xref: /dokuwiki/inc/search.php (revision 4befca0c401cd5d13a1de3ac5e17477757a8c284) 
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 * @author  Andreas Gohr <andi@splitbrain.org>
19 */
20function search(&$data,$base,$func,$opts,$dir='',$lvl=1){
21  $dirs   = array();
22  $files  = array();
23
24  //read in directories and files
25  $dh = @opendir($base.'/'.$dir);
26  if(!$dh) return;
27  while(($file = readdir($dh)) !== false){
28    if(preg_match('/^[\._]/',$file)) continue; //skip hidden files and upper dirs
29    if(is_dir($base.'/'.$dir.'/'.$file)){
30      $dirs[] = $dir.'/'.$file;
31      continue;
32    }elseif(substr($file,-5) == '.lock'){
33      //skip lockfiles
34      continue;
35    }
36    $files[] = $dir.'/'.$file;
37  }
38  closedir($dh);
39  sort($files);
40  sort($dirs);
41
42  //give directories to userfunction then recurse
43  foreach($dirs as $dir){
44    if ($func($data,$base,$dir,'d',$lvl,$opts)){
45      search($data,$base,$func,$opts,$dir,$lvl+1);
46    }
47  }
48  //now handle the files
49  foreach($files as $file){
50    $func($data,$base,$file,'f',$lvl,$opts);
51  }
52}
53
54/**
55 * The following functions are userfunctions to use with the search
56 * function above. This function is called for every found file or
57 * directory. When a directory is given to the function it has to
58 * decide if this directory should be traversed (true) or not (false)
59 * The function has to accept the following parameters:
60 *
61 * &$data - Reference to the result data structure
62 * $base  - Base usually $conf['datadir']
63 * $file  - current file or directory relative to $base
64 * $type  - Type either 'd' for directory or 'f' for file
65 * $lvl   - Current recursion depht
66 * $opts  - option array as given to search()
67 *
68 * return values for files are ignored
69 *
70 * All functions should check the ACL for document READ rights
71 * namespaces (directories) are NOT checked as this would break
72 * the recursion (You can have an nonreadable dir over a readable
73 * one deeper nested)
74 */
75
76/**
77 * Searches for pages beginning with the given query
78 *
79 * @author Andreas Gohr <andi@splitbrain.org>
80 */
81function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
82  $item = array();
83
84  if($type == 'd'){
85    return false; //no handling yet
86  }
87
88  //get id
89  $id = pathID($file);
90
91  //check if it matches the query
92  if(!preg_match('/^'.preg_quote($opts['query'],'/').'/u',$id)){
93    return false;
94  }
95
96  //check ACL
97  if(auth_quickaclcheck($id) < AUTH_READ){
98    return false;
99  }
100
101  $data[]=array( 'id'    => $id,
102                 'type'  => $type,
103                 'level' => 1,
104                 'open'  => true);
105  return true;
106}
107
108/**
109 * Build the browsable index of pages
110 *
111 * $opts['ns'] is the current namespace
112 *
113 * @author  Andreas Gohr <andi@splitbrain.org>
114 */
115function search_index(&$data,$base,$file,$type,$lvl,$opts){
116  $return = true;
117
118  $item = array();
119
120  if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
121    //add but don't recurse
122    $return = false;
123  }elseif($type == 'f' && !preg_match('#\.txt$#',$file)){
124    //don't add
125    return false;
126  }
127
128  $id = pathID($file);
129
130  //check hidden
131  if($type=='f' && isHiddenPage($id)){
132    return false;
133  }
134
135  //check ACL
136  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
137    return false;
138  }
139
140  $data[]=array( 'id'    => $id,
141                 'type'  => $type,
142                 'level' => $lvl,
143                 'open'  => $return );
144  return $return;
145}
146
147/**
148 * List all namespaces
149 *
150 * @author  Andreas Gohr <andi@splitbrain.org>
151 */
152function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
153  if($type == 'f') return true; //nothing to do on files
154
155  $id = pathID($file);
156  $data[]=array( 'id'    => $id,
157                 'type'  => $type,
158                 'level' => $lvl );
159  return true;
160}
161
162/**
163 * List all mediafiles in a namespace
164 *
165 * @author  Andreas Gohr <andi@splitbrain.org>
166 */
167function search_media(&$data,$base,$file,$type,$lvl,$opts){
168  //we do nothing with directories
169  if($type == 'd') return false;
170
171  $info         = array();
172  $info['id']   = pathID($file,true);
173
174  //check ACL for namespace (we have no ACL for mediafiles)
175  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
176    return false;
177  }
178
179  $info['file'] = basename($file);
180  $info['size'] = filesize($base.'/'.$file);
181  if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
182    $info['isimg'] = true;
183    require_once(DOKU_INC.'inc/JpegMeta.php');
184    $info['meta']  = new JpegMeta($base.'/'.$file);
