xref: /dokuwiki/inc/search.php (revision b14a97aef26bc5ae134ea206ab27467c3a30828e) 
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  $item = array();
95
96  if($type == 'd'){
97    return false; //no handling yet
98  }
99
100  //only search txt files
101  if(substr($file,-4) != '.txt') return false;
102
103  //get id
104  $id = pathID($file);
105
106  //check if it matches the query
107  if(!preg_match('/^'.preg_quote($opts['query'],'/').'/u',$id)){
108    return false;
109  }
110
111  //check ACL
112  if(auth_quickaclcheck($id) < AUTH_READ){
113    return false;
114  }
115
116  $data[]=array( 'id'    => $id,
117                 'type'  => $type,
118                 'level' => 1,
119                 'open'  => true);
120  return true;
121}
122
123/**
124 * Build the browsable index of pages
125 *
126 * $opts['ns'] is the current namespace
127 *
128 * @author  Andreas Gohr <andi@splitbrain.org>
129 */
130function search_index(&$data,$base,$file,$type,$lvl,$opts){
131  global $conf;
132  $return = true;
133
134  $item = array();
135
136  if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
137    //add but don't recurse
138    $return = false;
139  }elseif($type == 'f' && ($opts['nofiles'] || substr($file,-4) != '.txt')){
140    //don't add
141    return false;
142  }
143
144  $id = pathID($file);
145
146  if($type=='d' && $conf['sneaky_index'] && auth_quickaclcheck($id.':') < AUTH_READ){
147    return false;
148  }
149
150  //check hidden
151  if(isHiddenPage($id)){
152    return false;
153  }
154
155  //check ACL
156  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
157    return false;
158  }
159
160  $data[]=array( 'id'    => $id,
161                 'type'  => $type,
162                 'level' => $lvl,
163                 'open'  => $return );
164  return $return;
165}
166
167/**
168 * List all namespaces
169 *
170 * @author  Andreas Gohr <andi@splitbrain.org>
171 */
172function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
173  if($type == 'f') return true; //nothing to do on files
174
175  $id = pathID($file);
176  $data[]=array( 'id'    => $id,
177                 'type'  => $type,
178                 'level' => $lvl );
179  return true;
180}
181
182/**
183 * List all mediafiles in a namespace
184 *
185 * @author  Andreas Gohr <andi@splitbrain.org>
186 */
187function search_media(&$data,$base,$file,$type,$lvl,$opts){
188  //we do nothing with directories
189  if($type == 'd') {
190    return ($opts['recursive']);
191  }
192
193  $info         = array();
194  $info['id']   = pathID($file,true);
195  if($info['id'] != cleanID($info['id'])){
196    if($opts['showmsg'])
197      msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
198    return false; // skip non-valid files
199  }
200
201  //check ACL for namespace (we have no ACL for mediafiles)
202  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
203    return false;
204  }
205
206  $info['file'] = basename($file);
207  $info['size'] = filesize($base.'/'.$file);
208  $info['mtime'] = filemtime($base.'/'.$file);
209  $info['writable'] = is_writable($base.'/'.$file);
210  if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
211    $info['isimg'] = true;
212    require_once(DOKU_INC.'inc/JpegMeta.php');
213    $info['meta']  = new JpegMeta($base.'/'.$file);
214  }else{
215    $info['isimg'] = false;
216  }
217  $data[] = $info;
218
219  return false;
220}
221
222/**
223 * This function just lists documents (for RSS namespace export)
224 *
225 * @author  Andreas Gohr <andi@splitbrain.org>
226 */
227function search_list(&$data,$base,$file,$type,$lvl,$opts){
228  //we do nothing with directories
229  if($type == 'd') return false;
230  //only search txt files
231  if(substr($file,-4) == '.txt'){
232    //check ACL
233    $id = pathID($file);
234    if(auth_quickaclcheck($id) < AUTH_READ){
235      return false;
236    }
237    $data[]['id'] = $id;
238  }
239  return false;
240}
241
242/**
243 * Quicksearch for searching matching pagenames
244 *
245 * $opts['query'] is the search query
246 *
247 * @author  Andreas Gohr <andi@splitbrain.org>
248 */
249function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
250  //we do nothing with directories
251  if($type == 'd') return true;
252  //only search txt files
253  if(substr($file,-4) != '.txt') return true;
254
255  //simple stringmatching
256  if (!empty($opts['query'])){
257    if(strpos($file,$opts['query']) !== false){
258      //check ACL
259      $id = pathID($file);
260      if(auth_quickaclcheck($id) < AUTH_READ){
261        return false;
262      }
263      $data[]['id'] = $id;
264    }
265  }
266  return true;
267}
268
269/**
270 * Just lists all documents
271 *
272 * $opts['depth']  recursion level, 0 for all
273 * $opts['hash']   do md5 sum of content?
