xref: /dokuwiki/inc/search.php (revision 76e0afa4d89152ba2dc95db3f4043b540b998be4)
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    $dirs   = array();
29    $files  = array();
30    $filepaths = array();
31
32    // safeguard against runaways #1452
33    if($base == '' || $base == '/') {
34        throw new RuntimeException('No valid $base passed to search() - possible misconfiguration or bug');
35    }
36
37    //read in directories and files
38    $dh = @opendir($base.'/'.$dir);
39    if(!$dh) return;
40    while(($file = readdir($dh)) !== false){
41        if(preg_match('/^[\._]/',$file)) continue; //skip hidden files and upper dirs
42        if(is_dir($base.'/'.$dir.'/'.$file)){
43            $dirs[] = $dir.'/'.$file;
44            continue;
45        }
46        $files[] = $dir.'/'.$file;
47        $filepaths[] = $base.'/'.$dir.'/'.$file;
48    }
49    closedir($dh);
50    if (!empty($sort)) {
51        if ($sort == 'date') {
52            @array_multisort(array_map('filemtime', $filepaths), SORT_NUMERIC, SORT_DESC, $files);
53        } else /* natural */ {
54            Sort::asortFN($files);
55        }
56        Sort::asortFN($dirs);
57    }
58
59    //give directories to userfunction then recurse
60    foreach($dirs as $dir){
61        if (call_user_func_array($func, array(&$data,$base,$dir,'d',$lvl,$opts))){
62            search($data,$base,$func,$opts,$dir,$lvl+1,$sort);
63        }
64    }
65    //now handle the files
66    foreach($files as $file){
67        call_user_func_array($func, array(&$data,$base,$file,'f',$lvl,$opts));
68    }
69}
70
71/**
72 * The following functions are userfunctions to use with the search
73 * function above. This function is called for every found file or
74 * directory. When a directory is given to the function it has to
75 * decide if this directory should be traversed (true) or not (false)
76 * The function has to accept the following parameters:
77 *
78 * array &$data  - Reference to the result data structure
79 * string $base  - Base usually $conf['datadir']
80 * string $file  - current file or directory relative to $base
81 * string $type  - Type either 'd' for directory or 'f' for file
82 * int    $lvl   - Current recursion depht
83 * array  $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 *
99 * @param array $data
100 * @param string $base
101 * @param string $file
102 * @param string $type
103 * @param integer $lvl
104 * @param array $opts
105 *
106 * @return bool
107 */
108function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
109    $opts = array(
110            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
111            'listfiles' => true,
112            'pagesonly' => true,
113            );
114    return search_universal($data,$base,$file,$type,$lvl,$opts);
115}
116
117/**
118 * Build the browsable index of pages
119 *
120 * $opts['ns'] is the currently viewed namespace
121 *
122 * @author  Andreas Gohr <andi@splitbrain.org>
123 *
124 * @param array $data
125 * @param string $base
126 * @param string $file
127 * @param string $type
128 * @param integer $lvl
129 * @param array $opts
130 *
131 * @return bool
132 */
133function search_index(&$data,$base,$file,$type,$lvl,$opts){
134    global $conf;
135    $opts = array(
136        'pagesonly' => true,
137        'listdirs' => true,
138        'listfiles' => empty($opts['nofiles']),
139        'sneakyacl' => $conf['sneaky_index'],
140        // Hacky, should rather use recmatch
141        'depth' => preg_match('#^'.preg_quote($file, '#').'(/|$)#','/'.$opts['ns']) ? 0 : -1
142    );
143
144    return search_universal($data, $base, $file, $type, $lvl, $opts);
145}
146
147/**
148 * List all namespaces
149 *
150 * @author  Andreas Gohr <andi@splitbrain.org>
151 *
152 * @param array $data
153 * @param string $base
154 * @param string $file
155 * @param string $type
156 * @param integer $lvl
157 * @param array $opts
158 *
159 * @return bool
160 */
161function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
162    $opts = array(
163            'listdirs' => true,
164            );
