xref: /dokuwiki/inc/search.php (revision 0b1bbbbb7d4e3c531cd255dbf878ce27d5967a0c)
1<?php
2
3/**
4 * DokuWiki search functions
5 *
6 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
7 * @author     Andreas Gohr <andi@splitbrain.org>
8 */
9
10use dokuwiki\Utf8\PhpString;
11use dokuwiki\File\MediaFile;
12use dokuwiki\Utf8\Sort;
13
14/**
15 * Recurse directory
16 *
17 * This function recurses into a given base directory
18 * and calls the supplied function for each file and directory
19 *
20 * @param   array    &$data The results of the search are stored here
21 * @param   string    $base Where to start the search
22 * @param   callback  $func Callback (function name or array with object,method)
23 * @param   array     $opts option array will be given to the Callback
24 * @param   string    $dir  Current directory beyond $base
25 * @param   int       $lvl  Recursion Level
26 * @param   mixed     $sort 'natural' to use natural order sorting (default);
27 *                          'date' to sort by filemtime; leave empty to skip sorting.
28 * @author  Andreas Gohr <andi@splitbrain.org>
29 */
30function search(&$data, $base, $func, $opts, $dir = '', $lvl = 1, $sort = 'natural')
31{
32    $dirs   = [];
33    $files  = [];
34    $filepaths = [];
35
36    // safeguard against runaways #1452
37    if ($base == '' || $base == '/') {
38        throw new RuntimeException('No valid $base passed to search() - possible misconfiguration or bug');
39    }
40
41    //read in directories and files
42    $dh = @opendir($base . '/' . $dir);
43    if (!$dh) return;
44    while (($file = readdir($dh)) !== false) {
45        if (preg_match('/^[\._]/', $file)) continue; //skip hidden files and upper dirs
46        if (is_dir($base . '/' . $dir . '/' . $file)) {
47            $dirs[] = $dir . '/' . $file;
48            continue;
49        }
50        $files[] = $dir . '/' . $file;
51        $filepaths[] = $base . '/' . $dir . '/' . $file;
52    }
53    closedir($dh);
54    if (!empty($sort)) {
55        if ($sort == 'date') {
56            @array_multisort(array_map(filemtime(...), $filepaths), SORT_NUMERIC, SORT_DESC, $files);
57        } else /* natural */ {
58            Sort::asortFN($files);
59        }
60        Sort::asortFN($dirs);
61    }
62
63    //give directories to userfunction then recurse
64    foreach ($dirs as $dir) {
65        if (call_user_func_array($func, [&$data, $base, $dir, 'd', $lvl, $opts])) {
66            search($data, $base, $func, $opts, $dir, $lvl + 1, $sort);
67        }
68    }
69    //now handle the files
70    foreach ($files as $file) {
71        call_user_func_array($func, [&$data, $base, $file, 'f', $lvl, $opts]);
72    }
73}
74
75/**
76 * The following functions are userfunctions to use with the search
77 * function above. This function is called for every found file or
78 * directory. When a directory is given to the function it has to
79 * decide if this directory should be traversed (true) or not (false)
80 * The function has to accept the following parameters:
81 *
82 * array &$data  - Reference to the result data structure
83 * string $base  - Base usually $conf['datadir']
84 * string $file  - current file or directory relative to $base
85 * string $type  - Type either 'd' for directory or 'f' for file
86 * int    $lvl   - Current recursion depht
87 * array  $opts  - option array as given to search()
88 *
89 * return values for files are ignored
90 *
91 * All functions should check the ACL for document READ rights
92 * namespaces (directories) are NOT checked (when sneaky_index is 0) as this
93 * would break the recursion (You can have an nonreadable dir over a readable
94 * one deeper nested) also make sure to check the file type (for example
95 * in case of lockfiles).
