            return false;
208        }
209        $data[]['id'] = $id;
210    }
211    return false;
212}
213
214/**
215 * Quicksearch for searching matching pagenames
216 *
217 * $opts['query'] is the search query
218 *
219 * @author  Andreas Gohr <andi@splitbrain.org>
220 */
221function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
222    //we do nothing with directories
223    if($type == 'd') return true;
224    //only search txt files
225    if(substr($file,-4) != '.txt') return true;
226
227    //simple stringmatching
228    if (!empty($opts['query'])){
229        if(strpos($file,$opts['query']) !== false){
230            //check ACL
231            $id = pathID($file);
232            if(auth_quickaclcheck($id) < AUTH_READ){
233                return false;
234            }
235            $data[]['id'] = $id;
236        }
237    }
238    return true;
239}
240
241/**
242 * Just lists all documents
243 *
244 * $opts['depth']   recursion level, 0 for all
245 * $opts['hash']    do md5 sum of content?
246 * $opts['skipacl'] list everything regardless of ACL
247 *
248 * @author  Andreas Gohr <andi@splitbrain.org>
249 */
250function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
251    if(isset($opts['depth']) && $opts['depth']){
252        $parts = explode('/',ltrim($file,'/'));
253        if(($type == 'd' && count($parts) > $opts['depth'])
254          || ($type != 'd' && count($parts) > $opts['depth'] + 1)){
255            return false; // depth reached
256        }
257    }
258
259    //we do nothing with directories
260    if($type == 'd'){
261        return true;
262    }
263
264    //only search txt files
265    if(substr($file,-4) != '.txt') return true;
266
267    $item['id']   = pathID($file);
268    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
269        return false;
270    }
271
272    $item['rev']   = filemtime($base.'/'.$file);
273    $item['mtime'] = $item['rev'];
274    $item['size']  = filesize($base.'/'.$file);
275    if($opts['hash']){
276        $item['hash'] = md5(trim(rawWiki($item['id'])));
277    }
278
279    $data[] = $item;
280    return true;
281}
282
283/**
284 * Search for backlinks to a given page
285 *
286 * $opts['ns']    namespace of the page
287 * $opts['name']  name of the page without namespace
288 *
289 * @author  Andreas Gohr <andi@splitbrain.org>
290 * @deprecated Replaced by ft_backlinks()
291 */
292function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
293    //we do nothing with directories
294    if($type == 'd') return true;
295    //only search txt files
296    if(substr($file,-4) != '.txt') return true;
297
298    //absolute search id
299    $sid = cleanID($opts['ns'].':'.$opts['name']);
300
301    //current id and namespace
302    $cid = pathID($file);
303    $cns = getNS($cid);
304
305    //check ACL
306    if(auth_quickaclcheck($cid) < AUTH_READ){
307        return false;
308    }
309
310    //fetch instructions
311    $instructions = p_cached_instructions($base.$file,true);
312    if(is_null($instructions)) return false;
313
314    global $conf;
315    //check all links for match
316    foreach($instructions as $ins){
317        if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
318            $mid = $ins[1][0];
319            resolve_pageid($cns,$mid,$exists); //exists is not used
320            if($mid == $sid){
321                //we have a match - finish
322                $data[]['id'] = $cid;
323                break;
324            }
325        }
326    }
327
328    return false;
329}
330
331/**
332 * Fulltextsearch
333 *
334 * $opts['query'] is the search query
335 *
336 * @author  Andreas Gohr <andi@splitbrain.org>
337 * @deprecated - fulltext indexer is used instead
338 */
339function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
340    //we do nothing with directories
341    if($type == 'd') return true;
342    //only search txt files
343    if(substr($file,-4) != '.txt') return true;
344
345    //check ACL
346    $id = pathID($file);
347    if(auth_quickaclcheck($id) < AUTH_READ){
348        return false;
349    }
350
351    //create regexp from queries
352    $poswords = array();
353    $negwords = array();
354    $qpreg = preg_split('/\s+/',$opts['query']);
355
356    foreach($qpreg as $word){
357        switch(substr($word,0,1)){
358            case '-':
359                if(strlen($word) > 1){  // catch single '-'
360                    array_push($negwords,preg_quote(substr($word,1),'#'));
361                }
362                break;
363            case '+':
364                if(strlen($word) > 1){  // catch single '+'
365                    array_push($poswords,preg_quote(substr($word,1),'#'));
366                }
367                break;
368            default:
369                array_push($poswords,preg_quote($word,'#'));
370                break;
371        }
372    }
373
374    // a search without any posword is useless
375    if (!count($poswords)) return true;
376
377    $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
378            $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
379            search_regex($data,$base,$file,$reg,$poswords);
380            return true;
381            }
382
383            /**
384             * Reference search
385             * This fuction searches for existing references to a given media file
386             * and returns an array with the found pages. It doesn't pay any
387             * attention to ACL permissions to find every reference. The caller
388             * must check if the user has the appropriate rights to see the found
389             * page and eventually have to prevent the result from displaying.
