xref: /dokuwiki/inc/search.php (revision c04912f6c4e1c1c8bda182a7dc03c86bcb415d90) 
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.');
10require_once(DOKU_INC.'inc/common.php');
11
12/**
13 * recurse direcory
14 *
15 * This function recurses into a given base directory
16 * and calls the supplied function for each file and directory
17 *
18 * @param   array ref $data The results of the search are stored here
19 * @param   string    $base Where to start the search
20 * @param   callback  $func Callback (function name or arayy with object,method)
21 * @param   string    $dir  Current directory beyond $base
22 * @param   int       $lvl  Recursion Level
23 * @author  Andreas Gohr <andi@splitbrain.org>
24 */
25function search(&$data,$base,$func,$opts,$dir='',$lvl=1){
26    $dirs   = array();
27    $files  = 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    }
40    closedir($dh);
41    sort($files);
42    sort($dirs);
43
44    //give directories to userfunction then recurse
45    foreach($dirs as $dir){
46        if (call_user_func_array($func, array(&$data,$base,$dir,'d',$lvl,$opts))){
47            search($data,$base,$func,$opts,$dir,$lvl+1);
48        }
49    }
50    //now handle the files
51    foreach($files as $file){
52        call_user_func_array($func, array(&$data,$base,$file,'f',$lvl,$opts));
53    }
54}
55
56/**
57 * Wrapper around call_user_func_array.
58 *
59 * @deprecated
60 */
61function search_callback($func,&$data,$base,$file,$type,$lvl,$opts){
62    return call_user_func_array($func, array(&$data,$base,$file,$type,$lvl,$opts));
63}
64
65/**
66 * The following functions are userfunctions to use with the search
67 * function above. This function is called for every found file or
68 * directory. When a directory is given to the function it has to
69 * decide if this directory should be traversed (true) or not (false)
70 * The function has to accept the following parameters:
71 *
72 * &$data - Reference to the result data structure
73 * $base  - Base usually $conf['datadir']
74 * $file  - current file or directory relative to $base
75 * $type  - Type either 'd' for directory or 'f' for file
76 * $lvl   - Current recursion depht
77 * $opts  - option array as given to search()
78 *
79 * return values for files are ignored
80 *
81 * All functions should check the ACL for document READ rights
82 * namespaces (directories) are NOT checked as this would break
83 * the recursion (You can have an nonreadable dir over a readable
84 * one deeper nested) also make sure to check the file type (for example
85 * in case of lockfiles).
86 */
87
88/**
89 * Searches for pages beginning with the given query
90 *
91 * @author Andreas Gohr <andi@splitbrain.org>
92 */
93function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
94    $opts = array(
95            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
96            'listfiles' => true,
97            'pagesonly' => true,
98            );
99    return search_universal($data,$base,$file,$type,$lvl,$opts);
100}
101
102/**
103 * Build the browsable index of pages
104 *
105 * $opts['ns'] is the current namespace
106 *
107 * @author  Andreas Gohr <andi@splitbrain.org>
108 */
109function search_index(&$data,$base,$file,$type,$lvl,$opts){
110    global $conf;
111    $return = true;
112
113    $item = array();
114
115    if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
116        //add but don't recurse
117        $return = false;
118    }elseif($type == 'f' && ($opts['nofiles'] || substr($file,-4) != '.txt')){
119        //don't add
120        return false;
121    }
122
123    $id = pathID($file);
124
125    if($type=='d' && $conf['sneaky_index'] && auth_quickaclcheck($id.':') < AUTH_READ){
126        return false;
127    }
128
129    //check hidden
130    if(isHiddenPage($id)){
131        return false;
132    }
133
134    //check ACL
135    if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
136        return false;
137    }
138
139    $data[]=array( 'id'    => $id,
140            'type'  => $type,
141            'level' => $lvl,
142            'open'  => $return );
143    return $return;
144}
145
146/**
147 * List all namespaces
148 *
149 * @author  Andreas Gohr <andi@splitbrain.org>
150 */
151function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
152    $opts = array(
153            'listdirs' => true,
154            );
155    return search_universal($data,$base,$file,$type,$lvl,$opts);
156}
157
158/**
159 * List all mediafiles in a namespace
160 *
161 * @author  Andreas Gohr <andi@splitbrain.org>
162 */
163function search_media(&$data,$base,$file,$type,$lvl,$opts){
164
165    //we do nothing with directories
166    if($type == 'd') {
167        if(!$opts['depth']) return true; // recurse forever
168        $depth = substr_count($file,'/');
169        if($depth >= $opts['depth']) return false; // depth reached
170        return true;
171    }
172
173    $info         = array();
174    $info['id']   = pathID($file,true);
175    if($info['id'] != cleanID($info['id'])){