96 */
97
98/**
99 * Searches for pages beginning with the given query
100 *
101 * @author Andreas Gohr <andi@splitbrain.org>
102 *
103 * @param array $data
104 * @param string $base
105 * @param string $file
106 * @param string $type
107 * @param integer $lvl
108 * @param array $opts
109 *
110 * @return bool
111 */
112function search_qsearch(&$data, $base, $file, $type, $lvl, $opts)
113{
114    $opts = [
115        'idmatch'   => '(^|:)' . preg_quote($opts['query'], '/') . '/',
116        'listfiles' => true,
117        'pagesonly' => true
118    ];
119    return search_universal($data, $base, $file, $type, $lvl, $opts);
120}
121
122/**
123 * Build the browsable index of pages
124 *
125 * $opts['ns'] is the currently viewed namespace
126 *
127 * @author  Andreas Gohr <andi@splitbrain.org>
128 *
129 * @param array $data
130 * @param string $base
131 * @param string $file
132 * @param string $type
133 * @param integer $lvl
134 * @param array $opts
135 *
136 * @return bool
137 */
138function search_index(&$data, $base, $file, $type, $lvl, $opts)
139{
140    global $conf;
141    $ns = $opts['ns'] ?? '';
142    $opts = [
143        'pagesonly' => true,
144        'listdirs' => true,
145        'listfiles' => empty($opts['nofiles']),
146        'sneakyacl' => $conf['sneaky_index'],
147        // Hacky, should rather use recmatch
148        'depth' => preg_match('#^' . preg_quote($file, '#') . '(/|$)#', '/' . $ns) ? 0 : -1,
149    ];
150
151    return search_universal($data, $base, $file, $type, $lvl, $opts);
152}
153
154/**
155 * List all namespaces
156 *
157 * @author  Andreas Gohr <andi@splitbrain.org>
158 *
159 * @param array $data
160 * @param string $base
161 * @param string $file
162 * @param string $type
163 * @param integer $lvl
164 * @param array $opts
165 *
166 * @return bool
167 */
168function search_namespaces(&$data, $base, $file, $type, $lvl, $opts)
169{
170    $opts = ['listdirs' => true];
171    return search_universal($data, $base, $file, $type, $lvl, $opts);
172}
173
174/**
175 * List all mediafiles in a namespace
176 *   $opts['depth']     recursion level, 0 for all
177 *   $opts['showmsg']   shows message if invalid media id is used
178 *   $opts['skipacl']   skip acl checking
179 *   $opts['pattern']   check given pattern
180 *   $opts['hash']      add hashes to result list
181 *
182 * @author  Andreas Gohr <andi@splitbrain.org>
183 *
184 * @param array $data
185 * @param string $base
186 * @param string $file
187 * @param string $type
188 * @param integer $lvl
189 * @param array $opts
190 *
191 * @return bool
192 */
193function search_media(&$data, $base, $file, $type, $lvl, $opts)
194{
195    //we do nothing with directories
196    if ($type == 'd') {
197        if (empty($opts['depth'])) return true; // recurse forever
198        $depth = substr_count($file, '/');
199        if ($depth >= $opts['depth']) return false; // depth reached
200        return true;
201    }
202
203    $info         = [];
204    $info['id']   = pathID($file, true);
205    if ($info['id'] !== cleanID($info['id'])) {
206        if (!empty($opts['showmsg']))
207            msg(hsc($info['id']) . ' is not a valid file name for DokuWiki - skipped', -1);
208        return false; // skip non-valid files
209    }
210
211    //check ACL for namespace (we have no ACL for mediafiles)
212    $info['perm'] = auth_quickaclcheck(getNS($info['id']) . ':*');
213    if (empty($opts['skipacl']) && $info['perm'] < AUTH_READ) {
214        return false;
215    }
216
217    //check pattern filter
218    if (!empty($opts['pattern']) && !@preg_match($opts['pattern'], $info['id'])) {
219        return false;
220    }
221
222    $info['file']     = PhpString::basename($file);
223    $info['size']     = filesize($base . '/' . $file);
224    $info['mtime']    = filemtime($base . '/' . $file);
225    $info['writable'] = is_writable($base . '/' . $file);
226    if (preg_match("/\.(jpe?g|gif|png)$/", $file)) {
