xref: /dokuwiki/inc/search.php (revision 965d96926bd3b834e7814174763a94560f8bc2f0) 
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' => $conf['sneaky_index'],
121        // Hacky, should rather use recmatch
122        'depth' => preg_match('#^'.$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    $info['file']     = basename($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,$opts){
250    //we do nothing with directories
251    if($type == 'd'){
252        if(!$opts['depth']) return true; // recurse forever
253        $parts = explode('/',ltrim($file,'/'));
254        if(count($parts) == $opts['depth']) return false; // depth reached
255        return true;
256    }
257
258    //only search txt files
259    if(substr($file,-4) != '.txt') return true;
260
261    $item['id']   = pathID($file);
262    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
263        return false;
264    }
265
266    $item['rev']   = filemtime($base.'/'.$file);
267    $item['mtime'] = $item['rev'];
268    $item['size']  = filesize($base.'/'.$file);
269    if($opts['hash']){
270        $item['hash'] = md5(trim(rawWiki($item['id'])));
271    }
272
273    $data[] = $item;
274    return true;
275}
276
277/**
278 * Search for backlinks to a given page
279 *
280 * $opts['ns']    namespace of the page
281 * $opts['name']  name of the page without namespace
282 *
283 * @author  Andreas Gohr <andi@splitbrain.org>
284 * @deprecated Replaced by ft_backlinks()
285 */
286function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
287    //we do nothing with directories
288    if($type == 'd') return true;
289    //only search txt files
290    if(substr($file,-4) != '.txt') return true;
291
292    //absolute search id
293    $sid = cleanID($opts['ns'].':'.$opts['name']);
294
295    //current id and namespace
296    $cid = pathID($file);
297    $cns = getNS($cid);
298
299    //check ACL
300    if(auth_quickaclcheck($cid) < AUTH_READ){
301        return false;
302    }
303
304    //fetch instructions
305    $instructions = p_cached_instructions($base.$file,true);
306    if(is_null($instructions)) return false;
307
308    global $conf;
309    //check all links for match
310    foreach($instructions as $ins){
311        if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
312            $mid = $ins[1][0];
313            resolve_pageid($cns,$mid,$exists); //exists is not used
314            if($mid == $sid){
315                //we have a match - finish
316                $data[]['id'] = $cid;
317                break;
318            }
319        }
320    }
321
322    return false;
323}
324
325/**
326 * Fulltextsearch
327 *
328 * $opts['query'] is the search query
329 *
330 * @author  Andreas Gohr <andi@splitbrain.org>
331 * @deprecated - fulltext indexer is used instead
332 */
333function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
334    //we do nothing with directories
335    if($type == 'd') return true;
336    //only search txt files
337    if(substr($file,-4) != '.txt') return true;
338
339    //check ACL
340    $id = pathID($file);
341    if(auth_quickaclcheck($id) < AUTH_READ){
342        return false;
343    }
344
345    //create regexp from queries
346    $poswords = array();
347    $negwords = array();
348    $qpreg = preg_split('/\s+/',$opts['query']);
349
350    foreach($qpreg as $word){
351        switch(substr($word,0,1)){
352            case '-':
353                if(strlen($word) > 1){  // catch single '-'
354                    array_push($negwords,preg_quote(substr($word,1),'#'));
355                }
356                break;
357            case '+':
358                if(strlen($word) > 1){  // catch single '+'
359                    array_push($poswords,preg_quote(substr($word,1),'#'));
360                }
361                break;
362            default:
363                array_push($poswords,preg_quote($word,'#'));
364                break;
365        }
366    }
367
368    // a search without any posword is useless
369    if (!count($poswords)) return true;
370
371    $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
372            $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
373            search_regex($data,$base,$file,$reg,$poswords);
374            return true;
375            }
376
377            /**
378             * Reference search
379             * This fuction searches for existing references to a given media file
380             * and returns an array with the found pages. It doesn't pay any
381             * attention to ACL permissions to find every reference. The caller
382             * must check if the user has the appropriate rights to see the found
383             * page and eventually have to prevent the result from displaying.
384             *
385             * @param array  $data Reference to the result data structure
386             * @param string $base Base usually $conf['datadir']
387             * @param string $file current file or directory relative to $base
388             * @param char   $type Type either 'd' for directory or 'f' for file
389             * @param int    $lvl  Current recursion depht
390             * @param mixed  $opts option array as given to search()
391             *
392             * $opts['query'] is the demanded media file name
393             *
394             * @author  Andreas Gohr <andi@splitbrain.org>
395             * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
396             */
397function search_reference(&$data,$base,$file,$type,$lvl,$opts){
398    global $conf;
399
400    //we do nothing with directories
401    if($type == 'd') return true;
402
403    //only search txt files
404    if(substr($file,-4) != '.txt') return true;
405
406    //we finish after 'cnt' references found. The return value
407    //'false' will skip subdirectories to speed search up.
408    $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
409    if(count($data) >= $cnt) return false;
410
411    $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
412    search_regex($data,$base,$file,$reg,array($opts['query']));
413    return true;
414}
415
416/* ------------- helper functions below -------------- */
417
418/**
419 * fulltext search helper
420 * searches a text file with a given regular expression
421 * no ACL checks are performed. This have to be done by
422 * the caller if necessary.