176        if($opts['showmsg'])
177            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
178        return false; // skip non-valid files
179    }
180
181    //check ACL for namespace (we have no ACL for mediafiles)
182    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
183    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
184        return false;
185    }
186
187    //check pattern filter
188    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
189        return false;
190    }
191
192    $info['file']     = basename($file);
193    $info['size']     = filesize($base.'/'.$file);
194    $info['mtime']    = filemtime($base.'/'.$file);
195    $info['writable'] = is_writable($base.'/'.$file);
196    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
197        $info['isimg'] = true;
198        require_once(DOKU_INC.'inc/JpegMeta.php');
199        $info['meta']  = new JpegMeta($base.'/'.$file);
200    }else{
201        $info['isimg'] = false;
202    }
203    if($opts['hash']){
204        $info['hash'] = md5(io_readFile(wikiFN($info['id']),false));
205    }
206
207    $data[] = $info;
208
209    return false;
210}
211
212/**
213 * This function just lists documents (for RSS namespace export)
214 *
215 * @author  Andreas Gohr <andi@splitbrain.org>
216 */
217function search_list(&$data,$base,$file,$type,$lvl,$opts){
218    //we do nothing with directories
219    if($type == 'd') return false;
220    //only search txt files
221    if(substr($file,-4) == '.txt'){
222        //check ACL
223        $id = pathID($file);
224        if(auth_quickaclcheck($id) < AUTH_READ){
225            return false;
226        }
227        $data[]['id'] = $id;
228    }
229    return false;
230}
231
232/**
233 * Quicksearch for searching matching pagenames
234 *
235 * $opts['query'] is the search query
236 *
237 * @author  Andreas Gohr <andi@splitbrain.org>
238 */
239function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
240    //we do nothing with directories
241    if($type == 'd') return true;
242    //only search txt files
243    if(substr($file,-4) != '.txt') return true;
244
245    //simple stringmatching
246    if (!empty($opts['query'])){
247        if(strpos($file,$opts['query']) !== false){
248            //check ACL
249            $id = pathID($file);
250            if(auth_quickaclcheck($id) < AUTH_READ){
251                return false;
252            }
253            $data[]['id'] = $id;
254        }
255    }
256    return true;
257}
258
259/**
260 * Just lists all documents
261 *
262 * $opts['depth']   recursion level, 0 for all
263 * $opts['hash']    do md5 sum of content?
264 * $opts['skipacl'] list everything regardless of ACL
265 *
266 * @author  Andreas Gohr <andi@splitbrain.org>
267 */
268function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
269    //we do nothing with directories
270    if($type == 'd'){
271        if(!$opts['depth']) return true; // recurse forever
272        $parts = explode('/',ltrim($file,'/'));
273        if(count($parts) == $opts['depth']) return false; // depth reached
274        return true;
275    }
276
277    //only search txt files
278    if(substr($file,-4) != '.txt') return true;
279
280    $item['id']   = pathID($file);
281    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
282        return false;
283    }
284
285    $item['rev']   = filemtime($base.'/'.$file);
286    $item['mtime'] = $item['rev'];
287    $item['size']  = filesize($base.'/'.$file);
288    if($opts['hash']){
289        $item['hash'] = md5(trim(rawWiki($item['id'])));
290    }
291
292    $data[] = $item;
293    return true;
294}
295
296/**
297 * Search for backlinks to a given page
298 *
299 * $opts['ns']    namespace of the page
300 * $opts['name']  name of the page without namespace
301 *
302 * @author  Andreas Gohr <andi@splitbrain.org>
303 * @deprecated Replaced by ft_backlinks()
304 */
305function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
306    //we do nothing with directories
307    if($type == 'd') return true;
308    //only search txt files
309    if(substr($file,-4) != '.txt') return true;
310
311    //absolute search id
312    $sid = cleanID($opts['ns'].':'.$opts['name']);
313
314    //current id and namespace
315    $cid = pathID($file);
316    $cns = getNS($cid);
317
318    //check ACL
319    if(auth_quickaclcheck($cid) < AUTH_READ){
320        return false;
321    }
322
323    //fetch instructions
324    require_once(DOKU_INC.'inc/parserutils.php');
325    $instructions = p_cached_instructions($base.$file,true);
326    if(is_null($instructions)) return false;
327
328    //check all links for match
329    foreach($instructions as $ins){
330        if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
331            $mid = $ins[1][0];
332            resolve_pageid($cns,$mid,$exists); //exists is not used
333            if($mid == $sid){
334                //we have a match - finish
335                $data[]['id'] = $cid;
336                break;
337            }
338        }
339    }