227        $info['isimg'] = true;
228        $info['meta']  = new JpegMeta($base . '/' . $file);
229    } else {
230        $info['isimg'] = false;
231    }
232    if (!empty($opts['hash'])) {
233        $info['hash'] = md5(io_readFile(mediaFN($info['id']), false));
234    }
235
236    $data[] = $info;
237
238    return false;
239}
240
241/**
242 * List all mediafiles in a namespace
243 *   $opts['depth']     recursion level, 0 for all
244 *   $opts['showmsg']   shows message if invalid media id is used
245 *   $opts['skipacl']   skip acl checking
246 *   $opts['pattern']   check given pattern
247 *   $opts['hash']      add hashes to result list
248 *
249 * @todo This is a temporary copy of search_media returning a list of MediaFile intances
250 *
251 * @param array $data
252 * @param string $base
253 * @param string $file
254 * @param string $type
255 * @param integer $lvl
256 * @param array $opts
257 *
258 * @return bool
259 */
260function search_mediafiles(&$data, $base, $file, $type, $lvl, $opts)
261{
262
263    //we do nothing with directories
264    if ($type == 'd') {
265        if (empty($opts['depth'])) return true; // recurse forever
266        $depth = substr_count($file, '/');
267        if ($depth >= $opts['depth']) return false; // depth reached
268        return true;
269    }
270
271    $id   = pathID($file, true);
272    if ($id != cleanID($id)) {
273        if ($opts['showmsg'])
274            msg(hsc($id) . ' is not a valid file name for DokuWiki - skipped', -1);
275        return false; // skip non-valid files
276    }
277
278    //check ACL for namespace (we have no ACL for mediafiles)
279    $info['perm'] = auth_quickaclcheck(getNS($id) . ':*');
280    if (empty($opts['skipacl']) && $info['perm'] < AUTH_READ) {
281        return false;
282    }
283
284    //check pattern filter
285    if (!empty($opts['pattern']) && !@preg_match($opts['pattern'], $id)) {
286        return false;
287    }
288
289    $data[] = new MediaFile($id);
290    return false;
291}
292
293
294/**
295 * This function just lists documents (for RSS namespace export)
296 *
297 * @author  Andreas Gohr <andi@splitbrain.org>
298 *
299 * @param array $data
300 * @param string $base
301 * @param string $file
302 * @param string $type
303 * @param integer $lvl
304 * @param array $opts
305 *
306 * @return bool
307 */
308function search_list(&$data, $base, $file, $type, $lvl, $opts)
309{
310    //we do nothing with directories
311    if ($type == 'd') return false;
312    //only search txt files
313    if (str_ends_with($file,  '.txt') ) {
314        //check ACL
315        $id = pathID($file);
316        if (auth_quickaclcheck($id) < AUTH_READ) {
317            return false;
318        }
319        $data[]['id'] = $id;
320    }
321    return false;
322}
323
324/**
325 * Quicksearch for searching matching pagenames
326 *
327 * $opts['query'] is the search query
328 *
329 * @author  Andreas Gohr <andi@splitbrain.org>
330 *
331 * @param array $data
332 * @param string $base
333 * @param string $file
334 * @param string $type
335 * @param integer $lvl
336 * @param array $opts
337 *
338 * @return bool
339 */
340function search_pagename(&$data, $base, $file, $type, $lvl, $opts)
341{
342    //we do nothing with directories
343    if ($type == 'd') return true;
344    //only search txt files
345    if (!str_ends_with($file,  '.txt')) return true;
346
347    //simple stringmatching
348    if (!empty($opts['query'])) {
349        if (str_contains($file, (string) $opts['query'])) {
350            //check ACL
351            $id = pathID($file);
352            if (auth_quickaclcheck($id) < AUTH_READ) {
353                return false;
354            }
355            $data[]['id'] = $id;
356        }
357    }
358    return true;
359}
360
361/**
362 * Just lists all documents
363 *
364 * $opts['depth']   recursion level, 0 for all
365 * $opts['hash']    do md5 sum of content?