165    return search_universal($data,$base,$file,$type,$lvl,$opts);
166}
167
168/**
169 * List all mediafiles in a namespace
170 *   $opts['depth']     recursion level, 0 for all
171 *   $opts['showmsg']   shows message if invalid media id is used
172 *   $opts['skipacl']   skip acl checking
173 *   $opts['pattern']   check given pattern
174 *   $opts['hash']      add hashes to result list
175 *
176 * @author  Andreas Gohr <andi@splitbrain.org>
177 *
178 * @param array $data
179 * @param string $base
180 * @param string $file
181 * @param string $type
182 * @param integer $lvl
183 * @param array $opts
184 *
185 * @return bool
186 */
187function search_media(&$data,$base,$file,$type,$lvl,$opts){
188
189    //we do nothing with directories
190    if($type == 'd') {
191        if(empty($opts['depth'])) return true; // recurse forever
192        $depth = substr_count($file,'/');
193        if($depth >= $opts['depth']) return false; // depth reached
194        return true;
195    }
196
197    $info         = array();
198    $info['id']   = pathID($file,true);
199    if($info['id'] != cleanID($info['id'])){
200        if($opts['showmsg'])
201            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
202        return false; // skip non-valid files
203    }
204
205    //check ACL for namespace (we have no ACL for mediafiles)
206    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
207    if(empty($opts['skipacl']) && $info['perm'] < AUTH_READ){
208        return false;
209    }
210
211    //check pattern filter
212    if(!empty($opts['pattern']) && !@preg_match($opts['pattern'], $info['id'])){
213        return false;
214    }
215
216    $info['file']     = \dokuwiki\Utf8\PhpString::basename($file);
217    $info['size']     = filesize($base.'/'.$file);
218    $info['mtime']    = filemtime($base.'/'.$file);
219    $info['writable'] = is_writable($base.'/'.$file);
220    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
221        $info['isimg'] = true;
222        $info['meta']  = new JpegMeta($base.'/'.$file);
223    }else{
224        $info['isimg'] = false;
225    }
226    if(!empty($opts['hash'])){
227        $info['hash'] = md5(io_readFile(mediaFN($info['id']),false));
228    }
229
230    $data[] = $info;
231
232    return false;
233}
234
235/**
236 * This function just lists documents (for RSS namespace export)
237 *
238 * @author  Andreas Gohr <andi@splitbrain.org>
239 *
240 * @param array $data
241 * @param string $base
242 * @param string $file
243 * @param string $type
244 * @param integer $lvl
245 * @param array $opts
246 *
247 * @return bool
248 */
249function search_list(&$data,$base,$file,$type,$lvl,$opts){
250    //we do nothing with directories
251    if($type == 'd') return false;
252    //only search txt files
253    if(substr($file,-4) == '.txt'){
254        //check ACL
255        $id = pathID($file);
256        if(auth_quickaclcheck($id) < AUTH_READ){
257            return false;
258        }
259        $data[]['id'] = $id;
260    }
261    return false;
262}
263
264/**
265 * Quicksearch for searching matching pagenames
266 *
267 * $opts['query'] is the search query
268 *
269 * @author  Andreas Gohr <andi@splitbrain.org>
270 *
271 * @param array $data
272 * @param string $base
273 * @param string $file
274 * @param string $type
275 * @param integer $lvl
276 * @param array $opts
277 *
278 * @return bool
279 */
280function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
281    //we do nothing with directories
282    if($type == 'd') return true;
283    //only search txt files
284    if(substr($file,-4) != '.txt') return true;
285
286    //simple stringmatching
287    if (!empty($opts['query'])){
288        if(strpos($file,$opts['query']) !== false){
289            //check ACL
290            $id = pathID($file);
291            if(auth_quickaclcheck($id) < AUTH_READ){
292                return false;
293            }
294            $data[]['id'] = $id;
295        }
296    }
297    return true;
298}
299
300/**
301 * Just lists all documents
302 *
303 * $opts['depth']   recursion level, 0 for all
304 * $opts['hash']    do md5 sum of content?
