xref: /dokuwiki/inc/search.php (revision 20ce88839c3cd1579ecba9654bb67dc8f8c8f967) 
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        if(!$opts['depth']) return true; // recurse forever
191        $parts = explode('/',ltrim($file,'/'));
192        if(count($parts) == $opts['depth']) return false; // depth reached
193        return true;
194    }
195
196    $info         = array();
197    $info['id']   = pathID($file,true);
198    if($info['id'] != cleanID($info['id'])){
199        if($opts['showmsg'])
200            msg(hsc($info['id']).' is not a valid file name for DokuWiki - skipped',-1);
201        return false; // skip non-valid files
202    }
203
204    //check ACL for namespace (we have no ACL for mediafiles)
205    $info['perm'] = auth_quickaclcheck(getNS($info['id']).':*');
206    if(!$opts['skipacl'] && $info['perm'] < AUTH_READ){
207        return false;
208    }
209
210    //check pattern filter
211    if($opts['pattern'] && !@preg_match($opts['pattern'], $info['id'])){
212        return false;
213    }
214
215    $info['file']     = basename($file);
216    $info['size']     = filesize($base.'/'.$file);
217    $info['mtime']    = filemtime($base.'/'.$file);
218    $info['writable'] = is_writable($base.'/'.$file);
219    if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
220        $info['isimg'] = true;
221        require_once(DOKU_INC.'inc/JpegMeta.php');
222        $info['meta']  = new JpegMeta($base.'/'.$file);
223    }else{
224        $info['isimg'] = false;
225    }
226    if($opts['hash']){
227        $info['hash'] = md5(io_readFile(wikiFN($info['id']),false));
228    }
229
230    $data[] = $info;
231
232    return false;
233}
234
235/**
236 * This function just lists documents (for RSS namespace export)
237 *
238 * @author  Andreas Gohr <andi@splitbrain.org>
239 */
240function search_list(&$data,$base,$file,$type,$lvl,$opts){
241  //we do nothing with directories
242  if($type == 'd') return false;
243  //only search txt files
244  if(substr($file,-4) == '.txt'){
245    //check ACL
246    $id = pathID($file);
247    if(auth_quickaclcheck($id) < AUTH_READ){
248      return false;
249    }
250    $data[]['id'] = $id;
251  }
252  return false;
253}
254
255/**
256 * Quicksearch for searching matching pagenames
257 *
258 * $opts['query'] is the search query
259 *
260 * @author  Andreas Gohr <andi@splitbrain.org>
261 */
262function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
263  //we do nothing with directories
264  if($type == 'd') return true;
265  //only search txt files
266  if(substr($file,-4) != '.txt') return true;
267
268  //simple stringmatching
269  if (!empty($opts['query'])){
270    if(strpos($file,$opts['query']) !== false){
271      //check ACL
272      $id = pathID($file);
273      if(auth_quickaclcheck($id) < AUTH_READ){
274        return false;
275      }
276      $data[]['id'] = $id;
277    }
278  }
279  return true;
280}
281
282/**
283 * Just lists all documents
284 *
285 * $opts['depth']   recursion level, 0 for all
286 * $opts['hash']    do md5 sum of content?
287 * $opts['skipacl'] list everything regardless of ACL
288 *
289 * @author  Andreas Gohr <andi@splitbrain.org>
290 */
291function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
292    //we do nothing with directories
293    if($type == 'd'){
294        if(!$opts['depth']) return true; // recurse forever
295        $parts = explode('/',ltrim($file,'/'));
296        if(count($parts) == $opts['depth']) return false; // depth reached
297        return true;
298    }
299
300    //only search txt files
301    if(substr($file,-4) != '.txt') return true;
302
303    $item['id']   = pathID($file);
304    if(!$opts['skipacl'] && auth_quickaclcheck($item['id']) < AUTH_READ){
305      return false;
306    }
307
308    $item['rev']   = filemtime($base.'/'.$file);
309    $item['mtime'] = $item['rev'];
310    $item['size']  = filesize($base.'/'.$file);
311    if($opts['hash']){
312        $item['hash'] = md5(trim(rawWiki($item['id'])));
313    }
314
315    $data[] = $item;
316    return true;
317}
318
319/**
320 * Search for backlinks to a given page
321 *
322 * $opts['ns']    namespace of the page
323 * $opts['name']  name of the page without namespace
324 *
325 * @author  Andreas Gohr <andi@splitbrain.org>
326 * @deprecated Replaced by ft_backlinks()
327 */
328function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
329  //we do nothing with directories
330  if($type == 'd') return true;
331  //only search txt files
332  if(substr($file,-4) != '.txt') return true;
333
334  //absolute search id
335  $sid = cleanID($opts['ns'].':'.$opts['name']);
336
337  //current id and namespace
338  $cid = pathID($file);
339  $cns = getNS($cid);
340
341  //check ACL
342  if(auth_quickaclcheck($cid) < AUTH_READ){
343    return false;
344  }
345
346  //fetch instructions
347  require_once(DOKU_INC.'inc/parserutils.php');
348  $instructions = p_cached_instructions($base.$file,true);
349  if(is_null($instructions)) return false;
350
351  //check all links for match
352  foreach($instructions as $ins){
353    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
354      $mid = $ins[1][0];
355      resolve_pageid($cns,$mid,$exists); //exists is not used
356      if($mid == $sid){
357        //we have a match - finish
358        $data[]['id'] = $cid;
359        break;
360      }
361    }
362  }
363
364  return false;
365}
366
367/**
368 * Fulltextsearch
369 *
370 * $opts['query'] is the search query
371 *
372 * @author  Andreas Gohr <andi@splitbrain.org>
373 * @deprecated - fulltext indexer is used instead
374 */
375function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
