xref: /dokuwiki/inc/search.php (revision b4f2363aa1360136c8a826f09aaebc6505211c73) 
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
9/**
10 * Recurse directory
11 *
12 * This function recurses into a given base directory
13 * and calls the supplied function for each file and directory
14 *
15 * @param   array    &$data The results of the search are stored here
16 * @param   string    $base Where to start the search
17 * @param   callback  $func Callback (function name or array with object,method)
18 * @param   array     $opts option array will be given to the Callback
19 * @param   string    $dir  Current directory beyond $base
20 * @param   int       $lvl  Recursion Level
21 * @param   mixed     $sort 'natural' to use natural order sorting (default); 'date' to sort by filemtime; leave empty to skip sorting.
22 * @author  Andreas Gohr <andi@splitbrain.org>
23 */
24function search(&$data,$base,$func,$opts,$dir='',$lvl=1,$sort='natural'){
25    $dirs   = array();
26    $files  = array();
27    $filepaths = array();
28
29    // safeguard against runaways #1452
30    if($base == '' || $base == '/') {
31        throw new RuntimeException('No valid $base passed to search() - possible misconfiguration or bug');
32    }
33
34    //read in directories and files
35    $dh = @opendir($base.'/'.$dir);
36    if(!$dh) return;
37    while(($file = readdir($dh)) !== false){
38        if(preg_match('/^[\._]/',$file)) continue; //skip hidden files and upper dirs
39        if(is_dir($base.'/'.$dir.'/'.$file)){
40            $dirs[] = $dir.'/'.$file;
41            continue;
42        }
43        $files[] = $dir.'/'.$file;
44        $filepaths[] = $base.'/'.$dir.'/'.$file;
45    }
46    closedir($dh);
47    if (!empty($sort)) {
48        if ($sort == 'date') {
49            @array_multisort(array_map('filemtime', $filepaths), SORT_NUMERIC, SORT_DESC, $files);
50        } else /* natural */ {
51            natsort($files);
52        }
53        natsort($dirs);
54    }
55
56    //give directories to userfunction then recurse
57    foreach($dirs as $dir){
58        if (call_user_func_array($func, array(&$data,$base,$dir,'d',$lvl,$opts))){
59            search($data,$base,$func,$opts,$dir,$lvl+1,$sort);
60        }
61    }
62    //now handle the files
63    foreach($files as $file){
64        call_user_func_array($func, array(&$data,$base,$file,'f',$lvl,$opts));
65    }
66}
67
68/**
69 * The following functions are userfunctions to use with the search
70 * function above. This function is called for every found file or
71 * directory. When a directory is given to the function it has to
72 * decide if this directory should be traversed (true) or not (false)
73 * The function has to accept the following parameters:
74 *
75 * array &$data  - Reference to the result data structure
76 * string $base  - Base usually $conf['datadir']
77 * string $file  - current file or directory relative to $base
78 * string $type  - Type either 'd' for directory or 'f' for file
79 * int    $lvl   - Current recursion depht
80 * array  $opts  - option array as given to search()
81 *
82 * return values for files are ignored
83 *
84 * All functions should check the ACL for document READ rights
85 * namespaces (directories) are NOT checked (when sneaky_index is 0) as this
86 * would break the recursion (You can have an nonreadable dir over a readable
87 * one deeper nested) also make sure to check the file type (for example
88 * in case of lockfiles).
