xref: /dokuwiki/inc/search.php (revision a8422a927454df870e8dd61ab6fcd6e2d1391aa6) 
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  if(!defined('DOKU_INC')) define('DOKU_INC',realpath(dirname(__FILE__).'/../').'/');
10  require_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    }elseif(substr($file,-5) == '.lock'){
38      //skip lockfiles
39      continue;
40    }
41    $files[] = $dir.'/'.$file;
42  }
43  closedir($dh);
44  sort($files);
45  sort($dirs);
46
47  //give directories to userfunction then recurse
48  foreach($dirs as $dir){
49    if (search_callback($func,$data,$base,$dir,'d',$lvl,$opts)){
50      search($data,$base,$func,$opts,$dir,$lvl+1);
51    }
52  }
53  //now handle the files
54  foreach($files as $file){
55    search_callback($func,$data,$base,$file,'f',$lvl,$opts);
56  }
57}
58
59/**
60 * Used to run the a user callback
61 *
62 * Makes sure the $data array is passed by reference (unlike when using
63 * call_user_func())
64 *
65 * @todo If this can be generalized it may be useful elsewhere in the code
66 * @author Andreas Gohr <andi@splitbrain.org>
67 */
68function search_callback($func,&$data,$base,$file,$type,$lvl,$opts){
69  if(is_array($func)){
70    if(is_object($func[0])){
71      // instanciated object
72      return $func[0]->$func[1]($data,$base,$file,$type,$lvl,$opts);
73    }else{
74      // static call
75      $f = $func[0].'::'.$func[1];
76      return $f($data,$base,$file,$type,$lvl,$opts);
77    }
78  }
79  // simple function call
80  return $func($data,$base,$file,$type,$lvl,$opts);
81}
82
83/**
84 * The following functions are userfunctions to use with the search
85 * function above. This function is called for every found file or
86 * directory. When a directory is given to the function it has to
87 * decide if this directory should be traversed (true) or not (false)
88 * The function has to accept the following parameters:
89 *
90 * &$data - Reference to the result data structure
91 * $base  - Base usually $conf['datadir']
92 * $file  - current file or directory relative to $base
93 * $type  - Type either 'd' for directory or 'f' for file
94 * $lvl   - Current recursion depht
95 * $opts  - option array as given to search()
96 *
97 * return values for files are ignored
98 *
99 * All functions should check the ACL for document READ rights
100 * namespaces (directories) are NOT checked as this would break
101 * the recursion (You can have an nonreadable dir over a readable
102 * one deeper nested)
103 */
104
105/**
106 * Searches for pages beginning with the given query
107 *
108 * @author Andreas Gohr <andi@splitbrain.org>
109 */
110function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
111  $item = array();
112
113  if($type == 'd'){
114    return false; //no handling yet
115  }
116
117  //get id
118  $id = pathID($file);
119
120  //check if it matches the query
121  if(!preg_match('/^'.preg_quote($opts['query'],'/').'/u',$id)){
122    return false;
123  }
124
125  //check ACL
126  if(auth_quickaclcheck($id) < AUTH_READ){
127    return false;
128  }
129
130  $data[]=array( 'id'    => $id,
131                 'type'  => $type,
132                 'level' => 1,
133                 'open'  => true);
134  return true;
135}
136
137/**
138 * Build the browsable index of pages
139 *
140 * $opts['ns'] is the current namespace
141 *
142 * @author  Andreas Gohr <andi@splitbrain.org>
143 */
144function search_index(&$data,$base,$file,$type,$lvl,$opts){
145  global $conf;
146  $return = true;
147
148  $item = array();
149
150  if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
151    //add but don't recurse
152    $return = false;
153  }elseif($type == 'f' && ($opts['nofiles'] || !preg_match('#\.txt$#',$file))){
154    //don't add
155    return false;
156  }
157
158  $id = pathID($file);
159
160  if($type=='d' && $conf['sneaky_index'] && auth_quickaclcheck($id.':') < AUTH_READ){
161    return false;
162  }
163
164  //check hidden
165  if(isHiddenPage($id)){
166    return false;
167  }
168
169  //check ACL
170  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
171    return false;
172  }
173
174  $data[]=array( 'id'    => $id,
175                 'type'  => $type,
176                 'level' => $lvl,
177                 'open'  => $return );
178  return $return;
179}
180
181/**
182 * List all namespaces
183 *
184 * @author  Andreas Gohr <andi@splitbrain.org>
185 */
186function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
187  if($type == 'f') return true; //nothing to do on files
188
189  $id = pathID($file);
190  $data[]=array( 'id'    => $id,
191                 'type'  => $type,
192                 'level' => $lvl );
193  return true;
194}
195
196/**
197 * List all mediafiles in a namespace
198 *
199 * @author  Andreas Gohr <andi@splitbrain.org>
200 */
201function search_media(&$data,$base,$file,$type,$lvl,$opts){
202  //we do nothing with directories
203  if($type == 'd') return false;
204
205  $info         = array();
206  $info['id']   = pathID($file,true);
207
208  //check ACL for namespace (we have no ACL for mediafiles)