305 * $opts['skipacl'] list everything regardless of ACL
306 *
307 * @author  Andreas Gohr <andi@splitbrain.org>
308 *
309 * @param array $data
310 * @param string $base
311 * @param string $file
312 * @param string $type
313 * @param integer $lvl
314 * @param array $opts
315 *
316 * @return bool
317 */
318function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
319    if(isset($opts['depth']) && $opts['depth']){
320        $parts = explode('/',ltrim($file,'/'));
321        if(($type == 'd' && count($parts) >= $opts['depth'])
322          || ($type != 'd' && count($parts) > $opts['depth'])){
323            return false; // depth reached
324        }
325    }
326
327    //we do nothing with directories
328    if($type == 'd'){
329        return true;
330    }
331
332    //only search txt files
333    if(substr($file,-4) != '.txt') return true;
334
335    $item = array();
336    $item['id']   = pathID($file);
337    if(empty($opts['skipacl']) && auth_quickaclcheck($item['id']) < AUTH_READ){
338        return false;
339    }
340
341    $item['rev']   = filemtime($base.'/'.$file);
342    $item['mtime'] = $item['rev'];
343    $item['size']  = filesize($base.'/'.$file);
344    if(!empty($opts['hash'])){
345        $item['hash'] = md5(trim(rawWiki($item['id'])));
346    }
347
348    $data[] = $item;
349    return true;
350}
351
352/* ------------- helper functions below -------------- */
353
354/**
355 * fulltext sort
356 *
357 * Callback sort function for use with usort to sort the data
358 * structure created by search_fulltext. Sorts descending by count
359 *
360 * @author  Andreas Gohr <andi@splitbrain.org>
361 *
362 * @param array $a
363 * @param array $b
364 *
365 * @return int
366 */
367function sort_search_fulltext($a,$b){
368    if($a['count'] > $b['count']){
369        return -1;
370    }elseif($a['count'] < $b['count']){
371        return 1;
372    }else{
373        return Sort::strcmp($a['id'],$b['id']);
374    }
375}
376
377/**
378 * translates a document path to an ID
379 *
380 * @author  Andreas Gohr <andi@splitbrain.org>
381 * @todo    move to pageutils
382 *
383 * @param string $path
384 * @param bool $keeptxt
385 *
386 * @return mixed|string
387 */
388function pathID($path,$keeptxt=false){
389    $id = utf8_decodeFN($path);
390    $id = str_replace('/',':',$id);
391    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
392    $id = trim($id, ':');
393    return $id;
394}
395
396
397/**
398 * This is a very universal callback for the search() function, replacing
399 * many of the former individual functions at the cost of a more complex
400 * setup.
401 *
402 * How the function behaves, depends on the options passed in the $opts
403 * array, where the following settings can be used.