390             *
391             * @param array  $data Reference to the result data structure
392             * @param string $base Base usually $conf['datadir']
393             * @param string $file current file or directory relative to $base
394             * @param char   $type Type either 'd' for directory or 'f' for file
395             * @param int    $lvl  Current recursion depht
396             * @param mixed  $opts option array as given to search()
397             *
398             * $opts['query'] is the demanded media file name
399             *
400             * @author  Andreas Gohr <andi@splitbrain.org>
401             * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
402             */
403function search_reference(&$data,$base,$file,$type,$lvl,$opts){
404    global $conf;
405
406    //we do nothing with directories
407    if($type == 'd') return true;
408
409    //only search txt files
410    if(substr($file,-4) != '.txt') return true;
411
412    //we finish after 'cnt' references found. The return value
413    //'false' will skip subdirectories to speed search up.
414    $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
415    if(count($data) >= $cnt) return false;
416
417    $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
418    search_regex($data,$base,$file,$reg,array($opts['query']));
419    return true;
420}
421
422/* ------------- helper functions below -------------- */
423
424/**
425 * fulltext search helper
426 * searches a text file with a given regular expression
427 * no ACL checks are performed. This have to be done by
428 * the caller if necessary.
429 *
430 * @param array  $data  reference to array for results
431 * @param string $base  base directory
432 * @param string $file  file name to search in
433 * @param string $reg   regular expression to search for
434 * @param array  $words words that should be marked in the results
435 *
436 * @author  Andreas Gohr <andi@splitbrain.org>
437 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
438 *
439 * @deprecated - fulltext indexer is used instead
440 */
441function search_regex(&$data,$base,$file,$reg,$words){
442
443    //get text
444    $text = io_readfile($base.'/'.$file);
445    //lowercase text (u modifier does not help with case)
446    $lctext = utf8_strtolower($text);
447
448    //do the fulltext search
449    $matches = array();
450    if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
451        //this is not the best way for snippet generation but the fastest I could find
452        $q = $words[0];  //use first word for snippet creation
453        $p = utf8_strpos($lctext,$q);
454        $f = $p - 100;
455        $l = utf8_strlen($q) + 200;
456        if($f < 0) $f = 0;
457        $snippet = '<span class="search_sep"> ... </span>'.
458            htmlspecialchars(utf8_substr($text,$f,$l)).
459            '<span class="search_sep"> ... </span>';
460        $mark    = '('.join('|', $words).')';
461        $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
462
463        $data[] = array(
464                'id'       => pathID($file),
465                'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
466                'poswords' => join(' ',$words),
467                'snippet'  => $snippet,
468                );
469    }
470
471    return true;
472}
473
474
475/**
476 * fulltext sort
477 *
478 * Callback sort function for use with usort to sort the data
479 * structure created by search_fulltext. Sorts descending by count
480 *
481 * @author  Andreas Gohr <andi@splitbrain.org>
482 */
483function sort_search_fulltext($a,$b){
484    if($a['count'] > $b['count']){
485        return -1;
486    }elseif($a['count'] < $b['count']){
487        return 1;
488    }else{
489        return strcmp($a['id'],$b['id']);
490    }
491}
492
493/**
494 * translates a document path to an ID
495 *
496 * @author  Andreas Gohr <andi@splitbrain.org>
497 * @todo    move to pageutils
498 */
499function pathID($path,$keeptxt=false){
500    $id = utf8_decodeFN($path);
501    $id = str_replace('/',':',$id);
502    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
503    $id = trim($id, ':');
504    return $id;
505}
506
507
508/**
509 * This is a very universal callback for the search() function, replacing
510 * many of the former individual functions at the cost of a more complex
511 * setup.