185  }else{
186    $info['isimg'] = false;
187  }
188  $data[] = $info;
189
190  return false;
191}
192
193/**
194 * This function just lists documents (for RSS namespace export)
195 *
196 * @author  Andreas Gohr <andi@splitbrain.org>
197 */
198function search_list(&$data,$base,$file,$type,$lvl,$opts){
199  //we do nothing with directories
200  if($type == 'd') return false;
201  if(preg_match('#\.txt$#',$file)){
202    //check ACL
203    $id = pathID($file);
204    if(auth_quickaclcheck($id) < AUTH_READ){
205      return false;
206    }
207    $data[]['id'] = $id;;
208  }
209  return false;
210}
211
212/**
213 * Quicksearch for searching matching pagenames
214 *
215 * $opts['query'] is the search query
216 *
217 * @author  Andreas Gohr <andi@splitbrain.org>
218 */
219function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
220  //we do nothing with directories
221  if($type == 'd') return true;
222  //only search txt files
223  if(!preg_match('#\.txt$#',$file)) return true;
224
225  //simple stringmatching
226  if (!empty($opts['query'])){
227    if(strpos($file,$opts['query']) !== false){
228      //check ACL
229      $id = pathID($file);
230      if(auth_quickaclcheck($id) < AUTH_READ){
231        return false;
232      }
233      $data[]['id'] = $id;
234    }
235  }
236  return true;
237}
238
239/**
240 * Just lists all documents
241 *
242 * @author  Andreas Gohr <andi@splitbrain.org>
243 */
244function search_allpages(&$data,$base,$file,$type,$lvl,$opts){
245  //we do nothing with directories
246  if($type == 'd') return true;
247  //only search txt files
248  if(!preg_match('#\.txt$#',$file)) return true;
249
250  $data[]['id'] = pathID($file);
251  return true;
252}
253
254/**
255 * Search for backlinks to a given page
256 *
257 * $opts['ns']    namespace of the page
258 * $opts['name']  name of the page without namespace
259 *
260 * @author  Andreas Gohr <andi@splitbrain.org>
261 * @deprecated Replaced by ft_backlinks()
262 */
263function search_backlinks(&$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(!preg_match('#\.txt$#',$file)) return true;;
268
269  //absolute search id
270  $sid = cleanID($opts['ns'].':'.$opts['name']);
271
272  //current id and namespace
273  $cid = pathID($file);
274  $cns = getNS($cid);
275
276  //check ACL
277  if(auth_quickaclcheck($cid) < AUTH_READ){
278    return false;
279  }
280
281  //fetch instructions
282  require_once(DOKU_INC.'inc/parserutils.php');
283  $instructions = p_cached_instructions($base.$file,true);
284  if(is_null($instructions)) return false;
285
286  //check all links for match
287  foreach($instructions as $ins){
288    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
289      $mid = $ins[1][0];
290      resolve_pageid($cns,$mid,$exists); //exists is not used
291      if($mid == $sid){
292        //we have a match - finish
293        $data[]['id'] = $cid;
294        break;
295      }
296    }
297  }
298
299  return false;
300}
301
302/**
303 * Fulltextsearch
304 *
305 * $opts['query'] is the search query
306 *
307 * @author  Andreas Gohr <andi@splitbrain.org>
308 * @deprecated - fulltext indexer is used instead
309 */
310function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
311  //we do nothing with directories
312  if($type == 'd') return true;;
313  //only search txt files
314  if(!preg_match('#\.txt$#',$file)) return true;;
315
316  //check ACL
317  $id = pathID($file);
318  if(auth_quickaclcheck($id) < AUTH_READ){
319    return false;
320  }
321
322  //create regexp from queries
323  $poswords = array();
324  $negwords = array();
325  $qpreg = preg_split('/\s+/',$opts['query']);
326
327  foreach($qpreg as $word){
328    switch(substr($word,0,1)){
329      case '-':
330        if(strlen($word) > 1){  // catch single '-'
331          array_push($negwords,preg_quote(substr($word,1),'#'));
332        }
333        break;
334      case '+':
335        if(strlen($word) > 1){  // catch single '+'
336          array_push($poswords,preg_quote(substr($word,1),'#'));
337        }
338        break;
339      default:
340        array_push($poswords,preg_quote($word,'#'));
341        break;
342    }
343  }
344
345  // a search without any posword is useless
346  if (!count($poswords)) return true;
347
348  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
349  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
350  search_regex($data,$base,$file,$reg,$poswords);
351  return true;
352}
353
354/**
355 * Reference search
356 * This fuction searches for existing references to a given media file
357 * and returns an array with the found pages. It doesn't pay any
358 * attention to ACL permissions to find every reference. The caller
359 * must check if the user has the appropriate rights to see the found
360 * page and eventually have to prevent the result from displaying.
