xref: /dokuwiki/inc/search.php (revision 7fcedc39d164da3013a88e9ad6660009940e851b) 
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
9use dokuwiki\Utf8\Sort;
10
11/**
12 * Recurse directory
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    &$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   array     $opts option array will be given to the Callback
21 * @param   string    $dir  Current directory beyond $base
22 * @param   int       $lvl  Recursion Level
23 * @param   mixed     $sort 'natural' to use natural order sorting (default);
24 *                          'date' to sort by filemtime; leave empty to skip sorting.
25 * @author  Andreas Gohr <andi@splitbrain.org>
26 */
27function search(&$data, $base, $func, $opts, $dir='', $lvl=1, $sort='natural')
28{
29    $dirs   = array();
30    $files  = array();
31    $filepaths = array();
32
33    // safeguard against runaways #1452
34    if ($base == '' || $base == '/') {
35        throw new RuntimeException('No valid $base passed to search() - possible misconfiguration or bug');
36    }
37
38    //read in directories and files
39    $dh = @opendir($base.'/'.$dir);
40    if (!$dh) return;
41    while (($file = readdir($dh)) !== false) {
42        if (preg_match('/^[\._]/', $file)) continue; //skip hidden files and upper dirs
43        if (is_dir($base.'/'.$dir.'/'.$file)) {
44            $dirs[] = $dir.'/'.$file;
45            continue;
46        }
47        $files[] = $dir.'/'.$file;
48        $filepaths[] = $base.'/'.$dir.'/'.$file;
49    }
50    closedir($dh);
51    if (!empty($sort)) {
52        if ($sort == 'date') {
53            @array_multisort(array_map('filemtime', $filepaths), SORT_NUMERIC, SORT_DESC, $files);
54        } else /* natural */ {
55            Sort::asortFN($files);
56        }
57        Sort::asortFN($dirs);
58    }
59
60    //give directories to userfunction then recurse
61    foreach ($dirs as $dir) {
62        if (call_user_func_array($func, array(&$data, $base, $dir, 'd', $lvl,$opts))) {
63            search($data, $base, $func, $opts, $dir, $lvl+1, $sort);
64        }
65    }
66    //now handle the files
67    foreach ($files as $file) {
68        call_user_func_array($func, array(&$data, $base, $file, 'f', $lvl, $opts));
69    }
70}
71
72/**
73 * The following functions are userfunctions to use with the search
74 * function above. This function is called for every found file or
75 * directory. When a directory is given to the function it has to
76 * decide if this directory should be traversed (true) or not (false)
77 * The function has to accept the following parameters:
78 *
79 * array &$data  - Reference to the result data structure
80 * string $base  - Base usually $conf['datadir']
81 * string $file  - current file or directory relative to $base
82 * string $type  - Type either 'd' for directory or 'f' for file
83 * int    $lvl   - Current recursion depht
84 * array  $opts  - option array as given to search()
85 *
86 * return values for files are ignored
87 *
88 * All functions should check the ACL for document READ rights
89 * namespaces (directories) are NOT checked (when sneaky_index is 0) as this
90 * would break the recursion (You can have an nonreadable dir over a readable
91 * one deeper nested) also make sure to check the file type (for example
92 * in case of lockfiles).