512 *
513 * How the function behaves, depends on the options passed in the $opts
514 * array, where the following settings can be used.
515 *
516 * depth      int     recursion depth. 0 for unlimited
517 * keeptxt    bool    keep .txt extension for IDs
518 * listfiles  bool    include files in listing
519 * listdirs   bool    include namespaces in listing
520 * pagesonly  bool    restrict files to pages
521 * skipacl    bool    do not check for READ permission
522 * sneakyacl  bool    don't recurse into nonreadable dirs
523 * hash       bool    create MD5 hash for files
524 * meta       bool    return file metadata
525 * filematch  string  match files against this regexp
526 * idmatch    string  match full ID against this regexp
527 * dirmatch   string  match directory against this regexp when adding
528 * nsmatch    string  match namespace against this regexp when adding
529 * recmatch   string  match directory against this regexp when recursing
530 * showmsg    bool    warn about non-ID files
531 * showhidden bool    show hidden files too
532 * firsthead  bool    return first heading for pages
533 *
534 * @author Andreas Gohr <gohr@cosmocode.de>
535 */
536function search_universal(&$data,$base,$file,$type,$lvl,$opts){
537    $item   = array();
538    $return = true;
539
540    // get ID and check if it is a valid one
541    $item['id'] = pathID($file,($type == 'd' || $opts['keeptxt']));
542    if($item['id'] != cleanID($item['id'])){
543        if($opts['showmsg'])
544            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
545        return false; // skip non-valid files
546    }
547    $item['ns']  = getNS($item['id']);
548
549    if($type == 'd') {
550        // decide if to recursion into this directory is wanted
551        if(!$opts['depth']){
552            $return = true; // recurse forever
553        }else{
554            $depth = substr_count($file,'/');
555            if($depth >= $opts['depth']){
556                $return = false; // depth reached
557            }else{
558                $return = true;
559            }
560        }
561        if($return && !preg_match('/'.$opts['recmatch'].'/',$file)){
562            $return = false; // doesn't match
563        }
564    }
565
566    // check ACL
567    if(!$opts['skipacl']){
568        if($type == 'd'){
569            $item['perm'] = auth_quickaclcheck($item['id'].':*');
570        }else{
571            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
572        }
573    }else{
574        $item['perm'] = AUTH_DELETE;
575    }
576
577    // are we done here maybe?
578    if($type == 'd'){
579        if(!$opts['listdirs']) return $return;
580        if(!$opts['skipacl'] && $opts['sneakyacl'] && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
581        if($opts['dirmatch'] && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
582        if($opts['nsmatch'] && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
583    }else{
584        if(!$opts['listfiles']) return $return;
585        if(!$opts['skipacl'] && $item['perm'] < AUTH_READ) return $return;
586        if($opts['pagesonly'] && (substr($file,-4) != '.txt')) return $return;
587        if(!$opts['showhidden'] && isHiddenPage($item['id'])) return $return;
588        if($opts['filematch'] && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
589        if($opts['idmatch'] && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
590    }
591
592    // still here? prepare the item
593    $item['type']  = $type;
594    $item['level'] = $lvl;
595    $item['open']  = $return;
596
597    if($opts['meta']){
598        $item['file']       = utf8_basename($file);
599        $item['size']       = filesize($base.'/'.$file);
600        $item['mtime']      = filemtime($base.'/'.$file);
601        $item['rev']        = $item['mtime'];
602        $item['writable']   = is_writable($base.'/'.$file);
603        $item['executable'] = is_executable($base.'/'.$file);
604    }
605
606    if($type == 'f'){
607        if($opts['hash']) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
608        if($opts['firsthead']) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
609    }
610
611    // finally add the item
612    $data[] = $item;
613    return $return;
614}
615
616//Setup VIM: ex: et ts=4 :
617