xref: /dokuwiki/inc/search.php (revision 36df6fa3069d55db3c35724c4f465bde9c50b53a) 
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    require_once(DOKU_INC.'inc/JpegMeta.php');
178    $info['meta']  = new JpegMeta($base.'/'.$file);
179  }else{
180    $info['isimg'] = false;
181  }
182  $data[] = $info;
183
184  return false;
185}
186
187/**
188 * This function just lists documents (for RSS namespace export)
189 *
190 * @author  Andreas Gohr <andi@splitbrain.org>
191 */
192function search_list(&$data,$base,$file,$type,$lvl,$opts){
193  //we do nothing with directories
194  if($type == 'd') return false;
195  if(preg_match('#\.txt$#',$file)){
196    //check ACL
197    $id = pathID($file);
198    if(auth_quickaclcheck($id) < AUTH_READ){
199      return false;
200    }
201    $data[]['id'] = $id;;
202  }
203  return false;
204}
205
206/**
207 * Quicksearch for searching matching pagenames
208 *
209 * $opts['query'] is the search query
210 *
211 * @author  Andreas Gohr <andi@splitbrain.org>
212 */
213function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
214  //we do nothing with directories
215  if($type == 'd') return true;
216  //only search txt files
217  if(!preg_match('#\.txt$#',$file)) return true;
218
219  //simple stringmatching
220  if (!empty($opts['query'])){
221    if(strpos($file,$opts['query']) !== false){
222      //check ACL
223      $id = pathID($file);
224      if(auth_quickaclcheck($id) < AUTH_READ){
225        return false;
226      }
227      $data[]['id'] = $id;
228    }
229  }
230  return true;
231}
232
233/**
234 * Search for backlinks to a given page
235 *
236 * $opts['ns']    namespace of the page
237 * $opts['name']  name of the page without namespace
238 *
239 * @author  Andreas Gohr <andi@splitbrain.org>
240 */
241function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
242  //we do nothing with directories
243  if($type == 'd') return true;;
244  //only search txt files
245  if(!preg_match('#\.txt$#',$file)) return true;;
246
247  //absolute search id
248  $sid = cleanID($opts['ns'].':'.$opts['name']);
249
250  //current id and namespace
251  $cid = pathID($file);
252  $cns = getNS($cid);
253
254  //check ACL
255  if(auth_quickaclcheck($cid) < AUTH_READ){
256    return false;
257  }
258
259  //fetch instructions
260  require_once(DOKU_INC.'inc/parserutils.php');
261  $instructions = p_cached_instructions($base.$file,true);
262  if(is_null($instructions)) return false;
263
264  //check all links for match
265  foreach($instructions as $ins){
266    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
267      $mid = $ins[1][0];
268      resolve_pageid($cns,$mid,$exists); //exists is not used
269      if($mid == $sid){
270        //we have a match - finish
271        $data[]['id'] = $cid;
272        break;
273      }
274    }
275  }
276
277  return false;
278}
279
280/**
281 * Fulltextsearch
282 *
283 * $opts['query'] is the search query
284 *
285 * @author  Andreas Gohr <andi@splitbrain.org>
286 */
287function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
288  //we do nothing with directories
289  if($type == 'd') return true;;
290  //only search txt files
291  if(!preg_match('#\.txt$#',$file)) return true;;
292
293  //check ACL
294  $id = pathID($file);
295  if(auth_quickaclcheck($id) < AUTH_READ){
296    return false;
297  }
298
299  //create regexp from queries
300  $poswords = array();
301  $negwords = array();
302  $qpreg = preg_split('/\s+/',$opts['query']);
303
304  foreach($qpreg as $word){
305    switch(substr($word,0,1)){
306      case '-':
307        if(strlen($word) > 1){  // catch single '-'
308          array_push($negwords,preg_quote(substr($word,1),'#'));
309        }
310        break;
311      case '+':
312        if(strlen($word) > 1){  // catch single '+'
313          array_push($poswords,preg_quote(substr($word,1),'#'));
314        }
315        break;
316      default:
317        array_push($poswords,preg_quote($word,'#'));
318        break;
319    }
320  }
321
322  // a search without any posword is useless
323  if (!count($poswords)) return true;
324
325  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
326  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
327  search_regex($data,$base,$file,$reg,$poswords);
328  return true;
329}
330
331/**
332 * Reference search
333 * This fuction searches for existing references to a given media file
334 * and returns an array with the found pages. It doesn't pay any
335 * attention to ACL permissions to find every reference. The caller
336 * must check if the user has the appropriate rights to see the found
337 * page and eventually have to prevent the result from displaying.