93 */
94
95/**
96 * Searches for pages beginning with the given query
97 *
98 * @author Andreas Gohr <andi@splitbrain.org>
99 *
100 * @param array $data
101 * @param string $base
102 * @param string $file
103 * @param string $type
104 * @param integer $lvl
105 * @param array $opts
106 *
107 * @return bool
108 */
109function search_qsearch(&$data, $base, $file, $type, $lvl, $opts)
110{
111    $opts = array(
112            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
113            'listfiles' => true,
114            'pagesonly' => true,
115            );
116    return search_universal($data, $base, $file, $type, $lvl, $opts);
117}
118
119/**
120 * Build the browsable index of pages
121 *
122 * $opts['ns'] is the currently viewed namespace
123 *
124 * @author  Andreas Gohr <andi@splitbrain.org>
125 *
126 * @param array $data
127 * @param string $base
128 * @param string $file
129 * @param string $type
130 * @param integer $lvl
131 * @param array $opts
132 *
133 * @return bool
134 */
135function search_index(&$data, $base, $file, $type, $lvl, $opts)
136{
137    global $conf;
138    $ns = isset($opts['ns']) ? $opts['ns'] : '';
139    $opts = array(
140        'pagesonly' => true,
141        'listdirs' => true,
142        'listfiles' => empty($opts['nofiles']),
143        'sneakyacl' => $conf['sneaky_index'],
144        // Hacky, should rather use recmatch
145        'depth' => preg_match('#^'.preg_quote($file, '#').'(/|$)#','/'.$ns) ? 0 : -1
146    );
147
148    return search_universal($data, $base, $file, $type, $lvl, $opts);
149}
150
151/**
152 * List all namespaces
153 *
154 * @author  Andreas Gohr <andi@splitbrain.org>
155 *
156 * @param array $data
157 * @param string $base
158 * @param string $file
159 * @param string $type
160 * @param integer $lvl
161 * @param array $opts
162 *
163 * @return bool
164 */
165function search_namespaces(&$data, $base, $file, $type, $lvl, $opts)
166{
167    $opts = array(
168            'listdirs' => true,
169            );
170    return search_universal($data, $base, $file, $type, $lvl, $opts);
171}
172
173/**
174 * List all mediafiles in a namespace
175 *   $opts['depth']     recursion level, 0 for all
176 *   $opts['showmsg']   shows message if invalid media id is used
177 *   $opts['skipacl']   skip acl checking
178 *   $opts['pattern']   check given pattern
179 *   $opts['hash']      add hashes to result list
180 *
181 * @author  Andreas Gohr <andi@splitbrain.org>
182 *
183 * @param array $data
184 * @param string $base
185 * @param string $file
186 * @param string $type
187 * @param integer $lvl
188 * @param array $opts
189 *
190 * @return bool
191 */
192function search_media(&$data, $base, $file, $type, $lvl, $opts)
193{
194    //we do nothing with directories
195    if ($type == 'd') {
196        if (empty($opts['depth'])) return true; // recurse forever
197        $depth = substr_count($file,'/');
198        if ($depth >= $opts['depth']) return false; // depth reached
199        return true;
200    }
201
202    $info         = array();
203    $info['id']   = pathID($file, true);
204    if ($info['id'] != cleanID($info['id'])) {
205        if ($opts['showmsg'])
206            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
207        return false; // skip non-valid files
208    }
209
210    //check ACL for namespace (we have no ACL for mediafiles)
211    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
212    if (empty($opts['skipacl']) && $info['perm'] < AUTH_READ) {
213        return false;
214    }
215
216    //check pattern filter
217    if (!empty($opts['pattern']) && !@preg_match($opts['pattern'], $info['id'])) {
218        return false;
219    }
220
221    $info['file']     = \dokuwiki\Utf8\PhpString::basename($file);
222    $info['size']     = filesize($base.'/'.$file);
223    $info['mtime']    = filemtime($base.'/'.$file);
224    $info['writable'] = is_writable($base.'/'.$file);