209  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
210    return false;
211  }
212
213  $info['file'] = basename($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    require_once(DOKU_INC.'inc/JpegMeta.php');
220    $info['meta']  = new JpegMeta($base.'/'.$file);
221  }else{
222    $info['isimg'] = false;
223  }
224  $data[] = $info;
225
226  return false;
227}
228
229/**
230 * This function just lists documents (for RSS namespace export)
231 *
232 * @author  Andreas Gohr <andi@splitbrain.org>
233 */
234function search_list(&$data,$base,$file,$type,$lvl,$opts){
235  //we do nothing with directories
236  if($type == 'd') return false;
237  if(preg_match('#\.txt$#',$file)){
238    //check ACL
239    $id = pathID($file);
240    if(auth_quickaclcheck($id) < AUTH_READ){
241      return false;
242    }
243    $data[]['id'] = $id;;
244  }
245  return false;
246}
247
248/**
249 * Quicksearch for searching matching pagenames
250 *
251 * $opts['query'] is the search query
252 *
253 * @author  Andreas Gohr <andi@splitbrain.org>
254 */
255function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
256  //we do nothing with directories
257  if($type == 'd') return true;
258  //only search txt files
259  if(!preg_match('#\.txt$#',$file)) return true;
260
261  //simple stringmatching
262  if (!empty($opts['query'])){
263    if(strpos($file,$opts['query']) !== false){
264      //check ACL
265      $id = pathID($file);
266      if(auth_quickaclcheck($id) < AUTH_READ){
267        return false;
268      }
269      $data[]['id'] = $id;
270    }
271  }
272  return true;
273}
274
275/**
276 * Just lists all documents
277 *
278 * @author  Andreas Gohr <andi@splitbrain.org>
279 */
280function search_allpages(&$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(!preg_match('#\.txt$#',$file)) return true;
285
286  $data[]['id'] = pathID($file);
287  return true;
288}
289
290/**
291 * Search for backlinks to a given page
292 *
293 * $opts['ns']    namespace of the page
294 * $opts['name']  name of the page without namespace
295 *
296 * @author  Andreas Gohr <andi@splitbrain.org>
297 * @deprecated Replaced by ft_backlinks()
298 */
299function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
300  //we do nothing with directories
301  if($type == 'd') return true;;
302  //only search txt files
303  if(!preg_match('#\.txt$#',$file)) return true;;
304
305  //absolute search id
306  $sid = cleanID($opts['ns'].':'.$opts['name']);
307
308  //current id and namespace
309  $cid = pathID($file);
310  $cns = getNS($cid);
311
312  //check ACL
313  if(auth_quickaclcheck($cid) < AUTH_READ){
314    return false;
315  }
316
317  //fetch instructions
318  require_once(DOKU_INC.'inc/parserutils.php');
319  $instructions = p_cached_instructions($base.$file,true);
320  if(is_null($instructions)) return false;
321
322  //check all links for match
323  foreach($instructions as $ins){
324    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
325      $mid = $ins[1][0];
326      resolve_pageid($cns,$mid,$exists); //exists is not used
327      if($mid == $sid){
328        //we have a match - finish
329        $data[]['id'] = $cid;
330        break;
331      }
332    }
333  }
334
335  return false;
336}
337
338/**
339 * Fulltextsearch
340 *
341 * $opts['query'] is the search query
342 *
343 * @author  Andreas Gohr <andi@splitbrain.org>
344 * @deprecated - fulltext indexer is used instead
345 */
346function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
347  //we do nothing with directories
348  if($type == 'd') return true;;
349  //only search txt files
350  if(!preg_match('#\.txt$#',$file)) return true;;
351
352  //check ACL
353  $id = pathID($file);
354  if(auth_quickaclcheck($id) < AUTH_READ){
355    return false;
356  }
357
358  //create regexp from queries
359  $poswords = array();
360  $negwords = array();
361  $qpreg = preg_split('/\s+/',$opts['query']);
362
363  foreach($qpreg as $word){
364    switch(substr($word,0,1)){
365      case '-':
366        if(strlen($word) > 1){  // catch single '-'
367          array_push($negwords,preg_quote(substr($word,1),'#'));
368        }
369        break;
370      case '+':
371        if(strlen($word) > 1){  // catch single '+'
372          array_push($poswords,preg_quote(substr($word,1),'#'));
373        }
374        break;
375      default:
376        array_push($poswords,preg_quote($word,'#'));
377        break;
378    }
379  }
380
381  // a search without any posword is useless
382  if (!count($poswords)) return true;
383
384  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
385  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
386  search_regex($data,$base,$file,$reg,$poswords);
387  return true;
388}
389
390/**
391 * Reference search
392 * This fuction searches for existing references to a given media file
393 * and returns an array with the found pages. It doesn't pay any
394 * attention to ACL permissions to find every reference. The caller
395 * must check if the user has the appropriate rights to see the found
396 * page and eventually have to prevent the result from displaying.