274 *
275 * @author  Andreas Gohr <andi@splitbrain.org>
276 */
277function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
278    //we do nothing with directories
279    if($type == 'd'){
280        if(!$opts['depth']) return true; // recurse forever
281        $parts = explode('/',$file);
282        if(count($parts) == $opts['depth']) return false; // depth reached
283        return true;
284    }
285
286    //only search txt files
287    if(substr($file,-4) != '.txt') return true;
288
289    $item['id']   = pathID($file);
290    if(!$opts['skipacl'] && auth_quickaclcheck($id) < AUTH_READ){
291      return false;
292    }
293
294    $item['rev']  = filemtime($base.'/'.$file);
295    $item['size'] = filesize($base.'/'.$file);
296    if($opts['hash']){
297        $item['hash'] = md5(trim(rawWiki($item['id'])));
298    }
299
300    $data[] = $item;
301    return true;
302}
303
304/**
305 * Search for backlinks to a given page
306 *
307 * $opts['ns']    namespace of the page
308 * $opts['name']  name of the page without namespace
309 *
310 * @author  Andreas Gohr <andi@splitbrain.org>
311 * @deprecated Replaced by ft_backlinks()
312 */
313function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
314  //we do nothing with directories
315  if($type == 'd') return true;
316  //only search txt files
317  if(substr($file,-4) != '.txt') return true;
318
319  //absolute search id
320  $sid = cleanID($opts['ns'].':'.$opts['name']);
321
322  //current id and namespace
323  $cid = pathID($file);
324  $cns = getNS($cid);
325
326  //check ACL
327  if(auth_quickaclcheck($cid) < AUTH_READ){
328    return false;
329  }
330
331  //fetch instructions
332  require_once(DOKU_INC.'inc/parserutils.php');
333  $instructions = p_cached_instructions($base.$file,true);
334  if(is_null($instructions)) return false;
335
336  //check all links for match
337  foreach($instructions as $ins){
338    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
339      $mid = $ins[1][0];
340      resolve_pageid($cns,$mid,$exists); //exists is not used
341      if($mid == $sid){
342        //we have a match - finish
343        $data[]['id'] = $cid;
344        break;
345      }
346    }
347  }
348
349  return false;
350}
351
352/**
353 * Fulltextsearch
354 *
355 * $opts['query'] is the search query
356 *
357 * @author  Andreas Gohr <andi@splitbrain.org>
358 * @deprecated - fulltext indexer is used instead
359 */
360function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
361  //we do nothing with directories
362  if($type == 'd') return true;
363  //only search txt files
364  if(substr($file,-4) != '.txt') return true;
365
366  //check ACL
367  $id = pathID($file);
368  if(auth_quickaclcheck($id) < AUTH_READ){
369    return false;
370  }
371
372  //create regexp from queries
373  $poswords = array();
374  $negwords = array();
375  $qpreg = preg_split('/\s+/',$opts['query']);
376
377  foreach($qpreg as $word){
378    switch(substr($word,0,1)){
379      case '-':
380        if(strlen($word) > 1){  // catch single '-'
381          array_push($negwords,preg_quote(substr($word,1),'#'));
382        }
383        break;
384      case '+':
385        if(strlen($word) > 1){  // catch single '+'
386          array_push($poswords,preg_quote(substr($word,1),'#'));
387        }
388        break;
389      default:
390        array_push($poswords,preg_quote($word,'#'));
391        break;
392    }
393  }
394
395  // a search without any posword is useless
396  if (!count($poswords)) return true;
397
398  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
399  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
400  search_regex($data,$base,$file,$reg,$poswords);
401  return true;
402}
403
404/**
405 * Reference search
406 * This fuction searches for existing references to a given media file
407 * and returns an array with the found pages. It doesn't pay any
408 * attention to ACL permissions to find every reference. The caller
409 * must check if the user has the appropriate rights to see the found
410 * page and eventually have to prevent the result from displaying.