338 *
339 * @param array  $data Reference to the result data structure
340 * @param string $base Base usually $conf['datadir']
341 * @param string $file current file or directory relative to $base
342 * @param char   $type Type either 'd' for directory or 'f' for file
343 * @param int    $lvl  Current recursion depht
344 * @param mixed  $opts option array as given to search()
345 *
346 * $opts['query'] is the demanded media file name
347 *
348 * @author  Andreas Gohr <andi@splitbrain.org>
349 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
350 */
351function search_reference(&$data,$base,$file,$type,$lvl,$opts){
352  global $conf;
353
354  //we do nothing with directories
355  if($type == 'd') return true;
356
357  //only search txt files
358  if(!preg_match('#\.txt$#',$file)) return true;
359
360  //we finish after 'cnt' references found. The return value
361  //'false' will skip subdirectories to speed search up.
362  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
363  if(count($data) >= $cnt) return false;
364
365  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
366  search_regex($data,$base,$file,$reg,array($opts['query']));
367  return true;
368}
369
370/* ------------- helper functions below -------------- */
371
372/**
373 * fulltext search helper
374 * searches a text file with a given regular expression
375 * no ACL checks are performed. This have to be done by
376 * the caller if necessary.
377 *
378 * @param array  $data  reference to array for results
379 * @param string $base  base directory
380 * @param string $file  file name to search in
381 * @param string $reg   regular expression to search for
382 * @param array  $words words that should be marked in the results
383 *
384 * @author  Andreas Gohr <andi@splitbrain.org>
385 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
386 */
387function search_regex(&$data,$base,$file,$reg,$words){
388
389  //get text
390  $text = io_readfile($base.'/'.$file);
391  //lowercase text (u modifier does not help with case)
392  $lctext = utf8_strtolower($text);
393
394  //do the fulltext search
395  $matches = array();
396  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
397    //this is not the best way for snippet generation but the fastest I could find
398    $q = $words[0];  //use first word for snippet creation
399    $p = utf8_strpos($lctext,$q);
400    $f = $p - 100;
401    $l = utf8_strlen($q) + 200;
402    if($f < 0) $f = 0;
403    $snippet = '<span class="search_sep"> ... </span>'.
404               htmlspecialchars(utf8_substr($text,$f,$l)).
405               '<span class="search_sep"> ... </span>';
406    $mark    = '('.join('|', $words).')';
407    $snippet = preg_replace('#'.$mark.'#si','<span class="search_hit">\\1</span>',$snippet);
408
409    $data[] = array(
410      'id'       => pathID($file),
411      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
412      'poswords' => join(' ',$words),
413      'snippet'  => $snippet,
414    );
415  }
416
417  return true;
418}
419
420
421/**
422 * fulltext sort
423 *
424 * Callback sort function for use with usort to sort the data
425 * structure created by search_fulltext. Sorts descending by count
426 *
427 * @author  Andreas Gohr <andi@splitbrain.org>
428 */
429function sort_search_fulltext($a,$b){
430  if($a['count'] > $b['count']){
431    return -1;
432  }elseif($a['count'] < $b['count']){
433    return 1;
434  }else{
435    return strcmp($a['id'],$b['id']);
436  }
437}
438
439/**
440 * translates a document path to an ID
441 *
442 * @author  Andreas Gohr <andi@splitbrain.org>
443 * @todo    move to pageutils
444 */
445function pathID($path,$keeptxt=false){
446  $id = utf8_decodeFN($path);
447  $id = str_replace('/',':',$id);
448  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
449  $id = preg_replace('#^:+#','',$id);
450  $id = preg_replace('#:+$#','',$id);
451  return $id;
452}
453
454
455//Setup VIM: ex: et ts=2 enc=utf-8 :
456