397 *
398 * @param array  $data Reference to the result data structure
399 * @param string $base Base usually $conf['datadir']
400 * @param string $file current file or directory relative to $base
401 * @param char   $type Type either 'd' for directory or 'f' for file
402 * @param int    $lvl  Current recursion depht
403 * @param mixed  $opts option array as given to search()
404 *
405 * $opts['query'] is the demanded media file name
406 *
407 * @author  Andreas Gohr <andi@splitbrain.org>
408 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
409 */
410function search_reference(&$data,$base,$file,$type,$lvl,$opts){
411  global $conf;
412
413  //we do nothing with directories
414  if($type == 'd') return true;
415
416  //only search txt files
417  if(!preg_match('#\.txt$#',$file)) return true;
418
419  //we finish after 'cnt' references found. The return value
420  //'false' will skip subdirectories to speed search up.
421  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
422  if(count($data) >= $cnt) return false;
423
424  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
425  search_regex($data,$base,$file,$reg,array($opts['query']));
426  return true;
427}
428
429/* ------------- helper functions below -------------- */
430
431/**
432 * fulltext search helper
433 * searches a text file with a given regular expression
434 * no ACL checks are performed. This have to be done by
435 * the caller if necessary.
436 *
437 * @param array  $data  reference to array for results
438 * @param string $base  base directory
439 * @param string $file  file name to search in
440 * @param string $reg   regular expression to search for
441 * @param array  $words words that should be marked in the results
442 *
443 * @author  Andreas Gohr <andi@splitbrain.org>
444 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
445 *
446 * @deprecated - fulltext indexer is used instead
447 */
448function search_regex(&$data,$base,$file,$reg,$words){
449
450  //get text
451  $text = io_readfile($base.'/'.$file);
452  //lowercase text (u modifier does not help with case)
453  $lctext = utf8_strtolower($text);
454
455  //do the fulltext search
456  $matches = array();
457  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
458    //this is not the best way for snippet generation but the fastest I could find
459    $q = $words[0];  //use first word for snippet creation
460    $p = utf8_strpos($lctext,$q);
461    $f = $p - 100;
462    $l = utf8_strlen($q) + 200;
463    if($f < 0) $f = 0;
464    $snippet = '<span class="search_sep"> ... </span>'.
465               htmlspecialchars(utf8_substr($text,$f,$l)).
466               '<span class="search_sep"> ... </span>';
467    $mark    = '('.join('|', $words).')';
468    $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
469
470    $data[] = array(
471      'id'       => pathID($file),
472      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
473      'poswords' => join(' ',$words),
474      'snippet'  => $snippet,
475    );
476  }
477
478  return true;
479}
480
481
482/**
483 * fulltext sort
484 *
485 * Callback sort function for use with usort to sort the data
486 * structure created by search_fulltext. Sorts descending by count
487 *
488 * @author  Andreas Gohr <andi@splitbrain.org>
489 */
490function sort_search_fulltext($a,$b){
491  if($a['count'] > $b['count']){
492    return -1;
493  }elseif($a['count'] < $b['count']){
494    return 1;
495  }else{
496    return strcmp($a['id'],$b['id']);
497  }
498}
499
500/**
501 * translates a document path to an ID
502 *
503 * @author  Andreas Gohr <andi@splitbrain.org>
504 * @todo    move to pageutils
505 */
506function pathID($path,$keeptxt=false){
507  $id = utf8_decodeFN($path);
508  $id = str_replace('/',':',$id);
509  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
510  $id = preg_replace('#^:+#','',$id);
511  $id = preg_replace('#:+$#','',$id);
512  return $id;
513}
514
515
516//Setup VIM: ex: et ts=2 enc=utf-8 :
517