xref: /dokuwiki/inc/search.php (revision d67ca2c0f7f6b75608386c226d33afdda872757c) 
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  //check ACL
129  $id = pathID($file);
130  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
131    return false;
132  }
133
134  $data[]=array( 'id'    => $id,
135                 'type'  => $type,
136                 'level' => $lvl,
137                 'open'  => $return );
138  return $return;
139}
140
141/**
142 * List all namespaces
143 *
144 * @author  Andreas Gohr <andi@splitbrain.org>
145 */
146function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
147  if($type == 'f') return true; //nothing to do on files
148
149  $id = pathID($file);
150  $data[]=array( 'id'    => $id,
151                 'type'  => $type,
152                 'level' => $lvl );
153  return true;
154}
155
156/**
157 * List all mediafiles in a namespace
158 *
159 * @author  Andreas Gohr <andi@splitbrain.org>
160 */
161function search_media(&$data,$base,$file,$type,$lvl,$opts){
162  //we do nothing with directories
163  if($type == 'd') return false;
164
165  $info         = array();
166  $info['id']   = pathID($file,true);
167
168  //check ACL for namespace (we have no ACL for mediafiles)
169  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
170    return false;
171  }
172
173  $info['file'] = basename($file);
174  $info['size'] = filesize($base.'/'.$file);
175  if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
176    $info['isimg'] = true;
177    $info['info']  = getimagesize($base.'/'.$file);
178  }else{
179    $info['isimg'] = false;
180  }
181  $data[] = $info;
182
183  return false;
184}
185
186/**
187 * This function just lists documents (for RSS namespace export)
188 *
189 * @author  Andreas Gohr <andi@splitbrain.org>
190 */
191function search_list(&$data,$base,$file,$type,$lvl,$opts){
192  //we do nothing with directories
193  if($type == 'd') return false;
194  if(preg_match('#\.txt$#',$file)){
195    //check ACL
196    $id = pathID($file);
197    if(auth_quickaclcheck($id) < AUTH_READ){
198      return false;
199    }
200    $data[]['id'] = $id;;
201  }
202  return false;
203}
204
205/**
206 * Quicksearch for searching matching pagenames
207 *
208 * $opts['query'] is the search query
209 *
210 * @author  Andreas Gohr <andi@splitbrain.org>
211 */
212function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
213  //we do nothing with directories
214  if($type == 'd') return true;
215  //only search txt files
216  if(!preg_match('#\.txt$#',$file)) return true;
217
218  //simple stringmatching
219  if(strpos($file,$opts['query']) !== false){
220    //check ACL
221    $id = pathID($file);
222    if(auth_quickaclcheck($id) < AUTH_READ){
223      return false;
224    }
225    $data[]['id'] = $id;
226  }
227
228  return true;
229}
230
231/**
232 * Search for backlinks to a given page
233 *
234 * $opts['ns']    namespace of the page
235 * $opts['name']  name of the page without namespace
236 *
237 * @author  Andreas Gohr <andi@splitbrain.org>
238 */
239function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
240  //we do nothing with directories
241  if($type == 'd') return true;;
242  //only search txt files
243  if(!preg_match('#\.txt$#',$file)) return true;;
244
245  //absolute search id
246  $sid = cleanID($opts['ns'].':'.$opts['name']);
247
248  //current id and namespace
249  $cid = pathID($file);
250  $cns = getNS($cid);
251
252  //check ACL
253  if(auth_quickaclcheck($cid) < AUTH_READ){
254    return false;
255  }
256
257  //fetch instructions
258  require_once(DOKU_INC.'inc/parserutils.php');
259  $instructions = p_cached_instructions($base.$file,true);
260  if(is_null($instructions)) return false;
261
262  //check all links for match
263  foreach($instructions as $ins){
264    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
265      $mid = $ins[1][0];
266      resolve_pageid($cns,$mid,$exists); //exists is not used
267      if($mid == $sid){
268        //we have a match - finish
269        $data[]['id'] = $cid;
270        break;
271      }
272    }
273  }
274
275  return false;
276}
277
278/**
279 * Fulltextsearch
280 *
281 * $opts['query'] is the search query
282 *
283 * @author  Andreas Gohr <andi@splitbrain.org>
284 */
285function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
286  //we do nothing with directories
287  if($type == 'd') return true;;
288  //only search txt files
289  if(!preg_match('#\.txt$#',$file)) return true;;
290
291  //check ACL
292  $id = pathID($file);
293  if(auth_quickaclcheck($id) < AUTH_READ){
294    return false;
295  }
296
297  //create regexp from queries
298  $poswords = array();
299  $negwords = array();
300  $qpreg = preg_split('/\s+/',$opts['query']);
301
302  foreach($qpreg as $word){
303    switch(substr($word,0,1)){
304      case '-':
305        array_push($negwords,preg_quote(substr($word,1),'#'));
306        break;
307      case '+':
308        array_push($poswords,preg_quote(substr($word,1),'#'));
309        break;
310      default:
311        array_push($poswords,preg_quote($word,'#'));
312        break;
313    }
314  }
315
316  // a search without any posword is useless
317  if (!count($poswords)) return true;
318
319  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
320  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
321  search_regex($data,$base,$file,$reg,$poswords);
322  return true;
323}
324
325/**
326 * Reference search
327 * This fuction searches for existing references to a given media file
328 * and returns an array with the found pages. It doesn't pay any
329 * attention to ACL permissions to find every reference. The caller
330 * must check if the user has the appropriate rights to see the found
331 * page and eventually have to prevent the result from displaying.