366 * $opts['skipacl'] list everything regardless of ACL
367 *
368 * @author  Andreas Gohr <andi@splitbrain.org>
369 *
370 * @param array $data
371 * @param string $base
372 * @param string $file
373 * @param string $type
374 * @param integer $lvl
375 * @param array $opts
376 *
377 * @return bool
378 */
379function search_allpages(&$data, $base, $file, $type, $lvl, $opts)
380{
381    if (($opts['depth'] ?? 0) > 0) {
382        $parts = explode('/', ltrim($file, '/'));
383        if (
384            ($type == 'd' && count($parts) >= $opts['depth'])
385            || ($type != 'd' && count($parts) > $opts['depth'])
386        ) {
387            return false; // depth reached
388        }
389    }
390
391    //we do nothing with directories
392    if ($type == 'd') {
393        return true;
394    }
395
396    //only search txt files
397    if (!str_ends_with($file,  '.txt')) return true;
398
399    $item = [];
400    $item['id']   = pathID($file);
401    if (empty($opts['skipacl']) && auth_quickaclcheck($item['id']) < AUTH_READ) {
402        return false;
403    }
404
405    $item['rev']   = filemtime($base . '/' . $file);
406    $item['mtime'] = $item['rev'];
407    $item['size']  = filesize($base . '/' . $file);
408    if (!empty($opts['hash'])) {
409        $item['hash'] = md5(trim(rawWiki($item['id'])));
410    }
411
412    $data[] = $item;
413    return true;
414}
415
416/* ------------- helper functions below -------------- */
417
418/**
419 * fulltext sort
420 *
421 * Callback sort function for use with usort to sort the data
422 * structure created by search_fulltext. Sorts descending by count
423 *
424 * @author  Andreas Gohr <andi@splitbrain.org>
425 *
426 * @param array $a
427 * @param array $b
428 *
429 * @return int
430 */
431function sort_search_fulltext($a, $b)
432{
433    if ($a['count'] > $b['count']) {
434        return -1;
435    } elseif ($a['count'] < $b['count']) {
436        return 1;
437    } else {
438        return Sort::strcmp($a['id'], $b['id']);
439    }
440}
441
442/**
443 * translates a document path to an ID
444 *
445 * @author  Andreas Gohr <andi@splitbrain.org>
446 * @todo    move to pageutils
447 *
448 * @param string $path
449 * @param bool $keeptxt
450 *
451 * @return string
452 */
453function pathID($path, $keeptxt = false)
454{
455    $id = utf8_decodeFN($path);
456    $id = str_replace('/', ':', $id);
457    if (!$keeptxt) $id = preg_replace('#\.txt$#', '', $id);
458    $id = trim($id, ':');
459    return $id;
460}
461
462
463/**
464 * This is a very universal callback for the search() function, replacing
465 * many of the former individual functions at the cost of a more complex
466 * setup.
467 *
468 * How the function behaves, depends on the options passed in the $opts
469 * array, where the following settings can be used.
470 *
471 * depth      int     recursion depth. 0 for unlimited                       (default: 0)
472 * keeptxt    bool    keep .txt extension for IDs                            (default: false)
473 * listfiles  bool    include files in listing                               (default: false)
474 * listdirs   bool    include namespaces in listing                          (default: false)
475 * pagesonly  bool    restrict files to pages                                (default: false)
476 * skipacl    bool    do not check for READ permission                       (default: false)
477 * sneakyacl  bool    don't recurse into nonreadable dirs                    (default: false)
478 * hash       bool    create MD5 hash for files                              (default: false)
479 * meta       bool    return file metadata                                   (default: false)
480 * filematch  string  match files against this regexp                        (default: '', so accept everything)
481 * idmatch    string  match full ID against this regexp                      (default: '', so accept everything)
482 * dirmatch   string  match directory against this regexp when adding        (default: '', so accept everything)
483 * nsmatch    string  match namespace against this regexp when adding        (default: '', so accept everything)
484 * recmatch   string  match directory against this regexp when recursing     (default: '', so accept everything)
485 * showmsg    bool    warn about non-ID files                                (default: false)