340
341    return false;
342}
343
344/**
345 * Fulltextsearch
346 *
347 * $opts['query'] is the search query
348 *
349 * @author  Andreas Gohr <andi@splitbrain.org>
350 * @deprecated - fulltext indexer is used instead
351 */
352function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
353    //we do nothing with directories
354    if($type == 'd') return true;
355    //only search txt files
356    if(substr($file,-4) != '.txt') return true;
357
358    //check ACL
359    $id = pathID($file);
360    if(auth_quickaclcheck($id) < AUTH_READ){
361        return false;
362    }
363
364    //create regexp from queries
365    $poswords = array();
366    $negwords = array();
367    $qpreg = preg_split('/\s+/',$opts['query']);
368
369    foreach($qpreg as $word){
370        switch(substr($word,0,1)){
371            case '-':
372                if(strlen($word) > 1){  // catch single '-'
373                    array_push($negwords,preg_quote(substr($word,1),'#'));
374                }
375                break;
376            case '+':
377                if(strlen($word) > 1){  // catch single '+'
378                    array_push($poswords,preg_quote(substr($word,1),'#'));
379                }
380                break;
381            default:
382                array_push($poswords,preg_quote($word,'#'));
383                break;
384        }
385    }
386
387    // a search without any posword is useless
388    if (!count($poswords)) return true;
389
390    $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
391            $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
392            search_regex($data,$base,$file,$reg,$poswords);
393            return true;
394            }
395
396            /**
397             * Reference search
398             * This fuction searches for existing references to a given media file
399             * and returns an array with the found pages. It doesn't pay any
400             * attention to ACL permissions to find every reference. The caller
401             * must check if the user has the appropriate rights to see the found
402             * page and eventually have to prevent the result from displaying.
403             *
404             * @param array  $data Reference to the result data structure
405             * @param string $base Base usually $conf['datadir']
406             * @param string $file current file or directory relative to $base
407             * @param char   $type Type either 'd' for directory or 'f' for file
408             * @param int    $lvl  Current recursion depht
409             * @param mixed  $opts option array as given to search()
410             *
411             * $opts['query'] is the demanded media file name
412             *
413             * @author  Andreas Gohr <andi@splitbrain.org>
414             * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
415             */
416function search_reference(&$data,$base,$file,$type,$lvl,$opts){
417    global $conf;
418
419    //we do nothing with directories
420    if($type == 'd') return true;
421
422    //only search txt files
423    if(substr($file,-4) != '.txt') return true;
424
425    //we finish after 'cnt' references found. The return value
426    //'false' will skip subdirectories to speed search up.
427    $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
428    if(count($data) >= $cnt) return false;
429
430    $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
431    search_regex($data,$base,$file,$reg,array($opts['query']));
432    return true;
433}
434
435/* ------------- helper functions below -------------- */
436
437/**
438 * fulltext search helper
439 * searches a text file with a given regular expression
440 * no ACL checks are performed. This have to be done by
441 * the caller if necessary.
442 *
443 * @param array  $data  reference to array for results
444 * @param string $base  base directory
445 * @param string $file  file name to search in
446 * @param string $reg   regular expression to search for
447 * @param array  $words words that should be marked in the results
448 *
449 * @author  Andreas Gohr <andi@splitbrain.org>
450 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
451 *
452 * @deprecated - fulltext indexer is used instead
453 */
454function search_regex(&$data,$base,$file,$reg,$words){
455
456    //get text
457    $text = io_readfile($base.'/'.$file);
458    //lowercase text (u modifier does not help with case)
459    $lctext = utf8_strtolower($text);
460
461    //do the fulltext search
462    $matches = array();
463    if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
464        //this is not the best way for snippet generation but the fastest I could find
465        $q = $words[0];  //use first word for snippet creation
466        $p = utf8_strpos($lctext,$q);
467        $f = $p - 100;
468        $l = utf8_strlen($q) + 200;
469        if($f < 0) $f = 0;
470        $snippet = '<span class="search_sep"> ... </span>'.
471            htmlspecialchars(utf8_substr($text,$f,$l)).