332 *
333 * @param array  $data Reference to the result data structure
334 * @param string $base Base usually $conf['datadir']
335 * @param string $file current file or directory relative to $base
336 * @param char   $type Type either 'd' for directory or 'f' for file
337 * @param int    $lvl  Current recursion depht
338 * @param mixed  $opts option array as given to search()
339 *
340 * $opts['query'] is the demanded media file name
341 *
342 * @author  Andreas Gohr <andi@splitbrain.org>
343 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
344 */
345function search_reference(&$data,$base,$file,$type,$lvl,$opts){
346  global $conf;
347
348  //we do nothing with directories
349  if($type == 'd') return true;
350
351  //only search txt files
352  if(!preg_match('#\.txt$#',$file)) return true;
353
354  //we finish after five references found. The return value
355  //'false' will skip subdirectories to speed search up.
356  if(count($data) >= $conf['refcount']) return false;
357
358  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
359  search_regex($data,$base,$file,$reg,array($opts['query']));
360  return true;
361}
362
363/* ------------- helper functions below -------------- */
364
365/**
366 * fulltext search helper
367 * searches a text file with a given regular expression
368 * no ACL checks are performed. This have to be done by
369 * the caller if necessary.
370 *
371 * @param array  $data  reference to array for results
372 * @param string $base  base directory
373 * @param string $file  file name to search in
374 * @param string $reg   regular expression to search for
375 * @param array  $words words that should be marked in the results
376 *
377 * @author  Andreas Gohr <andi@splitbrain.org>
378 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
379 */
380function search_regex(&$data,$base,$file,$reg,$words){
381
382  //get text
383  $text = io_readfile($base.'/'.$file);
384  //lowercase text (u modifier does not help with case)
385  $lctext = utf8_strtolower($text);
386
387  //do the fulltext search
388  $matches = array();
389  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
390    //this is not the best way for snippet generation but the fastest I could find
391    $q = $words[0];  //use first word for snippet creation
392    $p = utf8_strpos($lctext,$q);
393    $f = $p - 100;
394    $l = utf8_strlen($q) + 200;
395    if($f < 0) $f = 0;
396    $snippet = '<span class="search_sep"> ... </span>'.
397               htmlspecialchars(utf8_substr($text,$f,$l)).
398               '<span class="search_sep"> ... </span>';
399    $mark    = '('.join('|', $words).')';
400    $snippet = preg_replace('#'.$mark.'#si','<span class="search_hit">\\1</span>',$snippet);
401
402    $data[] = array(
403      'id'       => pathID($file),
404      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
405      'poswords' => join(' ',$words),
406      'snippet'  => $snippet,
407    );
408  }
409
410  return true;
411}
412
413
414/**
415 * fulltext sort
416 *
417 * Callback sort function for use with usort to sort the data
418 * structure created by search_fulltext. Sorts descending by count
419 *
420 * @author  Andreas Gohr <andi@splitbrain.org>
421 */
422function sort_search_fulltext($a,$b){
423  if($a['count'] > $b['count']){
424    return -1;
425  }elseif($a['count'] < $b['count']){
426    return 1;
427  }else{
428    return strcmp($a['id'],$b['id']);
429  }
430}
431
432/**
433 * translates a document path to an ID
434 *
435 * @author  Andreas Gohr <andi@splitbrain.org>
436 * @todo    move to pageutils
437 */
438function pathID($path,$keeptxt=false){
439  $id = utf8_decodeFN($path);
440  $id = str_replace('/',':',$id);
441  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
442  $id = preg_replace('#^:+#','',$id);
443  $id = preg_replace('#:+$#','',$id);
444  return $id;
445}
446
447
448//Setup VIM: ex: et ts=2 enc=utf-8 :
449