xref: /dokuwiki/inc/search.php (revision 529b04166c604b1d086cbfeac1bd676227d04872) 
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 * $opts  - option array as given to search()
83 *
84 * return values for files are ignored
85 *
86 * All functions should check the ACL for document READ rights
87 * namespaces (directories) are NOT checked (when sneaky_index is 0) as this
88 * would break the recursion (You can have an nonreadable dir over a readable
89 * one deeper nested) also make sure to check the file type (for example
90 * in case of lockfiles).
91 */
92
93/**
94 * Searches for pages beginning with the given query
95 *
96 * @author Andreas Gohr <andi@splitbrain.org>
97 */
98function search_qsearch(&$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 * $opts['ns'] is the currently viewed namespace
111 *
112 * @author  Andreas Gohr <andi@splitbrain.org>
113 */
114function search_index(&$data,$base,$file,$type,$lvl,$opts){
115    global $conf;
116    $opts = array(
117        'pagesonly' => true,
118        'listdirs' => true,
119        'listfiles' => !$opts['nofiles'],
120        'sneakyacl' => $conf['sneaky_index'],
121        // Hacky, should rather use recmatch
122        'depth' => preg_match('#^'.preg_quote($file, '#').'(/|$)#','/'.$opts['ns']) ? 0 : -1
123    );
124
125    return search_universal($data, $base, $file, $type, $lvl, $opts);
126}
127
128/**
129 * List all namespaces
130 *
131 * @author  Andreas Gohr <andi@splitbrain.org>
132 */
133function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
134    $opts = array(
135            'listdirs' => true,
136            );
137    return search_universal($data,$base,$file,$type,$lvl,$opts);
138}
139
140/**
141 * List all mediafiles in a namespace
142 *
143 * @author  Andreas Gohr <andi@splitbrain.org>
144 */
145function search_media(&$data,$base,$file,$type,$lvl,$opts){
146
147    //we do nothing with directories
148    if($type == 'd') {
149        if(!$opts['depth']) return true; // recurse forever
150        $depth = substr_count($file,'/');
151        if($depth >= $opts['depth']) return false; // depth reached
152        return true;
153    }
154
155    $info         = array();
156    $info['id']   = pathID($file,true);
157    if($info['id'] != cleanID($info['id'])){
158        if($opts['showmsg'])
159            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
160        return false; // skip non-valid files
161    }
162
163    //check ACL for namespace (we have no ACL for mediafiles)
164    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
165    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
166        return false;
167    }
168
169    //check pattern filter
170    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
171        return false;
172    }
173
174    $info['file']     = utf8_basename($file);
175    $info['size']     = filesize($base.'/'.$file);
176    $info['mtime']    = filemtime($base.'/'.$file);
177    $info['writable'] = is_writable($base.'/'.$file);
178    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
179        $info['isimg'] = true;
180        $info['meta']  = new JpegMeta($base.'/'.$file);
181    }else{
182        $info['isimg'] = false;
183    }
184    if($opts['hash']){
185        $info['hash'] = md5(io_readFile(mediaFN($info['id']),false));
186    }
187
188    $data[] = $info;
189
190    return false;
191}
192
193/**
194 * This function just lists documents (for RSS namespace export)
195 *
196 * @author  Andreas Gohr <andi@splitbrain.org>
197 */
198function search_list(&$data,$base,$file,$type,$lvl,$opts){
199    //we do nothing with directories
200    if($type == 'd') return false;
201    //only search txt files
202    if(substr($file,-4) == '.txt'){
203        //check ACL
204        $id = pathID($file);
205        if(auth_quickaclcheck($id) < AUTH_READ){
206            return false;
207        }
208        $data[]['id'] = $id;
209    }
210    return false;
211}
212
213/**
214 * Quicksearch for searching matching pagenames
215 *
216 * $opts['query'] is the search query
217 *
218 * @author  Andreas Gohr <andi@splitbrain.org>
219 */
220function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
221    //we do nothing with directories
222    if($type == 'd') return true;
223    //only search txt files
224    if(substr($file,-4) != '.txt') return true;
225
226    //simple stringmatching
227    if (!empty($opts['query'])){
228        if(strpos($file,$opts['query']) !== false){
229            //check ACL
230            $id = pathID($file);
231            if(auth_quickaclcheck($id) < AUTH_READ){
232                return false;
233            }
234            $data[]['id'] = $id;
235        }
236    }
237    return true;
238}
239
240/**
241 * Just lists all documents
242 *
243 * $opts['depth']   recursion level, 0 for all
244 * $opts['hash']    do md5 sum of content?
