xref: /dokuwiki/inc/search.php (revision 357e220245cc608a5421bebf402bd083991714ab) 
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,$opts){
188
189    //we do nothing with directories
190    if($type == 'd') {
191        if(!$opts['depth']) return true; // recurse forever
192        $depth = substr_count($file,'/');
193        if($depth >= $opts['depth']) return false; // depth reached
194        return true;
195    }
196
197    $info         = array();
198    $info['id']   = pathID($file,true);
199    if($info['id'] != cleanID($info['id'])){
200        if($opts['showmsg'])
201            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
202        return false; // skip non-valid files
203    }
204
205    //check ACL for namespace (we have no ACL for mediafiles)
206    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
207    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
208        return false;
209    }
210
211    //check pattern filter
212    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
213        return false;
214    }
215
216    $info['file']     = basename($file);
217    $info['size']     = filesize($base.'/'.$file);
218    $info['mtime']    = filemtime($base.'/'.$file);
219    $info['writable'] = is_writable($base.'/'.$file);
220    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
221        $info['isimg'] = true;
222        require_once(DOKU_INC.'inc/JpegMeta.php');
223        $info['meta']  = new JpegMeta($base.'/'.$file);
224    }else{
225        $info['isimg'] = false;
226    }
227    if($opts['hash']){
228        $info['hash'] = md5(io_readFile(wikiFN($info['id']),false));
229    }
230
231    $data[] = $info;
232
233    return false;
234}
235
236/**
237 * This function just lists documents (for RSS namespace export)
238 *
239 * @author  Andreas Gohr <andi@splitbrain.org>
240 */
241function search_list(&$data,$base,$file,$type,$lvl,$opts){
242  //we do nothing with directories
243  if($type == 'd') return false;
244  //only search txt files
245  if(substr($file,-4) == '.txt'){
246    //check ACL
247    $id = pathID($file);
248    if(auth_quickaclcheck($id) < AUTH_READ){
249      return false;
250    }
251    $data[]['id'] = $id;
252  }
253  return false;
254}
255
256/**
257 * Quicksearch for searching matching pagenames
258 *
259 * $opts['query'] is the search query
260 *
261 * @author  Andreas Gohr <andi@splitbrain.org>
262 */
263function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
264  //we do nothing with directories
265  if($type == 'd') return true;
266  //only search txt files
267  if(substr($file,-4) != '.txt') return true;
268
269  //simple stringmatching
270  if (!empty($opts['query'])){
271    if(strpos($file,$opts['query']) !== false){
272      //check ACL
273      $id = pathID($file);
274      if(auth_quickaclcheck($id) < AUTH_READ){
275        return false;
276      }
277      $data[]['id'] = $id;
278    }
279  }
280  return true;
281}
282
283/**
284 * Just lists all documents
285 *
286 * $opts['depth']   recursion level, 0 for all
287 * $opts['hash']    do md5 sum of content?
288 * $opts['skipacl'] list everything regardless of ACL
289 *
290 * @author  Andreas Gohr <andi@splitbrain.org>
291 */
292function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
293    //we do nothing with directories
294    if($type == 'd'){
295        if(!$opts['depth']) return true; // recurse forever
296        $parts = explode('/',ltrim($file,'/'));
297        if(count($parts) == $opts['depth']) return false; // depth reached
298        return true;
299    }
300
301    //only search txt files
302    if(substr($file,-4) != '.txt') return true;
303
304    $item['id']   = pathID($file);
305    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
306      return false;
307    }
308
309    $item['rev']   = filemtime($base.'/'.$file);
310    $item['mtime'] = $item['rev'];
311    $item['size']  = filesize($base.'/'.$file);
312    if($opts['hash']){
313        $item['hash'] = md5(trim(rawWiki($item['id'])));
314    }
315
316    $data[] = $item;
317    return true;
318}
319
320/**
321 * Search for backlinks to a given page
322 *
323 * $opts['ns']    namespace of the page
324 * $opts['name']  name of the page without namespace
325 *
326 * @author  Andreas Gohr <andi@splitbrain.org>
327 * @deprecated Replaced by ft_backlinks()
328 */
329function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
330  //we do nothing with directories
331  if($type == 'd') return true;
332  //only search txt files
333  if(substr($file,-4) != '.txt') return true;
334
335  //absolute search id
336  $sid = cleanID($opts['ns'].':'.$opts['name']);
337
338  //current id and namespace
339  $cid = pathID($file);
340  $cns = getNS($cid);
341
342  //check ACL
343  if(auth_quickaclcheck($cid) < AUTH_READ){
344    return false;
345  }
346
347  //fetch instructions
348  require_once(DOKU_INC.'inc/parserutils.php');
349  $instructions = p_cached_instructions($base.$file,true);
350  if(is_null($instructions)) return false;
351
352  //check all links for match
353  foreach($instructions as $ins){
354    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
355      $mid = $ins[1][0];
356      resolve_pageid($cns,$mid,$exists); //exists is not used
357      if($mid == $sid){
358        //we have a match - finish
359        $data[]['id'] = $cid;
360        break;
361      }
362    }
363  }
364
365  return false;
366}
367
368/**
369 * Fulltextsearch
370 *
371 * $opts['query'] is the search query
372 *
373 * @author  Andreas Gohr <andi@splitbrain.org>
374 * @deprecated - fulltext indexer is used instead
375 */