225    if (preg_match("/\.(jpe?g|gif|png)$/", $file)) {
226        $info['isimg'] = true;
227        $info['meta']  = new JpegMeta($base.'/'.$file);
228    } else {
229        $info['isimg'] = false;
230    }
231    if (!empty($opts['hash'])) {
232        $info['hash'] = md5(io_readFile(mediaFN($info['id']), false));
233    }
234
235    $data[] = $info;
236
237    return false;
238}
239
240/**
241 * List all mediafiles in a namespace
242 *   $opts['depth']     recursion level, 0 for all
243 *   $opts['showmsg']   shows message if invalid media id is used
244 *   $opts['skipacl']   skip acl checking
245 *   $opts['pattern']   check given pattern
246 *   $opts['hash']      add hashes to result list
247 *
248 * @todo This is a temporary copy of search_media returning a list of MediaFile intances
249 *
250 * @param array $data
251 * @param string $base
252 * @param string $file
253 * @param string $type
254 * @param integer $lvl
255 * @param array $opts
256 *
257 * @return bool
258 */
259function search_mediafiles(&$data,$base,$file,$type,$lvl,$opts){
260
261    //we do nothing with directories
262    if($type == 'd') {
263        if(empty($opts['depth'])) return true; // recurse forever
264        $depth = substr_count($file,'/');
265        if($depth >= $opts['depth']) return false; // depth reached
266        return true;
267    }
268
269    $id   = pathID($file,true);
270    if($id != cleanID($id)){
271        if($opts['showmsg'])
272            msg(hsc($id).' is not a valid file name for DokuWiki - skipped',-1);
273        return false; // skip non-valid files
274    }
275
276    //check ACL for namespace (we have no ACL for mediafiles)
277    $info['perm'] = auth_quickaclcheck(getNS($id).':*');
278    if(empty($opts['skipacl']) && $info['perm'] < AUTH_READ){
279        return false;
280    }
281
282    //check pattern filter
283    if(!empty($opts['pattern']) && !@preg_match($opts['pattern'], $id)){
284        return false;
285    }
286
287    $data[] = new \dokuwiki\File\MediaFile($id);
288    return false;
289}
290
291
292/**
293 * This function just lists documents (for RSS namespace export)
294 *
295 * @author  Andreas Gohr <andi@splitbrain.org>
296 *
297 * @param array $data
298 * @param string $base
299 * @param string $file
300 * @param string $type
301 * @param integer $lvl
302 * @param array $opts
303 *
304 * @return bool
305 */
306function search_list(&$data, $base, $file, $type, $lvl, $opts)
307{
308    //we do nothing with directories
309    if ($type == 'd') return false;
310    //only search txt files
311    if (substr($file, -4) == '.txt') {
312        //check ACL
313        $id = pathID($file);
314        if (auth_quickaclcheck($id) < AUTH_READ) {
315            return false;
316        }
317        $data[]['id'] = $id;
318    }
319    return false;
320}
321
322/**
323 * Quicksearch for searching matching pagenames
324 *
325 * $opts['query'] is the search query
326 *
327 * @author  Andreas Gohr <andi@splitbrain.org>
328 *
329 * @param array $data
330 * @param string $base
331 * @param string $file
332 * @param string $type
333 * @param integer $lvl
334 * @param array $opts
335 *
336 * @return bool
337 */
338function search_pagename(&$data, $base, $file, $type, $lvl, $opts)
339{
340    //we do nothing with directories
341    if ($type == 'd') return true;
342    //only search txt files
343    if (substr($file, -4) != '.txt') return true;
344
345    //simple stringmatching
346    if (!empty($opts['query'])) {
347        if (strpos($file, $opts['query']) !== false) {
348            //check ACL
349            $id = pathID($file);
350            if (auth_quickaclcheck($id) < AUTH_READ) {
351                return false;
352            }
353            $data[]['id'] = $id;
354        }
355    }
356    return true;
357}
358
359/**
360 * Just lists all documents
361 *
362 * $opts['depth']   recursion level, 0 for all
363 * $opts['hash']    do md5 sum of content?