245 * $opts['skipacl'] list everything regardless of ACL
246 *
247 * @author  Andreas Gohr <andi@splitbrain.org>
248 */
249function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
250    if(isset($opts['depth']) && $opts['depth']){
251        $parts = explode('/',ltrim($file,'/'));
252        if(($type == 'd' && count($parts) > $opts['depth'])
253          || ($type != 'd' && count($parts) > $opts['depth'] + 1)){
254            return false; // depth reached
255        }
256    }
257
258    //we do nothing with directories
259    if($type == 'd'){
260        return true;
261    }
262
263    //only search txt files
264    if(substr($file,-4) != '.txt') return true;
265
266    $item['id']   = pathID($file);
267    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
268        return false;
269    }
270
271    $item['rev']   = filemtime($base.'/'.$file);
272    $item['mtime'] = $item['rev'];
273    $item['size']  = filesize($base.'/'.$file);
274    if($opts['hash']){
275        $item['hash'] = md5(trim(rawWiki($item['id'])));
276    }
277
278    $data[] = $item;
279    return true;
280}
281
282/**
283 * Search for backlinks to a given page
284 *
285 * $opts['ns']    namespace of the page
286 * $opts['name']  name of the page without namespace
287 *
288 * @author  Andreas Gohr <andi@splitbrain.org>
289 * @deprecated Replaced by ft_backlinks()
290 */
291function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
292    //we do nothing with directories
293    if($type == 'd') return true;
294    //only search txt files
295    if(substr($file,-4) != '.txt') return true;
296
297    //absolute search id
298    $sid = cleanID($opts['ns'].':'.$opts['name']);
299
300    //current id and namespace
301    $cid = pathID($file);
302    $cns = getNS($cid);
303
304    //check ACL
305    if(auth_quickaclcheck($cid) < AUTH_READ){
306        return false;
307    }
308
309    //fetch instructions
310    $instructions = p_cached_instructions($base.$file,true);
311    if(is_null($instructions)) return false;
312
313    global $conf;
314    //check all links for match
315    foreach($instructions as $ins){
316        if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
317            $mid = $ins[1][0];
318            resolve_pageid($cns,$mid,$exists); //exists is not used
319            if($mid == $sid){
320                //we have a match - finish
321                $data[]['id'] = $cid;
322                break;
323            }
324        }
325    }
326
327    return false;
328}
329
330/**
331 * Fulltextsearch
332 *
333 * $opts['query'] is the search query
334 *
335 * @author  Andreas Gohr <andi@splitbrain.org>
336 * @deprecated - fulltext indexer is used instead
337 */
338function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
339    //we do nothing with directories
340    if($type == 'd') return true;
341    //only search txt files
342    if(substr($file,-4) != '.txt') return true;
343
344    //check ACL
345    $id = pathID($file);
346    if(auth_quickaclcheck($id) < AUTH_READ){
347        return false;
348    }
349
350    //create regexp from queries
351    $poswords = array();
352    $negwords = array();
353    $qpreg = preg_split('/\s+/',$opts['query']);
354
355    foreach($qpreg as $word){
356        switch(substr($word,0,1)){
357            case '-':
358                if(strlen($word) > 1){  // catch single '-'
359                    array_push($negwords,preg_quote(substr($word,1),'#'));
360                }
361                break;
362            case '+':
363                if(strlen($word) > 1){  // catch single '+'
364                    array_push($poswords,preg_quote(substr($word,1),'#'));
365                }
366                break;
367            default:
368                array_push($poswords,preg_quote($word,'#'));
369                break;
370        }
371    }
372
373    // a search without any posword is useless
374    if (!count($poswords)) return true;
375
376    $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
377            $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
378            search_regex($data,$base,$file,$reg,$poswords);
379            return true;
380            }
381
382            /**
383             * Reference search
384             * This fuction searches for existing references to a given media file
385             * and returns an array with the found pages. It doesn't pay any
386             * attention to ACL permissions to find every reference. The caller
387             * must check if the user has the appropriate rights to see the found
388             * page and eventually have to prevent the result from displaying.