411 *
412 * @param array  $data Reference to the result data structure
413 * @param string $base Base usually $conf['datadir']
414 * @param string $file current file or directory relative to $base
415 * @param char   $type Type either 'd' for directory or 'f' for file
416 * @param int    $lvl  Current recursion depht
417 * @param mixed  $opts option array as given to search()
418 *
419 * $opts['query'] is the demanded media file name
420 *
421 * @author  Andreas Gohr <andi@splitbrain.org>
422 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
423 */
424function search_reference(&$data,$base,$file,$type,$lvl,$opts){
425  global $conf;
426
427  //we do nothing with directories
428  if($type == 'd') return true;
429
430  //only search txt files
431  if(substr($file,-4) != '.txt') return true;
432
433  //we finish after 'cnt' references found. The return value
434  //'false' will skip subdirectories to speed search up.
435  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
436  if(count($data) >= $cnt) return false;
437
438  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
439  search_regex($data,$base,$file,$reg,array($opts['query']));
440  return true;
441}
442
443/* ------------- helper functions below -------------- */
444
445/**
446 * fulltext search helper
447 * searches a text file with a given regular expression
448 * no ACL checks are performed. This have to be done by
449 * the caller if necessary.
450 *
451 * @param array  $data  reference to array for results
452 * @param string $base  base directory
453 * @param string $file  file name to search in
454 * @param string $reg   regular expression to search for
455 * @param array  $words words that should be marked in the results
456 *
457 * @author  Andreas Gohr <andi@splitbrain.org>
458 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
459 *
460 * @deprecated - fulltext indexer is used instead
461 */
462function search_regex(&$data,$base,$file,$reg,$words){
463
464  //get text
465  $text = io_readfile($base.'/'.$file);
466  //lowercase text (u modifier does not help with case)
467  $lctext = utf8_strtolower($text);
468
469  //do the fulltext search
470  $matches = array();
471  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
472    //this is not the best way for snippet generation but the fastest I could find
473    $q = $words[0];  //use first word for snippet creation
474    $p = utf8_strpos($lctext,$q);
475    $f = $p - 100;
476    $l = utf8_strlen($q) + 200;
477    if($f < 0) $f = 0;
478    $snippet = '<span class="search_sep"> ... </span>'.
479               htmlspecialchars(utf8_substr($text,$f,$l)).
480               '<span class="search_sep"> ... </span>';
481    $mark    = '('.join('|', $words).')';
482    $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
483
484    $data[] = array(
485      'id'       => pathID($file),
486      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
487      'poswords' => join(' ',$words),
488      'snippet'  => $snippet,
489    );
490  }
491
492  return true;
493}
494
495
496/**
497 * fulltext sort
498 *
499 * Callback sort function for use with usort to sort the data
500 * structure created by search_fulltext. Sorts descending by count
501 *
502 * @author  Andreas Gohr <andi@splitbrain.org>
503 */
504function sort_search_fulltext($a,$b){
505  if($a['count'] > $b['count']){
506    return -1;
507  }elseif($a['count'] < $b['count']){
508    return 1;
509  }else{
510    return strcmp($a['id'],$b['id']);
511  }
512}
513
514/**
515 * translates a document path to an ID
516 *
517 * @author  Andreas Gohr <andi@splitbrain.org>
518 * @todo    move to pageutils
519 */
520function pathID($path,$keeptxt=false){
521  $id = utf8_decodeFN($path);
522  $id = str_replace('/',':',$id);
523  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
524  $id = preg_replace('#^:+#','',$id);
525  $id = preg_replace('#:+$#','',$id);
526  return $id;
527}
528
529
530//Setup VIM: ex: et ts=2 enc=utf-8 :
531