89 */
90
91/**
92 * Searches for pages beginning with the given query
93 *
94 * @author Andreas Gohr <andi@splitbrain.org>
95 *
96 * @param array $data
97 * @param string $base
98 * @param string $file
99 * @param string $type
100 * @param integer $lvl
101 * @param array $opts
102 *
103 * @return bool
104 */
105function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
106    $opts = array(
107            'idmatch'   => '(^|:)'.preg_quote($opts['query'],'/').'/',
108            'listfiles' => true,
109            'pagesonly' => true,
110            );
111    return search_universal($data,$base,$file,$type,$lvl,$opts);
112}
113
114/**
115 * Build the browsable index of pages
116 *
117 * $opts['ns'] is the currently viewed namespace
118 *
119 * @author  Andreas Gohr <andi@splitbrain.org>
120 *
121 * @param array $data
122 * @param string $base
123 * @param string $file
124 * @param string $type
125 * @param integer $lvl
126 * @param array $opts
127 *
128 * @return bool
129 */
130function search_index(&$data,$base,$file,$type,$lvl,$opts){
131    global $conf;
132    $opts = array(
133        'pagesonly' => true,
134        'listdirs' => true,
135        'listfiles' => empty($opts['nofiles']),
136        'sneakyacl' => $conf['sneaky_index'],
137        // Hacky, should rather use recmatch
138        'depth' => preg_match('#^'.preg_quote($file, '#').'(/|$)#','/'.$opts['ns']) ? 0 : -1
139    );
140
141    return search_universal($data, $base, $file, $type, $lvl, $opts);
142}
143
144/**
145 * List all namespaces
146 *
147 * @author  Andreas Gohr <andi@splitbrain.org>
148 *
149 * @param array $data
150 * @param string $base
151 * @param string $file
152 * @param string $type
153 * @param integer $lvl
154 * @param array $opts
155 *
156 * @return bool
157 */
158function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
159    $opts = array(
160            'listdirs' => true,
161            );
162    return search_universal($data,$base,$file,$type,$lvl,$opts);
163}
164
165/**
166 * List all mediafiles in a namespace
167 *   $opts['depth']     recursion level, 0 for all
168 *   $opts['showmsg']   shows message if invalid media id is used
169 *   $opts['skipacl']   skip acl checking
170 *   $opts['pattern']   check given pattern
171 *   $opts['hash']      add hashes to result list
172 *
173 * @author  Andreas Gohr <andi@splitbrain.org>
174 *
175 * @param array $data
176 * @param string $base
177 * @param string $file
178 * @param string $type
179 * @param integer $lvl
180 * @param array $opts
181 *
182 * @return bool
183 */
184function search_media(&$data,$base,$file,$type,$lvl,$opts){
185
186    //we do nothing with directories
187    if($type == 'd') {
188        if(empty($opts['depth'])) return true; // recurse forever
189        $depth = substr_count($file,'/');
190        if($depth >= $opts['depth']) return false; // depth reached
191        return true;
192    }
193
194    $info         = array();
195    $info['id']   = pathID($file,true);
196    if($info['id'] != cleanID($info['id'])){
197        if($opts['showmsg'])
198            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
199        return false; // skip non-valid files
200    }
201
202    //check ACL for namespace (we have no ACL for mediafiles)
203    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
204    if(empty($opts['skipacl']) && $info['perm'] < AUTH_READ){
205        return false;
206    }
207
208    //check pattern filter
209    if(!empty($opts['pattern']) && !@preg_match($opts['pattern'], $info['id'])){
210        return false;
211    }
212
213    $info['file']     = utf8_basename($file);
214    $info['size']     = filesize($base.'/'.$file);
215    $info['mtime']    = filemtime($base.'/'.$file);
216    $info['writable'] = is_writable($base.'/'.$file);
217    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
218        $info['isimg'] = true;
219        $info['meta']  = new JpegMeta($base.'/'.$file);
220    }else{
221        $info['isimg'] = false;
222    }
223    if(!empty($opts['hash'])){
224        $info['hash'] = md5(io_readFile(mediaFN($info['id']),false));
225    }
226
227    $data[] = $info;
228
229    return false;
230}
231
232/**
233 * This function just lists documents (for RSS namespace export)
234 *
235 * @author  Andreas Gohr <andi@splitbrain.org>
236 *
237 * @param array $data
238 * @param string $base
239 * @param string $file
240 * @param string $type
241 * @param integer $lvl
242 * @param array $opts
243 *
244 * @return bool
245 */
246function search_list(&$data,$base,$file,$type,$lvl,$opts){
247    //we do nothing with directories
248    if($type == 'd') return false;
249    //only search txt files
250    if(substr($file,-4) == '.txt'){
251        //check ACL
252        $id = pathID($file);
253        if(auth_quickaclcheck($id) < AUTH_READ){
254            return false;
255        }
256        $data[]['id'] = $id;
257    }
258    return false;
259}
260
261/**
262 * Quicksearch for searching matching pagenames
263 *
264 * $opts['query'] is the search query
265 *
266 * @author  Andreas Gohr <andi@splitbrain.org>
267 *
268 * @param array $data
269 * @param string $base
270 * @param string $file
271 * @param string $type
272 * @param integer $lvl
273 * @param array $opts
274 *
275 * @return bool
276 */
277function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
278    //we do nothing with directories
279    if($type == 'd') return true;
280    //only search txt files
281    if(substr($file,-4) != '.txt') return true;
282
283    //simple stringmatching
284    if (!empty($opts['query'])){
285        if(strpos($file,$opts['query']) !== false){
286            //check ACL
287            $id = pathID($file);
288            if(auth_quickaclcheck($id) < AUTH_READ){
289                return false;
290            }
291            $data[]['id'] = $id;
292        }
293    }
294    return true;
295}
296
297/**
298 * Just lists all documents
299 *
300 * $opts['depth']   recursion level, 0 for all
301 * $opts['hash']    do md5 sum of content?
