xref: /dokuwiki/inc/search.php (revision 63f2400befcdf463a6c6d34a9235e1c6e4b1e909) 
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    }
33    $files[] = $dir.'/'.$file;
34  }
35  closedir($dh);
36  sort($files);
37  sort($dirs);
38
39  //give directories to userfunction then recurse
40  foreach($dirs as $dir){
41    if ($func($data,$base,$dir,'d',$lvl,$opts)){
42      search($data,$base,$func,$opts,$dir,$lvl+1);
43    }
44  }
45  //now handle the files
46  foreach($files as $file){
47    $func($data,$base,$file,'f',$lvl,$opts);
48  }
49}
50
51/**
52 * The following functions are userfunctions to use with the search
53 * function above. This function is called for every found file or
54 * directory. When a directory is given to the function it has to
55 * decide if this directory should be traversed (true) or not (false)
56 * The function has to accept the following parameters:
57 *
58 * &$data - Reference to the result data structure
59 * $base  - Base usually $conf['datadir']
60 * $file  - current file or directory relative to $base
61 * $type  - Type either 'd' for directory or 'f' for file
62 * $lvl   - Current recursion depht
63 * $opts  - option array as given to search()
64 *
65 * return values for files are ignored
66 *
67 * All functions should check the ACL for document READ rights
68 * namespaces (directories) are NOT checked as this would break
69 * the recursion (You can have an nonreadable dir over a readable
70 * one deeper nested)
71 */
72
73/**
74 * Searches for pages beginning with the given query
75 *
76 * @author Andreas Gohr <andi@splitbrain.org>
77 */
78function search_qsearch(&$data,$base,$file,$type,$lvl,$opts){
79  $item = array();
80
81  if($type == 'd'){
82    return false; //no handling yet
83  }
84
85  //get id
86  $id = pathID($file);
87
88  //check if it matches the query
89  if(!preg_match('/^'.preg_quote($opts['query'],'/').'/u',$id)){
90    return false;
91  }
92
93  //check ACL
94  if(auth_quickaclcheck($id) < AUTH_READ){
95    return false;
96  }
97
98  $data[]=array( 'id'    => $id,
99                 'type'  => $type,
100                 'level' => 1,
101                 'open'  => true);
102  return true;
103}
104
105/**
106 * Build the browsable index of pages
107 *
108 * $opts['ns'] is the current namespace
109 *
110 * @author  Andreas Gohr <andi@splitbrain.org>
111 */
112function search_index(&$data,$base,$file,$type,$lvl,$opts){
113  $return = true;
114
115  $item = array();
116
117  if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
118    //add but don't recurse
119    $return = false;
120  }elseif($type == 'f' && !preg_match('#\.txt$#',$file)){
121    //don't add
122    return false;
123  }
124
125  //check ACL
126  $id = pathID($file);
127  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
128    return false;
129  }
130
131  $data[]=array( 'id'    => $id,
132                 'type'  => $type,
133                 'level' => $lvl,
134                 'open'  => $return );
135  return $return;
136}
137
138/**
139 * List all namespaces
140 *
141 * @author  Andreas Gohr <andi@splitbrain.org>
142 */
143function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
144  if($type == 'f') return true; //nothing to do on files
145
146  $id = pathID($file);
147  $data[]=array( 'id'    => $id,
148                 'type'  => $type,
149                 'level' => $lvl );
150  return true;
151}
152
153/**
154 * List all mediafiles in a namespace
155 *
156 * @author  Andreas Gohr <andi@splitbrain.org>
157 */
158function search_media(&$data,$base,$file,$type,$lvl,$opts){
159  //we do nothing with directories
160  if($type == 'd') return false;
161
162  $info         = array();
163  $info['id']   = pathID($file);
164
165  //check ACL for namespace (we have no ACL for mediafiles)
166  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
167    return false;
168  }
169
170  $info['file'] = basename($file);
171  $info['size'] = filesize($base.'/'.$file);
172  if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
173    $info['isimg'] = true;
174    $info['info']  = getimagesize($base.'/'.$file);
175  }else{
176    $info['isimg'] = false;
177  }
178  $data[] = $info;
179
180  return false;
181}
182
183/**
184 * This function just lists documents (for RSS namespace export)
185 *
186 * @author  Andreas Gohr <andi@splitbrain.org>
187 */
188function search_list(&$data,$base,$file,$type,$lvl,$opts){
189  //we do nothing with directories
190  if($type == 'd') return false;
191  if(preg_match('#\.txt$#',$file)){
192    //check ACL
193    $id = pathID($file);
194    if(auth_quickaclcheck($id) < AUTH_READ){
