xref: /dokuwiki/inc/search.php (revision ee33a34725046a56589daea1b2957f357d1836aa) 
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 (!empty($opts['query'])){
220    if(strpos($file,$opts['query']) !== false){
221      //check ACL
222      $id = pathID($file);
223      if(auth_quickaclcheck($id) < AUTH_READ){
224        return false;
225      }
226      $data[]['id'] = $id;
227    }
228  }
229  return true;
230}
231
232/**
233 * Search for backlinks to a given page
234 *
235 * $opts['ns']    namespace of the page
236 * $opts['name']  name of the page without namespace
237 *
238 * @author  Andreas Gohr <andi@splitbrain.org>
239 */
240function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
241  //we do nothing with directories
242  if($type == 'd') return true;;
243  //only search txt files
244  if(!preg_match('#\.txt$#',$file)) return true;;
245
246  //absolute search id
247  $sid = cleanID($opts['ns'].':'.$opts['name']);
248
249  //current id and namespace
250  $cid = pathID($file);
251  $cns = getNS($cid);
252
253  //check ACL
254  if(auth_quickaclcheck($cid) < AUTH_READ){
255    return false;
256  }
257
258  //fetch instructions
259  require_once(DOKU_INC.'inc/parserutils.php');
260  $instructions = p_cached_instructions($base.$file,true);
261  if(is_null($instructions)) return false;
262
263  //check all links for match
264  foreach($instructions as $ins){
265    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
266      $mid = $ins[1][0];
267      resolve_pageid($cns,$mid,$exists); //exists is not used
268      if($mid == $sid){
269        //we have a match - finish
270        $data[]['id'] = $cid;
271        break;
272      }
273    }
274  }
275
276  return false;
277}
278
279/**
280 * Fulltextsearch
281 *
282 * $opts['query'] is the search query
283 *
284 * @author  Andreas Gohr <andi@splitbrain.org>
285 */
286function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
287  //we do nothing with directories
288  if($type == 'd') return true;;
289  //only search txt files
290  if(!preg_match('#\.txt$#',$file)) return true;;
291
292  //check ACL
293  $id = pathID($file);
294  if(auth_quickaclcheck($id) < AUTH_READ){
295    return false;
296  }
297
298  //create regexp from queries
299  $poswords = array();
300  $negwords = array();
301  $qpreg = preg_split('/\s+/',$opts['query']);
302
303  foreach($qpreg as $word){
304    switch(substr($word,0,1)){
305      case '-':
306        if(strlen($word) > 1){  // catch single '-'
307          array_push($negwords,preg_quote(substr($word,1),'#'));
308        }
309        break;
310      case '+':
311        if(strlen($word) > 1){  // catch single '+'
312          array_push($poswords,preg_quote(substr($word,1),'#'));
313        }
314        break;
315      default:
316        array_push($poswords,preg_quote($word,'#'));
317        break;
318    }
319  }
320
321  // a search without any posword is useless
322  if (!count($poswords)) return true;
323
324  $reg  = '^(?=.*?'.join(')(?=.*?',$poswords).')';
325  $reg .= count($negwords) ? '((?!'.join('|',$negwords).').)*$' : '.*$';
326  search_regex($data,$base,$file,$reg,$poswords);
327  return true;
328}
329
330/**
331 * Reference search
332 * This fuction searches for existing references to a given media file
333 * and returns an array with the found pages. It doesn't pay any
334 * attention to ACL permissions to find every reference. The caller
335 * must check if the user has the appropriate rights to see the found
336 * page and eventually have to prevent the result from displaying.