302 * $opts['skipacl'] list everything regardless of ACL
303 *
304 * @author  Andreas Gohr <andi@splitbrain.org>
305 *
306 * @param array $data
307 * @param string $base
308 * @param string $file
309 * @param string $type
310 * @param integer $lvl
311 * @param array $opts
312 *
313 * @return bool
314 */
315function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
316    if(isset($opts['depth']) && $opts['depth']){
317        $parts = explode('/',ltrim($file,'/'));
318        if(($type == 'd' && count($parts) >= $opts['depth'])
319          || ($type != 'd' && count($parts) > $opts['depth'])){
320            return false; // depth reached
321        }
322    }
323
324    //we do nothing with directories
325    if($type == 'd'){
326        return true;
327    }
328
329    //only search txt files
330    if(substr($file,-4) != '.txt') return true;
331
332    $item = array();
333    $item['id']   = pathID($file);
334    if(isset($opts['skipacl']) && !$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
335        return false;
336    }
337
338    $item['rev']   = filemtime($base.'/'.$file);
339    $item['mtime'] = $item['rev'];
340    $item['size']  = filesize($base.'/'.$file);
341    if(!empty($opts['hash'])){
342        $item['hash'] = md5(trim(rawWiki($item['id'])));
343    }
344
345    $data[] = $item;
346    return true;
347}
348
349/* ------------- helper functions below -------------- */
350
351/**
352 * fulltext sort
353 *
354 * Callback sort function for use with usort to sort the data
355 * structure created by search_fulltext. Sorts descending by count
356 *
357 * @author  Andreas Gohr <andi@splitbrain.org>
358 *
359 * @param array $a
360 * @param array $b
361 *
362 * @return int
363 */
364function sort_search_fulltext($a,$b){
365    if($a['count'] > $b['count']){
366        return -1;
367    }elseif($a['count'] < $b['count']){
368        return 1;
369    }else{
370        return strcmp($a['id'],$b['id']);
371    }
372}
373
374/**
375 * translates a document path to an ID
376 *
377 * @author  Andreas Gohr <andi@splitbrain.org>
378 * @todo    move to pageutils
379 *
380 * @param string $path
381 * @param bool $keeptxt
382 *
383 * @return mixed|string
384 */
385function pathID($path,$keeptxt=false){
386    $id = utf8_decodeFN($path);
387    $id = str_replace('/',':',$id);
388    if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
389    $id = trim($id, ':');
390    return $id;
391}
392
393
394/**
395 * This is a very universal callback for the search() function, replacing
396 * many of the former individual functions at the cost of a more complex
397 * setup.
398 *
399 * How the function behaves, depends on the options passed in the $opts
400 * array, where the following settings can be used.