376  //we do nothing with directories
377  if($type == 'd') return true;
378  //only search txt files
379  if(substr($file,-4) != '.txt') return true;
380
381  //check ACL
382  $id = pathID($file);
383  if(auth_quickaclcheck($id) < AUTH_READ){
384    return false;
385  }
386
387  //create regexp from queries
388  $poswords = array();
389  $negwords = array();
390  $qpreg = preg_split('/\s+/',$opts['query']);
391
392  foreach($qpreg as $word){
393    switch(substr($word,0,1)){
394      case '-':
395        if(strlen($word) > 1){  // catch single '-'
396          array_push($negwords,preg_quote(substr($word,1),'#'));
397        }
398        break;
399      case '+':
400        if(strlen($word) > 1){  // catch single '+'
401          array_push($poswords,preg_quote(substr($word,1),'#'));
402        }
403        break;
404      default:
405        array_push($poswords,preg_quote($word,'#'));
406        break;
407    }
408  }
409
410  // a search without any posword is useless
411  if (!count($poswords)) return true;
412
413  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
414  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
415  search_regex($data,$base,$file,$reg,$poswords);
416  return true;
417}
418
419/**
420 * Reference search
421 * This fuction searches for existing references to a given media file
422 * and returns an array with the found pages. It doesn't pay any
423 * attention to ACL permissions to find every reference. The caller
424 * must check if the user has the appropriate rights to see the found
425 * page and eventually have to prevent the result from displaying.
426 *
427 * @param array  $data Reference to the result data structure
428 * @param string $base Base usually $conf['datadir']
429 * @param string $file current file or directory relative to $base
430 * @param char   $type Type either 'd' for directory or 'f' for file
431 * @param int    $lvl  Current recursion depht
432 * @param mixed  $opts option array as given to search()
433 *
434 * $opts['query'] is the demanded media file name
435 *
436 * @author  Andreas Gohr <andi@splitbrain.org>
437 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
438 */
439function search_reference(&$data,$base,$file,$type,$lvl,$opts){
440  global $conf;
441
442  //we do nothing with directories
443  if($type == 'd') return true;
444
445  //only search txt files
446  if(substr($file,-4) != '.txt') return true;
447
448  //we finish after 'cnt' references found. The return value
449  //'false' will skip subdirectories to speed search up.
450  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
451  if(count($data) >= $cnt) return false;
452
453  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
454  search_regex($data,$base,$file,$reg,array($opts['query']));
455  return true;
456}
457
458/* ------------- helper functions below -------------- */
459
460/**
461 * fulltext search helper
462 * searches a text file with a given regular expression
463 * no ACL checks are performed. This have to be done by
464 * the caller if necessary.
465 *
466 * @param array  $data  reference to array for results
467 * @param string $base  base directory
468 * @param string $file  file name to search in
469 * @param string $reg   regular expression to search for
470 * @param array  $words words that should be marked in the results
471 *
472 * @author  Andreas Gohr <andi@splitbrain.org>
473 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
474 *
475 * @deprecated - fulltext indexer is used instead
476 */
477function search_regex(&$data,$base,$file,$reg,$words){
478
479  //get text
480  $text = io_readfile($base.'/'.$file);
481  //lowercase text (u modifier does not help with case)
482  $lctext = utf8_strtolower($text);
483
484  //do the fulltext search
485  $matches = array();
486  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
487    //this is not the best way for snippet generation but the fastest I could find
488    $q = $words[0];  //use first word for snippet creation
489    $p = utf8_strpos($lctext,$q);
490    $f = $p - 100;
491    $l = utf8_strlen($q) + 200;
492    if($f < 0) $f = 0;
493    $snippet = '<span class="search_sep"> ... </span>'.
494               htmlspecialchars(utf8_substr($text,$f,$l)).
495               '<span class="search_sep"> ... </span>';
496    $mark    = '('.join('|', $words).')';
497    $snippet = preg_replace('#'.$mark.'#si','<strong class="search_hit">\\1</strong>',$snippet);
498
499    $data[] = array(
500      'id'       => pathID($file),
501      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
502      'poswords' => join(' ',$words),
503      'snippet'  => $snippet,
504    );
505  }
506
507  return true;
508}
509
510
511/**
512 * fulltext sort
513 *
514 * Callback sort function for use with usort to sort the data
515 * structure created by search_fulltext. Sorts descending by count
516 *
517 * @author  Andreas Gohr <andi@splitbrain.org>
518 */
519function sort_search_fulltext($a,$b){
520  if($a['count'] > $b['count']){
521    return -1;
522  }elseif($a['count'] < $b['count']){
523    return 1;
524  }else{
525    return strcmp($a['id'],$b['id']);
526  }
527}
528
529/**
530 * translates a document path to an ID
531 *
532 * @author  Andreas Gohr <andi@splitbrain.org>
533 * @todo    move to pageutils
534 */
535function pathID($path,$keeptxt=false){
536  $id = utf8_decodeFN($path);
537  $id = str_replace('/',':',$id);
538  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
539  $id = preg_replace('#^:+#','',$id);
540  $id = preg_replace('#:+$#','',$id);
541  return $id;
542}
543
544
545//Setup VIM: ex: et ts=2 enc=utf-8 :
546