486 * showhidden bool    show hidden files(e.g. by hidepages config) too        (default: false)
487 * firsthead  bool    return first heading for pages                         (default: false)
488 *
489 * @param array &$data  - Reference to the result data structure
490 * @param string $base  - Base usually $conf['datadir']
491 * @param string $file  - current file or directory relative to $base
492 * @param string $type  - Type either 'd' for directory or 'f' for file
493 * @param int    $lvl   - Current recursion depht
494 * @param array  $opts  - option array as given to search()
495 * @return bool if this directory should be traversed (true) or not (false)
496 *              return value is ignored for files
497 *
498 * @author Andreas Gohr <gohr@cosmocode.de>
499 */
500function search_universal(&$data, $base, $file, $type, $lvl, $opts)
501{
502    $item   = [];
503    $return = true;
504
505    // get ID and check if it is a valid one
506    $item['id'] = pathID($file, ($type == 'd' || !empty($opts['keeptxt'])));
507    if ($item['id'] !== cleanID($item['id'])) {
508        if (!empty($opts['showmsg'])) {
509            msg(hsc($item['id']) . ' is not a valid file name for DokuWiki - skipped', -1);
510        }
511        return false; // skip non-valid files
512    }
513    $item['ns']  = getNS($item['id']);
514
515    if ($type == 'd') {
516        // decide if to recursion into this directory is wanted
517        if (empty($opts['depth'])) {
518            $return = true; // recurse forever
519        } else {
520            $depth = substr_count($file, '/');
521            if ($depth >= $opts['depth']) {
522                $return = false; // depth reached
523            } else {
524                $return = true;
525            }
526        }
527
528        if ($return) {
529            $match = empty($opts['recmatch']) || preg_match('/' . $opts['recmatch'] . '/', $file);
530            if (!$match) {
531                return false; // doesn't match
532            }
533        }
534    }
535
536    // check ACL
537    if (empty($opts['skipacl'])) {
538        if ($type == 'd') {
539            $item['perm'] = auth_quickaclcheck($item['id'] . ':*');
540        } else {
541            $item['perm'] = auth_quickaclcheck($item['id']); //FIXME check namespace for media files
542        }
543    } else {
544        $item['perm'] = AUTH_DELETE;
545    }
546
547    // are we done here maybe?
548    if ($type == 'd') {
549        if (empty($opts['listdirs'])) return $return;
550        //neither list nor recurse forbidden items:
551        if (empty($opts['skipacl']) && !empty($opts['sneakyacl']) && $item['perm'] < AUTH_READ) return false;
552        if (!empty($opts['dirmatch']) && !preg_match('/'.$opts['dirmatch'].'/', $file)) return $return;
553        if (!empty($opts['nsmatch']) && !preg_match('/'.$opts['nsmatch'].'/', $item['ns'])) return $return;
554    } else {
555        if (empty($opts['listfiles'])) return $return;
556        if (empty($opts['skipacl']) && $item['perm'] < AUTH_READ) return $return;
557        if (!empty($opts['pagesonly']) && !str_ends_with($file,  '.txt')) return $return;
558        if (empty($opts['showhidden']) && isHiddenPage($item['id'])) return $return;
559        if (!empty($opts['filematch']) && !preg_match('/'.$opts['filematch'].'/', $file)) return $return;
560        if (!empty($opts['idmatch']) && !preg_match('/'.$opts['idmatch'].'/', $item['id'])) return $return;
561    }
562
563    // still here? prepare the item
564    $item['type']  = $type;
565    $item['level'] = $lvl;
566    $item['open']  = $return;
567
568    if (!empty($opts['meta'])) {
569        $item['file']       = PhpString::basename($file);
570        $item['size']       = filesize($base . '/' . $file);
571        $item['mtime']      = filemtime($base . '/' . $file);
572        $item['rev']        = $item['mtime'];
573        $item['writable']   = is_writable($base . '/' . $file);
574        $item['executable'] = is_executable($base . '/' . $file);
575    }
576
577    if ($type == 'f') {
578        if (!empty($opts['hash'])) $item['hash'] = md5(io_readFile($base . '/' . $file, false));
579        if (!empty($opts['firsthead'])) {
580            $item['title'] = p_get_first_heading($item['id'], METADATA_DONT_RENDER);
581        }
582    }
583
584    // finally add the item
585    $data[] = $item;
586    return $return;
587}
588
589//Setup VIM: ex: et ts=4 :
590