376function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
377  //we do nothing with directories
378  if($type == 'd') return true;
379  //only search txt files
380  if(substr($file,-4) != '.txt') return true;
381
382  //check ACL
383  $id = pathID($file);
384  if(auth_quickaclcheck($id) < AUTH_READ){
385    return false;
386  }
387
388  //create regexp from queries
389  $poswords = array();
390  $negwords = array();
391  $qpreg = preg_split('/\s+/',$opts['query']);
392
393  foreach($qpreg as $word){
394    switch(substr($word,0,1)){
395      case '-':
396        if(strlen($word) > 1){  // catch single '-'
397          array_push($negwords,preg_quote(substr($word,1),'#'));
398        }
399        break;
400      case '+':
401        if(strlen($word) > 1){  // catch single '+'
402          array_push($poswords,preg_quote(substr($word,1),'#'));
403        }
404        break;
405      default:
406        array_push($poswords,preg_quote($word,'#'));
407        break;
408    }
409  }
410
411  // a search without any posword is useless
412  if (!count($poswords)) return true;
413
414  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
415  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
416  search_regex($data,$base,$file,$reg,$poswords);
417  return true;
418}
419
420/**
421 * Reference search
422 * This fuction searches for existing references to a given media file
423 * and returns an array with the found pages. It doesn't pay any
424 * attention to ACL permissions to find every reference. The caller
425 * must check if the user has the appropriate rights to see the found
426 * page and eventually have to prevent the result from displaying.
427 *
428 * @param array  $data Reference to the result data structure
429 * @param string $base Base usually $conf['datadir']
430 * @param string $file current file or directory relative to $base
431 * @param char   $type Type either 'd' for directory or 'f' for file
432 * @param int    $lvl  Current recursion depht
433 * @param mixed  $opts option array as given to search()
434 *
435 * $opts['query'] is the demanded media file name
436 *
437 * @author  Andreas Gohr <andi@splitbrain.org>
438 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
439 */
440function search_reference(&$data,$base,$file,$type,$lvl,$opts){
441  global $conf;
442
443  //we do nothing with directories
444  if($type == 'd') return true;
445
446  //only search txt files
447  if(substr($file,-4) != '.txt') return true;
448
449  //we finish after 'cnt' references found. The return value
450  //'false' will skip subdirectories to speed search up.
451  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
452  if(count($data) >= $cnt) return false;
453
454  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
455  search_regex($data,$base,$file,$reg,array($opts['query']));
456  return true;
457}
458
459/* ------------- helper functions below -------------- */
460
461/**
462 * fulltext search helper
463 * searches a text file with a given regular expression
464 * no ACL checks are performed. This have to be done by
465 * the caller if necessary.
466 *
467 * @param array  $data  reference to array for results
468 * @param string $base  base directory
469 * @param string $file  file name to search in
470 * @param string $reg   regular expression to search for
471 * @param array  $words words that should be marked in the results
472 *
473 * @author  Andreas Gohr <andi@splitbrain.org>
474 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
475 *
476 * @deprecated - fulltext indexer is used instead
477 */
478function search_regex(&$data,$base,$file,$reg,$words){
479
480  //get text
481  $text = io_readfile($base.'/'.$file);
482  //lowercase text (u modifier does not help with case)
483  $lctext = utf8_strtolower($text);
484
485  //do the fulltext search
486  $matches = array();
487  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
488    //this is not the best way for snippet generation but the fastest I could find
489    $q = $words[0];  //use first word for snippet creation
490    $p = utf8_strpos($lctext,$q);
491    $f = $p - 100;
492    $l = utf8_strlen($q) + 200;
493    if($f < 0) $f = 0;
494    $snippet = '<span class="search_sep"> ... </span>'.
495               htmlspecialchars(utf8_substr($text,$f,$l)).
496               '<span class="search_sep"> ... </span>';
497    $mark    = '('.join('|', $words).')';
498    $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
499
500    $data[] = array(
501      'id'       => pathID($file),
502      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
503      'poswords' => join(' ',$words),
504      'snippet'  => $snippet,
505    );
506  }
507
508  return true;
509}
510
511
512/**
513 * fulltext sort
514 *
515 * Callback sort function for use with usort to sort the data
516 * structure created by search_fulltext. Sorts descending by count
517 *
518 * @author  Andreas Gohr <andi@splitbrain.org>
519 */
520function sort_search_fulltext($a,$b){
521  if($a['count'] > $b['count']){
522    return -1;
523  }elseif($a['count'] < $b['count']){
524    return 1;
525  }else{
526    return strcmp($a['id'],$b['id']);
527  }
528}
529
530/**
531 * translates a document path to an ID
532 *
533 * @author  Andreas Gohr <andi@splitbrain.org>
534 * @todo    move to pageutils
535 */
536function pathID($path,$keeptxt=false){
537  $id = utf8_decodeFN($path);
538  $id = str_replace('/',':',$id);
539  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
540  $id = preg_replace('#^:+#','',$id);
541  $id = preg_replace('#:+$#','',$id);
542  return $id;
543}
544
545
546//Setup VIM: ex: et ts=2 enc=utf-8 :
547