364 * $opts['skipacl'] list everything regardless of ACL
365 *
366 * @author  Andreas Gohr <andi@splitbrain.org>
367 *
368 * @param array $data
369 * @param string $base
370 * @param string $file
371 * @param string $type
372 * @param integer $lvl
373 * @param array $opts
374 *
375 * @return bool
376 */
377function search_allpages(&$data, $base, $file, $type, $lvl, $opts)
378{
379    if (isset($opts['depth']) && $opts['depth']) {
380        $parts = explode('/',ltrim($file, '/'));
381        if (($type == 'd' && count($parts) >= $opts['depth'])
382            || ($type != 'd' && count($parts) > $opts['depth'])
383        ){
384            return false; // depth reached
385        }
386    }
387
388    //we do nothing with directories
389    if ($type == 'd') {
390        return true;
391    }
392
393    //only search txt files
394    if (substr($file, -4) != '.txt') return true;
395
396    $item = array();
397    $item['id']   = pathID($file);
398    if (empty($opts['skipacl']) && auth_quickaclcheck($item['id']) < AUTH_READ) {
399        return false;
400    }
401
402    $item['rev']   = filemtime($base.'/'.$file);
403    $item['mtime'] = $item['rev'];
404    $item['size']  = filesize($base.'/'.$file);
405    if (!empty($opts['hash'])) {
406        $item['hash'] = md5(trim(rawWiki($item['id'])));
407    }
408
409    $data[] = $item;
410    return true;
411}
412
413/* ------------- helper functions below -------------- */
414
415/**
416 * fulltext sort
417 *
418 * Callback sort function for use with usort to sort the data
419 * structure created by search_fulltext. Sorts descending by count
420 *
421 * @author  Andreas Gohr <andi@splitbrain.org>
422 *
423 * @param array $a
424 * @param array $b
425 *
426 * @return int
427 */
428function sort_search_fulltext($a, $b)
429{
430    if ($a['count'] > $b['count']) {
431        return -1;
432    } elseif ($a['count'] < $b['count']) {
433        return 1;
434    }else{
435        return Sort::strcmp($a['id'],$b['id']);
436    }
437}
438
439/**
440 * translates a document path to an ID
441 *
442 * @author  Andreas Gohr <andi@splitbrain.org>
443 * @todo    move to pageutils
444 *
445 * @param string $path
446 * @param bool $keeptxt
447 *
448 * @return mixed|string
449 */
450function pathID($path, $keeptxt=false)
451{
452    $id = utf8_decodeFN($path);
453    $id = str_replace('/', ':', $id);
454    if (!$keeptxt) $id = preg_replace('#\.txt$#', '', $id);
455    $id = trim($id, ':');
456    return $id;
457}
458
459
460/**
461 * This is a very universal callback for the search() function, replacing
462 * many of the former individual functions at the cost of a more complex
463 * setup.
464 *
465 * How the function behaves, depends on the options passed in the $opts
466 * array, where the following settings can be used.
467 *
468 * depth      int     recursion depth. 0 for unlimited                       (default: 0)
469 * keeptxt    bool    keep .txt extension for IDs                            (default: false)
470 * listfiles  bool    include files in listing                               (default: false)
471 * listdirs   bool    include namespaces in listing                          (default: false)
472 * pagesonly  bool    restrict files to pages                                (default: false)
473 * skipacl    bool    do not check for READ permission                       (default: false)
474 * sneakyacl  bool    don't recurse into nonreadable dirs                    (default: false)
475 * hash       bool    create MD5 hash for files                              (default: false)
476 * meta       bool    return file metadata                                   (default: false)
477 * filematch  string  match files against this regexp                        (default: '', so accept everything)
478 * idmatch    string  match full ID against this regexp                      (default: '', so accept everything)
479 * dirmatch   string  match directory against this regexp when adding        (default: '', so accept everything)
480 * nsmatch    string  match namespace against this regexp when adding        (default: '', so accept everything)
481 * recmatch   string  match directory against this regexp when recursing     (default: '', so accept everything)
482 * showmsg    bool    warn about non-ID files                                (default: false)