401 *
402 * depth      int     recursion depth. 0 for unlimited                       (default: 0)
403 * keeptxt    bool    keep .txt extension for IDs                            (default: false)
404 * listfiles  bool    include files in listing                               (default: false)
405 * listdirs   bool    include namespaces in listing                          (default: false)
406 * pagesonly  bool    restrict files to pages                                (default: false)
407 * skipacl    bool    do not check for READ permission                       (default: false)
408 * sneakyacl  bool    don't recurse into nonreadable dirs                    (default: false)
409 * hash       bool    create MD5 hash for files                              (default: false)
410 * meta       bool    return file metadata                                   (default: false)
411 * filematch  string  match files against this regexp                        (default: '', so accept everything)
412 * idmatch    string  match full ID against this regexp                      (default: '', so accept everything)
413 * dirmatch   string  match directory against this regexp when adding        (default: '', so accept everything)
414 * nsmatch    string  match namespace against this regexp when adding        (default: '', so accept everything)
415 * recmatch   string  match directory against this regexp when recursing     (default: '', so accept everything)
416 * showmsg    bool    warn about non-ID files                                (default: false)
417 * showhidden bool    show hidden files(e.g. by hidepages config) too        (default: false)
418 * firsthead  bool    return first heading for pages                         (default: false)
419 *
420 * @param array &$data  - Reference to the result data structure
421 * @param string $base  - Base usually $conf['datadir']
422 * @param string $file  - current file or directory relative to $base
423 * @param string $type  - Type either 'd' for directory or 'f' for file
424 * @param int    $lvl   - Current recursion depht
425 * @param array  $opts  - option array as given to search()
426 * @return bool if this directory should be traversed (true) or not (false)
427 *              return value is ignored for files
428 *
429 * @author Andreas Gohr <gohr@cosmocode.de>
430 */
431function search_universal(&$data,$base,$file,$type,$lvl,$opts){
432    $item   = array();
433    $return = true;
434
435    // get ID and check if it is a valid one
436    $item['id'] = pathID($file,($type == 'd' || !empty($opts['keeptxt'])));
437    if($item['id'] != cleanID($item['id'])){
438        if(!empty($opts['showmsg'])){
439            msg(hsc($item['id']).' is not a valid file name for DokuWiki - skipped',-1);
440        }
441        return false; // skip non-valid files
442    }
443    $item['ns']  = getNS($item['id']);
444
445    if($type == 'd') {
446        // decide if to recursion into this directory is wanted
447        if(empty($opts['depth'])){
448            $return = true; // recurse forever
449        }else{
450            $depth = substr_count($file,'/');
451            if($depth >= $opts['depth']){
452                $return = false; // depth reached
453            }else{
454                $return = true;
455            }
456        }
457
458        if ($return) {
459            $match = empty($opts['recmatch']) || preg_match('/'.$opts['recmatch'].'/',$file);
460            if (!$match) {
461                return false; // doesn't match
462            }
463        }
464    }
465
466    // check ACL
467    if(empty($opts['skipacl'])){
468        if($type == 'd'){
469            $item['perm'] = auth_quickaclcheck($item['id'].':*');
470        }else{
471            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
472        }
473    }else{
474        $item['perm'] = AUTH_DELETE;
475    }
476
477    // are we done here maybe?
478    if($type == 'd'){
479        if(empty($opts['listdirs'])) return $return;
480        if(empty($opts['skipacl']) && !empty($opts['sneakyacl']) && $item['perm'] < AUTH_READ) return false; //neither list nor recurse
481        if(!empty($opts['dirmatch']) && !preg_match('/'.$opts['dirmatch'].'/',$file)) return $return;
482        if(!empty($opts['nsmatch']) && !preg_match('/'.$opts['nsmatch'].'/',$item['ns'])) return $return;
483    }else{
484        if(empty($opts['listfiles'])) return $return;
485        if(empty($opts['skipacl']) && $item['perm'] < AUTH_READ) return $return;
486        if(!empty($opts['pagesonly']) && (substr($file,-4) != '.txt')) return $return;
487        if(empty($opts['showhidden']) && isHiddenPage($item['id'])) return $return;
488        if(!empty($opts['filematch']) && !preg_match('/'.$opts['filematch'].'/',$file)) return $return;
489        if(!empty($opts['idmatch']) && !preg_match('/'.$opts['idmatch'].'/',$item['id'])) return $return;
490    }
491
492    // still here? prepare the item
493    $item['type']  = $type;
494    $item['level'] = $lvl;
495    $item['open']  = $return;
496
497    if(!empty($opts['meta'])){
498        $item['file']       = utf8_basename($file);
499        $item['size']       = filesize($base.'/'.$file);
500        $item['mtime']      = filemtime($base.'/'.$file);
501        $item['rev']        = $item['mtime'];
502        $item['writable']   = is_writable($base.'/'.$file);
503        $item['executable'] = is_executable($base.'/'.$file);
504    }
505
506    if($type == 'f'){
507        if(!empty($opts['hash'])) $item['hash'] = md5(io_readFile($base.'/'.$file,false));
508        if(!empty($opts['firsthead'])) $item['title'] = p_get_first_heading($item['id'],METADATA_DONT_RENDER);
509    }
510
511    // finally add the item
512    $data[] = $item;
513    return $return;
514}
515
516//Setup VIM: ex: et ts=4 :
517