xref: /dokuwiki/inc/search.php (revision 9e7ed8c421085cfaaecbf7b65648c79c78e996ad) 
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',fullpath(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    }
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'] = $id;
264    }
265  }
266  return true;
267}
268
269/**
270 * Just lists all documents
271 *
272 * @author  Andreas Gohr <andi@splitbrain.org>
273 */
274function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
275  //we do nothing with directories
276  if($type == 'd') return true;
277  //only search txt files
278  if(substr($file,-4) != '.txt') return true;
279
280  $data[]['id'] = pathID($file);
281  return true;
282}
283
284/**
285 * Search for backlinks to a given page
286 *
287 * $opts['ns']    namespace of the page
288 * $opts['name']  name of the page without namespace
289 *
290 * @author  Andreas Gohr <andi@splitbrain.org>
291 * @deprecated Replaced by ft_backlinks()
292 */
293function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
294  //we do nothing with directories
295  if($type == 'd') return true;
296  //only search txt files
297  if(substr($file,-4) != '.txt') return true;
298
299  //absolute search id
300  $sid = cleanID($opts['ns'].':'.$opts['name']);
301
302  //current id and namespace
303  $cid = pathID($file);
304  $cns = getNS($cid);
305
306  //check ACL
307  if(auth_quickaclcheck($cid) < AUTH_READ){
308    return false;
309  }
310
311  //fetch instructions
312  require_once(DOKU_INC.'inc/parserutils.php');
313  $instructions = p_cached_instructions($base.$file,true);
314  if(is_null($instructions)) return false;
315
316  //check all links for match
317  foreach($instructions as $ins){
318    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
319      $mid = $ins[1][0];
320      resolve_pageid($cns,$mid,$exists); //exists is not used
321      if($mid == $sid){
322        //we have a match - finish
323        $data[]['id'] = $cid;
324        break;
325      }
326    }
327  }
328
329  return false;
330}
331
332/**
333 * Fulltextsearch
334 *
335 * $opts['query'] is the search query
336 *
337 * @author  Andreas Gohr <andi@splitbrain.org>
338 * @deprecated - fulltext indexer is used instead
339 */
340function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
341  //we do nothing with directories
342  if($type == 'd') return true;
343  //only search txt files
344  if(substr($file,-4) != '.txt') return true;
345
346  //check ACL
347  $id = pathID($file);
348  if(auth_quickaclcheck($id) < AUTH_READ){
349    return false;
350  }
351
352  //create regexp from queries
353  $poswords = array();
354  $negwords = array();
355  $qpreg = preg_split('/\s+/',$opts['query']);
356
357  foreach($qpreg as $word){
358    switch(substr($word,0,1)){
359      case '-':
360        if(strlen($word) > 1){  // catch single '-'
361          array_push($negwords,preg_quote(substr($word,1),'#'));
362        }
363        break;
364      case '+':
365        if(strlen($word) > 1){  // catch single '+'
366          array_push($poswords,preg_quote(substr($word,1),'#'));
367        }
368        break;
369      default:
370        array_push($poswords,preg_quote($word,'#'));
371        break;
372    }
373  }
374
375  // a search without any posword is useless
376  if (!count($poswords)) return true;
377
378  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
379  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
380  search_regex($data,$base,$file,$reg,$poswords);
381  return true;
382}
383
384/**
385 * Reference search
386 * This fuction searches for existing references to a given media file
387 * and returns an array with the found pages. It doesn't pay any
388 * attention to ACL permissions to find every reference. The caller
389 * must check if the user has the appropriate rights to see the found
390 * page and eventually have to prevent the result from displaying.
391 *
392 * @param array  $data Reference to the result data structure
393 * @param string $base Base usually $conf['datadir']
394 * @param string $file current file or directory relative to $base
395 * @param char   $type Type either 'd' for directory or 'f' for file
396 * @param int    $lvl  Current recursion depht
397 * @param mixed  $opts option array as given to search()
398 *
399 * $opts['query'] is the demanded media file name
400 *
401 * @author  Andreas Gohr <andi@splitbrain.org>
402 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
403 */
404function search_reference(&$data,$base,$file,$type,$lvl,$opts){
405  global $conf;
406
407  //we do nothing with directories
408  if($type == 'd') return true;
409
410  //only search txt files
411  if(substr($file,-4) != '.txt') return true;
412
413  //we finish after 'cnt' references found. The return value
414  //'false' will skip subdirectories to speed search up.
415  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
416  if(count($data) >= $cnt) return false;
417
418  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
419  search_regex($data,$base,$file,$reg,array($opts['query']));
420  return true;
421}
422
423/* ------------- helper functions below -------------- */
424
425/**
426 * fulltext search helper
427 * searches a text file with a given regular expression
428 * no ACL checks are performed. This have to be done by
429 * the caller if necessary.
430 *
431 * @param array  $data  reference to array for results
432 * @param string $base  base directory
433 * @param string $file  file name to search in
434 * @param string $reg   regular expression to search for
435 * @param array  $words words that should be marked in the results
436 *
437 * @author  Andreas Gohr <andi@splitbrain.org>
438 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
439 *
440 * @deprecated - fulltext indexer is used instead
441 */
442function search_regex(&$data,$base,$file,$reg,$words){
443
444  //get text
445  $text = io_readfile($base.'/'.$file);
446  //lowercase text (u modifier does not help with case)
447  $lctext = utf8_strtolower($text);
448
449  //do the fulltext search
450  $matches = array();
451  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
452    //this is not the best way for snippet generation but the fastest I could find
453    $q = $words[0];  //use first word for snippet creation
454    $p = utf8_strpos($lctext,$q);
455    $f = $p - 100;
456    $l = utf8_strlen($q) + 200;
457    if($f < 0) $f = 0;
458    $snippet = '<span class="search_sep"> ... </span>'.
459               htmlspecialchars(utf8_substr($text,$f,$l)).
460               '<span class="search_sep"> ... </span>';
461    $mark    = '('.join('|', $words).')';
462    $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
463
464    $data[] = array(
465      'id'       => pathID($file),
466      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
467      'poswords' => join(' ',$words),
468      'snippet'  => $snippet,
469    );
470  }
471
472  return true;
473}
474
475
476/**
477 * fulltext sort
478 *
479 * Callback sort function for use with usort to sort the data
480 * structure created by search_fulltext. Sorts descending by count
481 *
482 * @author  Andreas Gohr <andi@splitbrain.org>
483 */
484function sort_search_fulltext($a,$b){
485  if($a['count'] > $b['count']){
486    return -1;
487  }elseif($a['count'] < $b['count']){
488    return 1;
489  }else{
490    return strcmp($a['id'],$b['id']);
491  }
492}
493
494/**
495 * translates a document path to an ID
496 *
497 * @author  Andreas Gohr <andi@splitbrain.org>
498 * @todo    move to pageutils
499 */
500function pathID($path,$keeptxt=false){
501  $id = utf8_decodeFN($path);
502  $id = str_replace('/',':',$id);
503  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
504  $id = preg_replace('#^:+#','',$id);
505  $id = preg_replace('#:+$#','',$id);
506  return $id;
507}
508
509
510//Setup VIM: ex: et ts=2 enc=utf-8 :
511