483 * showhidden bool    show hidden files(e.g. by hidepages config) too        (default: false)
484 * firsthead  bool    return first heading for pages                         (default: false)
485 *
486 * @param array &$data  - Reference to the result data structure
487 * @param string $base  - Base usually $conf['datadir']
488 * @param string $file  - current file or directory relative to $base
489 * @param string $type  - Type either 'd' for directory or 'f' for file
490 * @param int    $lvl   - Current recursion depht
491 * @param array  $opts  - option array as given to search()
492 * @return bool if this directory should be traversed (true) or not (false)
493 *              return value is ignored for files
494 *
495 * @author Andreas Gohr <gohr@cosmocode.de>
496 */
497function search_universal(&$data, $base, $file, $type, $lvl, $opts)
498{
499    $item   = array();
500    $return = true;
501
502    // get ID and check if it is a valid one
503    $item['id'] = pathID($file, ($type == 'd' || !empty($opts['keeptxt'])));
504    if ($item['id'] != cleanID($item['id'])){
505        if (!empty($opts['showmsg'])) {
506            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
507        }
508        return false; // skip non-valid files
509    }
510    $item['ns']  = getNS($item['id']);
511
512    if ($type == 'd') {
513        // decide if to recursion into this directory is wanted
514        if (empty($opts['depth'])) {
515            $return = true; // recurse forever
516        } else {
517            $depth = substr_count($file,'/');
518            if ($depth >= $opts['depth']) {
519                $return = false; // depth reached
520            } else {
521                $return = true;
522            }
523        }
524
525        if ($return) {
526            $match = empty($opts['recmatch']) || preg_match('/'.$opts['recmatch'].'/', $file);
527            if (!$match) {
528                return false; // doesn't match
529            }
530        }
531    }
532
533    // check ACL
534    if (empty($opts['skipacl'])) {
535        if ($type == 'd') {
536            $item['perm'] = auth_quickaclcheck($item['id'].':*');
537        } else {
538            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
539        }
540    } else {
541        $item['perm'] = AUTH_DELETE;
542    }
543
544    // are we done here maybe?
545    if ($type == 'd') {
546        if (empty($opts['listdirs'])) return $return;
547        //neither list nor recurse forbidden items:
548        if (empty($opts['skipacl']) && !empty($opts['sneakyacl']) && $item['perm'] < AUTH_READ) return false;
549        if (!empty($opts['dirmatch']) && !preg_match('/'.$opts['dirmatch'].'/', $file)) return $return;
550        if (!empty($opts['nsmatch']) && !preg_match('/'.$opts['nsmatch'].'/', $item['ns'])) return $return;
551    } else {
552        if (empty($opts['listfiles'])) return $return;
553        if (empty($opts['skipacl']) && $item['perm'] < AUTH_READ) return $return;
554        if (!empty($opts['pagesonly']) && (substr($file, -4) != '.txt')) return $return;
555        if (empty($opts['showhidden']) && isHiddenPage($item['id'])) return $return;
556        if (!empty($opts['filematch']) && !preg_match('/'.$opts['filematch'].'/', $file)) return $return;
557        if (!empty($opts['idmatch']) && !preg_match('/'.$opts['idmatch'].'/', $item['id'])) return $return;
558    }
559
560    // still here? prepare the item
561    $item['type']  = $type;
562    $item['level'] = $lvl;
563    $item['open']  = $return;
564
565    if (!empty($opts['meta'])) {
566        $item['file']       = \dokuwiki\Utf8\PhpString::basename($file);
567        $item['size']       = filesize($base.'/'.$file);
568        $item['mtime']      = filemtime($base.'/'.$file);
569        $item['rev']        = $item['mtime'];
570        $item['writable']   = is_writable($base.'/'.$file);
571        $item['executable'] = is_executable($base.'/'.$file);
572    }
573
574    if ($type == 'f') {
575        if (!empty($opts['hash'])) $item['hash'] = md5(io_readFile($base.'/'.$file, false));
576        if (!empty($opts['firsthead'])) {
577            $item['title'] = p_get_first_heading($item['id'], METADATA_DONT_RENDER);
578        }
579    }
580
581    // finally add the item
582    $data[] = $item;
583    return $return;
584}
585
586//Setup VIM: ex: et ts=4 :
587