472            '<span class="search_sep"> ... </span>';
473        $mark    = '('.join('|', $words).')';
474        $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
475
476        $data[] = array(
477                'id'       => pathID($file),
478                'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
479                'poswords' => join(' ',$words),
480                'snippet'  => $snippet,
481                );
482    }
483
484    return true;
485}
486
487
488/**
489 * fulltext sort
490 *
491 * Callback sort function for use with usort to sort the data
492 * structure created by search_fulltext. Sorts descending by count
493 *
494 * @author  Andreas Gohr <andi@splitbrain.org>
495 */
496function sort_search_fulltext($a,$b){
497    if($a['count'] > $b['count']){
498        return -1;
499    }elseif($a['count'] < $b['count']){
500        return 1;
501    }else{
502        return strcmp($a['id'],$b['id']);
503    }
504}
505
506/**
507 * translates a document path to an ID
508 *
509 * @author  Andreas Gohr <andi@splitbrain.org>
510 * @todo    move to pageutils
511 */
512function pathID($path,$keeptxt=false){
513    $id = utf8_decodeFN($path);
514    $id = str_replace('/',':',$id);
515    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
516    $id = preg_replace('#^:+#','',$id);
517    $id = preg_replace('#:+$#','',$id);
518    return $id;
519}
520
521
522/**
523 * This is a very universal callback for the search() function, replacing
524 * many of the former individual functions at the cost of a more complex
525 * setup.
526 *
527 * How the function behaves, depends on the options passed in the $opts
528 * array, where the following settings can be used.
529 *
530 * depth      int     recursion depth. 0 for unlimited
531 * keeptxt    bool    keep .txt extension for IDs
532 * listfiles  bool    include files in listing
533 * listdirs   bool    include namespaces in listing
534 * pagesonly  bool    restrict files to pages
535 * skipacl    bool    do not check for READ permission
536 * sneakyacl  bool    don't recurse into nonreadable dirs
537 * hash       bool    create MD5 hash for files
538 * meta       bool    return file metadata
539 * filematch  string  match files against this regexp
540 * idmatch    string  match full ID against this regexp
541 * dirmatch   string  match directory against this regexp when adding
542 * nsmatch    string  match namespace against this regexp when adding
543 * recmatch   string  match directory against this regexp when recursing
544 * showmsg    bool    warn about non-ID files
545 * showhidden bool    show hidden files too
546 * firsthead  bool    return first heading for pages
547 *
548 * @author Andreas Gohr <gohr@cosmocode.de>
549 */
550function search_universal(&$data,$base,$file,$type,$lvl,$opts){
551    $item   = array();
552    $return = true;
553
554    // get ID and check if it is a valid one
555    $item['id'] = pathID($file);
556    if($item['id'] != cleanID($item['id'])){
557        if($opts['showmsg'])
558            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
559        return false; // skip non-valid files
560    }
561    $item['ns']  = getNS($item['id']);
562
563    if($type == 'd') {
564        // decide if to recursion into this directory is wanted
565        if(!$opts['depth']){
566            $return = true; // recurse forever
567        }else{
568            $depth = substr_count($file,'/');
569            if($depth >= $opts['depth']){
570                $return = false; // depth reached
571            }else{
572                $return = true;
573            }
574        }
575        if($return && !preg_match('/'.$opts['recmatch'].'/',$file)){
576            $return = false; // doesn't match
577        }
578    }
579
580    // check ACL
581    if(!$opts['skipacl']){
582        if($type == 'd'){
583            $item['perm'] = auth_quickaclcheck($item['id'].':*');
584        }else{
585            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
586        }
587    }else{
588        $item['perm'] = AUTH_DELETE;
589    }
590
591    // are we done here maybe?
592    if($type == 'd'){
593        if(!$opts['listdirs']) return $return;
594        if(!$opts['skipacl'] && $opts['sneakyacl'] && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
595        if($opts['dirmatch'] && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
596        if($opts['nsmatch'] && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
597    }else{
598        if(!$opts['listfiles']) return $return;
599        if(!$opts['skipacl'] && $item['perm'] < AUTH_READ) return $return;
600        if($opts['pagesonly'] && (substr($file,-4) != '.txt')) return $return;
601        if(!$conf['showhidden'] && isHiddenPage($id)) return $return;
602        if($opts['filematch'] && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
603        if($opts['idmatch'] && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
604    }
605
606    // still here? prepare the item
607    $item['type']  = $type;
608    $item['level'] = $lvl;
609    $item['open']  = $return;
610
611    if($opts['meta']){
612        $item['file']       = basename($file);
613        $item['size']       = filesize($base.'/'.$file);
614        $item['mtime']      = filemtime($base.'/'.$file);
615        $item['rev']        = $item['mtime'];
616        $item['writable']   = is_writable($base.'/'.$file);
617        $item['executable'] = is_executable($base.'/'.$file);
618    }
619
620    if($type == 'f'){
621        if($opts['hash']) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
622        if($opts['firsthead']) $item['title'] = p_get_first_heading($item['id'],false);
623    }
624
625    // finally add the item
626    $data[] = $item;
627    return $return;
628}
629
630//Setup VIM: ex: et ts=4 enc=utf-8 :
631