404 *
405 * depth      int     recursion depth. 0 for unlimited                       (default: 0)
406 * keeptxt    bool    keep .txt extension for IDs                            (default: false)
407 * listfiles  bool    include files in listing                               (default: false)
408 * listdirs   bool    include namespaces in listing                          (default: false)
409 * pagesonly  bool    restrict files to pages                                (default: false)
410 * skipacl    bool    do not check for READ permission                       (default: false)
411 * sneakyacl  bool    don't recurse into nonreadable dirs                    (default: false)
412 * hash       bool    create MD5 hash for files                              (default: false)
413 * meta       bool    return file metadata                                   (default: false)
414 * filematch  string  match files against this regexp                        (default: '', so accept everything)
415 * idmatch    string  match full ID against this regexp                      (default: '', so accept everything)
416 * dirmatch   string  match directory against this regexp when adding        (default: '', so accept everything)
417 * nsmatch    string  match namespace against this regexp when adding        (default: '', so accept everything)
418 * recmatch   string  match directory against this regexp when recursing     (default: '', so accept everything)
419 * showmsg    bool    warn about non-ID files                                (default: false)
420 * showhidden bool    show hidden files(e.g. by hidepages config) too        (default: false)
421 * firsthead  bool    return first heading for pages                         (default: false)
422 *
423 * @param array &$data  - Reference to the result data structure
424 * @param string $base  - Base usually $conf['datadir']
425 * @param string $file  - current file or directory relative to $base
426 * @param string $type  - Type either 'd' for directory or 'f' for file
427 * @param int    $lvl   - Current recursion depht
428 * @param array  $opts  - option array as given to search()
429 * @return bool if this directory should be traversed (true) or not (false)
430 *              return value is ignored for files
431 *
432 * @author Andreas Gohr <gohr@cosmocode.de>
433 */
434function search_universal(&$data,$base,$file,$type,$lvl,$opts){
435    $item   = array();
436    $return = true;
437
438    // get ID and check if it is a valid one
439    $item['id'] = pathID($file,($type == 'd' || !empty($opts['keeptxt'])));
440    if($item['id'] != cleanID($item['id'])){
441        if(!empty($opts['showmsg'])){
442            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
443        }
444        return false; // skip non-valid files
445    }
446    $item['ns']  = getNS($item['id']);
447
448    if($type == 'd') {
449        // decide if to recursion into this directory is wanted
450        if(empty($opts['depth'])){
451            $return = true; // recurse forever
452        }else{
453            $depth = substr_count($file,'/');
454            if($depth >= $opts['depth']){
455                $return = false; // depth reached
456            }else{
457                $return = true;
458            }
459        }
460
461        if ($return) {
462            $match = empty($opts['recmatch']) || preg_match('/'.$opts['recmatch'].'/',$file);
463            if (!$match) {
464                return false; // doesn't match
465            }
466        }
467    }
468
469    // check ACL
470    if(empty($opts['skipacl'])){
471        if($type == 'd'){
472            $item['perm'] = auth_quickaclcheck($item['id'].':*');
473        }else{
474            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
475        }
476    }else{
477        $item['perm'] = AUTH_DELETE;
478    }
479
480    // are we done here maybe?
481    if($type == 'd'){
482        if(empty($opts['listdirs'])) return $return;
483        //neither list nor recurse forbidden items:
484        if(empty($opts['skipacl']) && !empty($opts['sneakyacl']) && $item['perm'] < AUTH_READ) return false;
485        if(!empty($opts['dirmatch']) && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
486        if(!empty($opts['nsmatch']) && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
487    }else{
488        if(empty($opts['listfiles'])) return $return;
489        if(empty($opts['skipacl']) && $item['perm'] < AUTH_READ) return $return;
490        if(!empty($opts['pagesonly']) && (substr($file,-4) != '.txt')) return $return;
491        if(empty($opts['showhidden']) && isHiddenPage($item['id'])) return $return;
492        if(!empty($opts['filematch']) && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
493        if(!empty($opts['idmatch']) && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
494    }
495
496    // still here? prepare the item
497    $item['type']  = $type;
498    $item['level'] = $lvl;
499    $item['open']  = $return;
500
501    if(!empty($opts['meta'])){
502        $item['file']       = \dokuwiki\Utf8\PhpString::basename($file);
503        $item['size']       = filesize($base.'/'.$file);
504        $item['mtime']      = filemtime($base.'/'.$file);
505        $item['rev']        = $item['mtime'];
506        $item['writable']   = is_writable($base.'/'.$file);
507        $item['executable'] = is_executable($base.'/'.$file);
508    }
509
510    if($type == 'f'){
511        if(!empty($opts['hash'])) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
512        if(!empty($opts['firsthead'])) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
513    }
514
515    // finally add the item
516    $data[] = $item;
517    return $return;
518}
519
520//Setup VIM: ex: et ts=4 :
521