337 *
338 * @param array  $data Reference to the result data structure
339 * @param string $base Base usually $conf['datadir']
340 * @param string $file current file or directory relative to $base
341 * @param char   $type Type either 'd' for directory or 'f' for file
342 * @param int    $lvl  Current recursion depht
343 * @param mixed  $opts option array as given to search()
344 *
345 * $opts['query'] is the demanded media file name
346 *
347 * @author  Andreas Gohr <andi@splitbrain.org>
348 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
349 */
350function search_reference(&$data,$base,$file,$type,$lvl,$opts){
351  global $conf;
352
353  //we do nothing with directories
354  if($type == 'd') return true;
355
356  //only search txt files
357  if(!preg_match('#\.txt$#',$file)) return true;
358
359  //we finish after 'cnt' references found. The return value
360  //'false' will skip subdirectories to speed search up.
361  $cnt = $conf['refshow'] > 0 ? $conf['refshow'] : 1;
362  if(count($data) >= $cnt) return false;
363
364  $reg = '\{\{ *\:?'.$opts['query'].' *(\|.*)?\}\}';
365  search_regex($data,$base,$file,$reg,array($opts['query']));
366  return true;
367}
368
369/* ------------- helper functions below -------------- */
370
371/**
372 * fulltext search helper
373 * searches a text file with a given regular expression
374 * no ACL checks are performed. This have to be done by
375 * the caller if necessary.
376 *
377 * @param array  $data  reference to array for results
378 * @param string $base  base directory
379 * @param string $file  file name to search in
380 * @param string $reg   regular expression to search for
381 * @param array  $words words that should be marked in the results
382 *
383 * @author  Andreas Gohr <andi@splitbrain.org>
384 * @author  Matthias Grimm <matthiasgrimm@users.sourceforge.net>
385 */
386function search_regex(&$data,$base,$file,$reg,$words){
387
388  //get text
389  $text = io_readfile($base.'/'.$file);
390  //lowercase text (u modifier does not help with case)
391  $lctext = utf8_strtolower($text);
392
393  //do the fulltext search
394  $matches = array();
395  if($cnt = preg_match_all('#'.$reg.'#usi',$lctext,$matches)){
396    //this is not the best way for snippet generation but the fastest I could find
397    $q = $words[0];  //use first word for snippet creation
398    $p = utf8_strpos($lctext,$q);
399    $f = $p - 100;
400    $l = utf8_strlen($q) + 200;
401    if($f < 0) $f = 0;
402    $snippet = '<span class="search_sep"> ... </span>'.
403               htmlspecialchars(utf8_substr($text,$f,$l)).
404               '<span class="search_sep"> ... </span>';
405    $mark    = '('.join('|', $words).')';
406    $snippet = preg_replace('#'.$mark.'#si','<span class="search_hit">\\1</span>',$snippet);
407
408    $data[] = array(
409      'id'       => pathID($file),
410      'count'    => preg_match_all('#'.$mark.'#usi',$lctext,$matches),
411      'poswords' => join(' ',$words),
412      'snippet'  => $snippet,
413    );
414  }
415
416  return true;
417}
418
419
420/**
421 * fulltext sort
422 *
423 * Callback sort function for use with usort to sort the data
424 * structure created by search_fulltext. Sorts descending by count
425 *
426 * @author  Andreas Gohr <andi@splitbrain.org>
427 */
428function sort_search_fulltext($a,$b){
429  if($a['count'] > $b['count']){
430    return -1;
431  }elseif($a['count'] < $b['count']){
432    return 1;
433  }else{
434    return strcmp($a['id'],$b['id']);
435  }
436}
437
438/**
439 * translates a document path to an ID
440 *
441 * @author  Andreas Gohr <andi@splitbrain.org>
442 * @todo    move to pageutils
443 */
444function pathID($path,$keeptxt=false){
445  $id = utf8_decodeFN($path);
446  $id = str_replace('/',':',$id);
447  if(!$keeptxt) $id = preg_replace('#\.txt$#','',$id);
448  $id = preg_replace('#^:+#','',$id);
449  $id = preg_replace('#:+$#','',$id);
450  return $id;
451}
452
453
454//Setup VIM: ex: et ts=2 enc=utf-8 :
455