423 *
424 * @param array  $data  reference to array for results
425 * @param string $base  base directory
426 * @param string $file  file name to search in
427 * @param string $reg   regular expression to search for
428 * @param array  $words words that should be marked in the results
429 *
430 * @author  Andreas Gohr <andi@splitbrain.org>
431 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
432 *
433 * @deprecated - fulltext indexer is used instead
434 */
435function search_regex(&$data,$base,$file,$reg,$words){
436
437    //get text
438    $text = io_readfile($base.'/'.$file);
439    //lowercase text (u modifier does not help with case)
440    $lctext = utf8_strtolower($text);
441
442    //do the fulltext search
443    $matches = array();
444    if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
445        //this is not the best way for snippet generation but the fastest I could find
446        $q = $words[0];  //use first word for snippet creation
447        $p = utf8_strpos($lctext,$q);
448        $f = $p - 100;
449        $l = utf8_strlen($q) + 200;
450        if($f < 0) $f = 0;
451        $snippet = '<span class="search_sep"> ... </span>'.
452            htmlspecialchars(utf8_substr($text,$f,$l)).
453            '<span class="search_sep"> ... </span>';
454        $mark    = '('.join('|', $words).')';
455        $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
456
457        $data[] = array(
458                'id'       => pathID($file),
459                'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
460                'poswords' => join(' ',$words),
461                'snippet'  => $snippet,
462                );
463    }
464
465    return true;
466}
467
468
469/**
470 * fulltext sort
471 *
472 * Callback sort function for use with usort to sort the data
473 * structure created by search_fulltext. Sorts descending by count
474 *
475 * @author  Andreas Gohr <andi@splitbrain.org>
476 */
477function sort_search_fulltext($a,$b){
478    if($a['count'] > $b['count']){
479        return -1;
480    }elseif($a['count'] < $b['count']){
481        return 1;
482    }else{
483        return strcmp($a['id'],$b['id']);
484    }
485}
486
487/**
488 * translates a document path to an ID
489 *
490 * @author  Andreas Gohr <andi@splitbrain.org>
491 * @todo    move to pageutils
492 */
493function pathID($path,$keeptxt=false){
494    $id = utf8_decodeFN($path);
495    $id = str_replace('/',':',$id);
496    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
497    $id = trim($id, ':');
498    return $id;
499}
500
501
502/**
503 * This is a very universal callback for the search() function, replacing
504 * many of the former individual functions at the cost of a more complex
505 * setup.
506 *
507 * How the function behaves, depends on the options passed in the $opts
508 * array, where the following settings can be used.
509 *
510 * depth      int     recursion depth. 0 for unlimited
511 * keeptxt    bool    keep .txt extension for IDs
512 * listfiles  bool    include files in listing
513 * listdirs   bool    include namespaces in listing
514 * pagesonly  bool    restrict files to pages
515 * skipacl    bool    do not check for READ permission
516 * sneakyacl  bool    don't recurse into nonreadable dirs
517 * hash       bool    create MD5 hash for files
518 * meta       bool    return file metadata
519 * filematch  string  match files against this regexp
520 * idmatch    string  match full ID against this regexp
521 * dirmatch   string  match directory against this regexp when adding
522 * nsmatch    string  match namespace against this regexp when adding
523 * recmatch   string  match directory against this regexp when recursing
524 * showmsg    bool    warn about non-ID files
525 * showhidden bool    show hidden files too
526 * firsthead  bool    return first heading for pages
527 *
528 * @author Andreas Gohr <gohr@cosmocode.de>
529 */
530function search_universal(&$data,$base,$file,$type,$lvl,$opts){
531    $item   = array();
532    $return = true;
533
534    // get ID and check if it is a valid one
535    $item['id'] = pathID($file,($type == 'd' || $opts['keeptxt']));
536    if($item['id'] != cleanID($item['id'])){
537        if($opts['showmsg'])
538            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
539        return false; // skip non-valid files
540    }
541    $item['ns']  = getNS($item['id']);
542
543    if($type == 'd') {
544        // decide if to recursion into this directory is wanted
545        if(!$opts['depth']){
546            $return = true; // recurse forever
547        }else{
548            $depth = substr_count($file,'/');
549            if($depth >= $opts['depth']){
550                $return = false; // depth reached
551            }else{
552                $return = true;
553            }
554        }
555        if($return && !preg_match('/'.$opts['recmatch'].'/',$file)){
556            $return = false; // doesn't match
557        }
558    }
559
560    // check ACL
561    if(!$opts['skipacl']){
562        if($type == 'd'){
563            $item['perm'] = auth_quickaclcheck($item['id'].':*');
564        }else{
565            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
566        }
567    }else{
568        $item['perm'] = AUTH_DELETE;
569    }
570
571    // are we done here maybe?
572    if($type == 'd'){
573        if(!$opts['listdirs']) return $return;
574        if(!$opts['skipacl'] && $opts['sneakyacl'] && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
575        if($opts['dirmatch'] && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
576        if($opts['nsmatch'] && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
577    }else{
578        if(!$opts['listfiles']) return $return;
579        if(!$opts['skipacl'] && $item['perm'] < AUTH_READ) return $return;
580        if($opts['pagesonly'] && (substr($file,-4) != '.txt')) return $return;
581        if(!$opts['showhidden'] && isHiddenPage($item['id'])) return $return;
582        if($opts['filematch'] && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
583        if($opts['idmatch'] && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
584    }
585
586    // still here? prepare the item
587    $item['type']  = $type;
588    $item['level'] = $lvl;
589    $item['open']  = $return;
590
591    if($opts['meta']){
592        $item['file']       = basename($file);
593        $item['size']       = filesize($base.'/'.$file);
594        $item['mtime']      = filemtime($base.'/'.$file);
595        $item['rev']        = $item['mtime'];
596        $item['writable']   = is_writable($base.'/'.$file);
597        $item['executable'] = is_executable($base.'/'.$file);
598    }
599
600    if($type == 'f'){
601        if($opts['hash']) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
602        if($opts['firsthead']) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
603    }
604
605    // finally add the item
606    $data[] = $item;
607    return $return;
608}
609
610//Setup VIM: ex: et ts=4 :
611