389             *
390             * @param array  $data Reference to the result data structure
391             * @param string $base Base usually $conf['datadir']
392             * @param string $file current file or directory relative to $base
393             * @param char   $type Type either 'd' for directory or 'f' for file
394             * @param int    $lvl  Current recursion depht
395             * @param mixed  $opts option array as given to search()
396             *
397             * $opts['query'] is the demanded media file name
398             *
399             * @author  Andreas Gohr <andi@splitbrain.org>
400             * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
401             */
402function search_reference(&$data,$base,$file,$type,$lvl,$opts){
403    global $conf;
404
405    //we do nothing with directories
406    if($type == 'd') return true;
407
408    //only search txt files
409    if(substr($file,-4) != '.txt') return true;
410
411    //we finish after 'cnt' references found. The return value
412    //'false' will skip subdirectories to speed search up.
413    $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
414    if(count($data) >= $cnt) return false;
415
416    $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
417    search_regex($data,$base,$file,$reg,array($opts['query']));
418    return true;
419}
420
421/* ------------- helper functions below -------------- */
422
423/**
424 * fulltext search helper
425 * searches a text file with a given regular expression
426 * no ACL checks are performed. This have to be done by
427 * the caller if necessary.
428 *
429 * @param array  $data  reference to array for results
430 * @param string $base  base directory
431 * @param string $file  file name to search in
432 * @param string $reg   regular expression to search for
433 * @param array  $words words that should be marked in the results
434 *
435 * @author  Andreas Gohr <andi@splitbrain.org>
436 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
437 *
438 * @deprecated - fulltext indexer is used instead
439 */
440function search_regex(&$data,$base,$file,$reg,$words){
441
442    //get text
443    $text = io_readfile($base.'/'.$file);
444    //lowercase text (u modifier does not help with case)
445    $lctext = utf8_strtolower($text);
446
447    //do the fulltext search
448    $matches = array();
449    if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
450        //this is not the best way for snippet generation but the fastest I could find
451        $q = $words[0];  //use first word for snippet creation
452        $p = utf8_strpos($lctext,$q);
453        $f = $p - 100;
454        $l = utf8_strlen($q) + 200;
455        if($f < 0) $f = 0;
456        $snippet = '<span class="search_sep"> ... </span>'.
457            htmlspecialchars(utf8_substr($text,$f,$l)).
458            '<span class="search_sep"> ... </span>';
459        $mark    = '('.join('|', $words).')';
460        $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
461
462        $data[] = array(
463                'id'       => pathID($file),
464                'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
465                'poswords' => join(' ',$words),
466                'snippet'  => $snippet,
467                );
468    }
469
470    return true;
471}
472
473
474/**
475 * fulltext sort
476 *
477 * Callback sort function for use with usort to sort the data
478 * structure created by search_fulltext. Sorts descending by count
479 *
480 * @author  Andreas Gohr <andi@splitbrain.org>
481 */
482function sort_search_fulltext($a,$b){
483    if($a['count'] > $b['count']){
484        return -1;
485    }elseif($a['count'] < $b['count']){
486        return 1;
487    }else{
488        return strcmp($a['id'],$b['id']);
489    }
490}
491
492/**
493 * translates a document path to an ID
494 *
495 * @author  Andreas Gohr <andi@splitbrain.org>
496 * @todo    move to pageutils
497 */
498function pathID($path,$keeptxt=false){
499    $id = utf8_decodeFN($path);
500    $id = str_replace('/',':',$id);
501    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
502    $id = trim($id, ':');
503    return $id;
504}
505
506
507/**
508 * This is a very universal callback for the search() function, replacing
509 * many of the former individual functions at the cost of a more complex
510 * setup.