361 *
362 * @param array  $data Reference to the result data structure
363 * @param string $base Base usually $conf['datadir']
364 * @param string $file current file or directory relative to $base
365 * @param char   $type Type either 'd' for directory or 'f' for file
366 * @param int    $lvl  Current recursion depht
367 * @param mixed  $opts option array as given to search()
368 *
369 * $opts['query'] is the demanded media file name
370 *
371 * @author  Andreas Gohr <andi@splitbrain.org>
372 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
373 */
374function search_reference(&$data,$base,$file,$type,$lvl,$opts){
375  global $conf;
376
377  //we do nothing with directories
378  if($type == 'd') return true;
379
380  //only search txt files
381  if(!preg_match('#\.txt$#',$file)) return true;
382
383  //we finish after 'cnt' references found. The return value
384  //'false' will skip subdirectories to speed search up.
385  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
386  if(count($data) >= $cnt) return false;
387
388  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
389  search_regex($data,$base,$file,$reg,array($opts['query']));
390  return true;
391}
392
393/* ------------- helper functions below -------------- */
394
395/**
396 * fulltext search helper
397 * searches a text file with a given regular expression
398 * no ACL checks are performed. This have to be done by
399 * the caller if necessary.
400 *
401 * @param array  $data  reference to array for results
402 * @param string $base  base directory
403 * @param string $file  file name to search in
404 * @param string $reg   regular expression to search for
405 * @param array  $words words that should be marked in the results
406 *
407 * @author  Andreas Gohr <andi@splitbrain.org>
408 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
409 *
410 * @deprecated - fulltext indexer is used instead
411 */
412function search_regex(&$data,$base,$file,$reg,$words){
413
414  //get text
415  $text = io_readfile($base.'/'.$file);
416  //lowercase text (u modifier does not help with case)
417  $lctext = utf8_strtolower($text);
418
419  //do the fulltext search
420  $matches = array();
421  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
422    //this is not the best way for snippet generation but the fastest I could find
423    $q = $words[0];  //use first word for snippet creation
424    $p = utf8_strpos($lctext,$q);
425    $f = $p - 100;
426    $l = utf8_strlen($q) + 200;
427    if($f < 0) $f = 0;
428    $snippet = '<span class="search_sep"> ... </span>'.
429               htmlspecialchars(utf8_substr($text,$f,$l)).
430               '<span class="search_sep"> ... </span>';
431    $mark    = '('.join('|', $words).')';
432    $snippet = preg_replace('#'.$mark.'#si','<span class="search_hit">\\1</span>',$snippet);
433
434    $data[] = array(
435      'id'       => pathID($file),
436      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
437      'poswords' => join(' ',$words),
438      'snippet'  => $snippet,
439    );
440  }
441
442  return true;
443}
444
445
446/**
447 * fulltext sort
448 *
449 * Callback sort function for use with usort to sort the data
450 * structure created by search_fulltext. Sorts descending by count
451 *
452 * @author  Andreas Gohr <andi@splitbrain.org>
453 */
454function sort_search_fulltext($a,$b){
455  if($a['count'] > $b['count']){
456    return -1;
457  }elseif($a['count'] < $b['count']){
458    return 1;
459  }else{
460    return strcmp($a['id'],$b['id']);
461  }
462}
463
464/**
465 * translates a document path to an ID
466 *
467 * @author  Andreas Gohr <andi@splitbrain.org>
468 * @todo    move to pageutils
469 */
470function pathID($path,$keeptxt=false){
471  $id = utf8_decodeFN($path);
472  $id = str_replace('/',':',$id);
473  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
474  $id = preg_replace('#^:+#','',$id);
475  $id = preg_replace('#:+$#','',$id);
476  return $id;
477}
478
479
480//Setup VIM: ex: et ts=2 enc=utf-8 :
481