195      return false;
196    }
197    $data[]['id'] = $id;;
198  }
199  return false;
200}
201
202/**
203 * Quicksearch for searching matching pagenames
204 *
205 * $opts['query'] is the search query
206 *
207 * @author  Andreas Gohr <andi@splitbrain.org>
208 */
209function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
210  //we do nothing with directories
211  if($type == 'd') return true;
212  //only search txt files
213  if(!preg_match('#\.txt$#',$file)) return true;
214
215  //simple stringmatching
216  if(strpos($file,$opts['query']) !== false){
217    //check ACL
218    $id = pathID($file);
219    if(auth_quickaclcheck($id) < AUTH_READ){
220      return false;
221    }
222    $data[]['id'] = $id;
223  }
224
225  return true;
226}
227
228/**
229 * Search for backlinks to a given page
230 *
231 * $opts['ns']    namespace of the page
232 * $opts['name']  name of the page without namespace
233 *
234 * @author  Andreas Gohr <andi@splitbrain.org>
235 */
236function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
237  //we do nothing with directories
238  if($type == 'd') return true;;
239  //only search txt files
240  if(!preg_match('#\.txt$#',$file)) return true;;
241
242  //absolute search id
243  $sid = cleanID($opts['ns'].':'.$opts['name']);
244
245  //current id and namespace
246  $cid = pathID($file);
247  $cns = getNS($cid);
248
249  //check ACL
250  if(auth_quickaclcheck($cid) < AUTH_READ){
251    return false;
252  }
253
254  //fetch instructions
255  require_once(DOKU_INC.'inc/parserutils.php');
256  $instructions = p_cached_instructions($base.$file,true);
257  if(is_null($instructions)) return false;
258
259  //check all links for match
260  foreach($instructions as $ins){
261    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
262      $mid = $ins[1][0];
263      resolve_pageid($cns,$mid,$exists); //exists is not used
264      if($mid == $sid){
265        //we have a match - finish
266        $data[]['id'] = $cid;
267        break;
268      }
269    }
270  }
271
272  return false;
273}
274
275/**
276 * Fulltextsearch
277 *
278 * $opts['query'] is the search query
279 *
280 * @author  Andreas Gohr <andi@splitbrain.org>
281 */
282function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
283  //we do nothing with directories
284  if($type == 'd') return true;;
285  //only search txt files
286  if(!preg_match('#\.txt$#',$file)) return true;;
287
288  //check ACL
289  $id = pathID($file);
290  if(auth_quickaclcheck($id) < AUTH_READ){
291    return false;
292  }
293
294  //get text
295  $text = io_readfile($base.'/'.$file);
296  //lowercase text (u modifier does not help with case)
297  $lctext = utf8_strtolower($text);
298
299  //create regexp from queries
300  $qpreg = preg_split('/\s+/',preg_quote($opts['query'],'#'));
301  $qpreg = '('.join('|',$qpreg).')';
302
303  //do the fulltext search
304  $matches = array();
305  if($cnt = preg_match_all('#'.$qpreg.'#usi',$lctext,$matches)){
306    //this is not the best way for snippet generation but the fastest I could find
307    //split query and only use the first token
308    $q = preg_split('/\s+/',$opts['query'],2);
309    $q = $q[0];
310    $p = utf8_strpos($lctext,$q);
311    $f = $p - 100;
312    $l = utf8_strlen($q) + 200;
313    if($f < 0) $f = 0;
314    $snippet = '<span class="search_sep"> ... </span>'.
315               htmlspecialchars(utf8_substr($text,$f,$l)).
316               '<span class="search_sep"> ... </span>';
317    $snippet = preg_replace('#'.$qpreg.'#si','<span class="search_hit">\\1</span>',$snippet);
318
319    $data[] = array(
320      'id'      => $id,
321      'count'   => $cnt,
322      'snippet' => $snippet,
323    );
324  }
325
326  return true;
327}
328
329/**
330 * fulltext sort
331 *
332 * Callback sort function for use with usort to sort the data
333 * structure created by search_fulltext. Sorts descending by count
334 *
335 * @author  Andreas Gohr <andi@splitbrain.org>
336 */
337function sort_search_fulltext($a,$b){
338  if($a['count'] > $b['count']){
339    return -1;
340  }elseif($a['count'] < $b['count']){
341    return 1;
342  }else{
343    return strcmp($a['id'],$b['id']);
344  }
345}
346
347/**
348 * translates a document path to an ID
349 *
350 * @author  Andreas Gohr <andi@splitbrain.org>
351 * @todo    move to pageutils
352 */
353function pathID($path){
354  $id = utf8_decodeFN($path);
355  $id = str_replace('/',':',$id);
356  $id = preg_replace('#\.txt$#','',$id);
357  $id = preg_replace('#^:+#','',$id);
358  $id = preg_replace('#:+$#','',$id);
359  return $id;
360}
361
362
363//Setup VIM: ex: et ts=2 enc=utf-8 :
364