511 *
512 * How the function behaves, depends on the options passed in the $opts
513 * array, where the following settings can be used.
514 *
515 * depth      int     recursion depth. 0 for unlimited
516 * keeptxt    bool    keep .txt extension for IDs
517 * listfiles  bool    include files in listing
518 * listdirs   bool    include namespaces in listing
519 * pagesonly  bool    restrict files to pages
520 * skipacl    bool    do not check for READ permission
521 * sneakyacl  bool    don't recurse into nonreadable dirs
522 * hash       bool    create MD5 hash for files
523 * meta       bool    return file metadata
524 * filematch  string  match files against this regexp
525 * idmatch    string  match full ID against this regexp
526 * dirmatch   string  match directory against this regexp when adding
527 * nsmatch    string  match namespace against this regexp when adding
528 * recmatch   string  match directory against this regexp when recursing
529 * showmsg    bool    warn about non-ID files
530 * showhidden bool    show hidden files too
531 * firsthead  bool    return first heading for pages
532 *
533 * @author Andreas Gohr <gohr@cosmocode.de>
534 */
535function search_universal(&$data,$base,$file,$type,$lvl,$opts){
536    $item   = array();
537    $return = true;
538
539    // get ID and check if it is a valid one
540    $item['id'] = pathID($file,($type == 'd' || $opts['keeptxt']));
541    if($item['id'] != cleanID($item['id'])){
542        if($opts['showmsg'])
543            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
544        return false; // skip non-valid files
545    }
546    $item['ns']  = getNS($item['id']);
547
548    if($type == 'd') {
549        // decide if to recursion into this directory is wanted
550        if(!$opts['depth']){
551            $return = true; // recurse forever
552        }else{
553            $depth = substr_count($file,'/');
554            if($depth >= $opts['depth']){
555                $return = false; // depth reached
556            }else{
557                $return = true;
558            }
559        }
560        if($return && !preg_match('/'.$opts['recmatch'].'/',$file)){
561            $return = false; // doesn't match
562        }
563    }
564
565    // check ACL
566    if(!$opts['skipacl']){
567        if($type == 'd'){
568            $item['perm'] = auth_quickaclcheck($item['id'].':*');
569        }else{
570            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
571        }
572    }else{
573        $item['perm'] = AUTH_DELETE;
574    }
575
576    // are we done here maybe?
577    if($type == 'd'){
578        if(!$opts['listdirs']) return $return;
579        if(!$opts['skipacl'] && $opts['sneakyacl'] && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
580        if($opts['dirmatch'] && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
581        if($opts['nsmatch'] && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
582    }else{
583        if(!$opts['listfiles']) return $return;
584        if(!$opts['skipacl'] && $item['perm'] < AUTH_READ) return $return;
585        if($opts['pagesonly'] && (substr($file,-4) != '.txt')) return $return;
586        if(!$opts['showhidden'] && isHiddenPage($item['id'])) return $return;
587        if($opts['filematch'] && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
588        if($opts['idmatch'] && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
589    }
590
591    // still here? prepare the item
592    $item['type']  = $type;
593    $item['level'] = $lvl;
594    $item['open']  = $return;
595
596    if($opts['meta']){
597        $item['file']       = utf8_basename($file);
598        $item['size']       = filesize($base.'/'.$file);
599        $item['mtime']      = filemtime($base.'/'.$file);
600        $item['rev']        = $item['mtime'];
601        $item['writable']   = is_writable($base.'/'.$file);
602        $item['executable'] = is_executable($base.'/'.$file);
603    }
604
605    if($type == 'f'){
606        if($opts['hash']) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
607        if($opts['firsthead']) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
608    }
609
610    // finally add the item
611    $data[] = $item;
612    return $return;
613}
614
615//Setup VIM: ex: et ts=4 :
616