xref: /dokuwiki/inc/search.php (revision 37e34a5ef6c436d6cf1525ba3643b1597cadc168) 
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 * Build the browsable index of pages
75 *
76 * $opts['ns'] is the current namespace
77 *
78 * @author  Andreas Gohr <andi@splitbrain.org>
79 */
80function search_index(&$data,$base,$file,$type,$lvl,$opts){
81  $return = true;
82
83  $item = array();
84
85  if($type == 'd' && !preg_match('#^'.$file.'(/|$)#','/'.$opts['ns'])){
86    //add but don't recurse
87    $return = false;
88  }elseif($type == 'f' && !preg_match('#\.txt$#',$file)){
89    //don't add
90    return false;
91  }
92
93  //check ACL
94  $id = pathID($file);
95  if($type=='f' && auth_quickaclcheck($id) < AUTH_READ){
96    return false;
97  }
98
99  $data[]=array( 'id'    => $id,
100                 'type'  => $type,
101                 'level' => $lvl,
102                 'open'  => $return );
103  return $return;
104}
105
106/**
107 * List all namespaces
108 *
109 * @author  Andreas Gohr <andi@splitbrain.org>
110 */
111function search_namespaces(&$data,$base,$file,$type,$lvl,$opts){
112  if($type == 'f') return true; //nothing to do on files
113
114  $id = pathID($file);
115  $data[]=array( 'id'    => $id,
116                 'type'  => $type,
117                 'level' => $lvl );
118  return true;
119}
120
121/**
122 * List all mediafiles in a namespace
123 *
124 * @author  Andreas Gohr <andi@splitbrain.org>
125 */
126function search_media(&$data,$base,$file,$type,$lvl,$opts){
127  //we do nothing with directories
128  if($type == 'd') return false;
129
130  $info         = array();
131  $info['id']   = pathID($file);
132
133  //check ACL for namespace (we have no ACL for mediafiles)
134  if(auth_quickaclcheck(getNS($info['id']).':*') < AUTH_READ){
135    return false;
136  }
137
138  $info['file'] = basename($file);
139  $info['size'] = filesize($base.'/'.$file);
140  if(preg_match("/\.(jpe?g|gif|png)$/",$file)){
141    $info['isimg'] = true;
142    $info['info']  = getimagesize($base.'/'.$file);
143  }else{
144    $info['isimg'] = false;
145  }
146  $data[] = $info;
147
148  return false;
149}
150
151/**
152 * This function just lists documents (for RSS namespace export)
153 *
154 * @author  Andreas Gohr <andi@splitbrain.org>
155 */
156function search_list(&$data,$base,$file,$type,$lvl,$opts){
157  //we do nothing with directories
158  if($type == 'd') return false;
159  if(preg_match('#\.txt$#',$file)){
160    //check ACL
161    $id = pathID($file);
162    if(auth_quickaclcheck($id) < AUTH_READ){
163      return false;
164    }
165    $data[]['id'] = $id;;
166  }
167  return false;
168}
169
170/**
171 * Quicksearch for searching matching pagenames
172 *
173 * $opts['query'] is the search query
174 *
175 * @author  Andreas Gohr <andi@splitbrain.org>
176 */
177function search_pagename(&$data,$base,$file,$type,$lvl,$opts){
178  //we do nothing with directories
179  if($type == 'd') return true;
180  //only search txt files
181  if(!preg_match('#\.txt$#',$file)) return true;
182
183  //simple stringmatching
184  if(strpos($file,$opts['query']) !== false){
185    //check ACL
186    $id = pathID($file);
187    if(auth_quickaclcheck($id) < AUTH_READ){
188      return false;
189    }
190    $data[]['id'] = $id;
191  }
192
193  return true;
194}
195
196/**
197 * Search for backlinks to a given page
198 *
199 * $opts['ns']    namespace of the page
200 * $opts['name']  name of the page without namespace
201 *
202 * @author  Andreas Gohr <andi@splitbrain.org>
203 */
204function search_backlinks(&$data,$base,$file,$type,$lvl,$opts){
205  //we do nothing with directories
206  if($type == 'd') return true;;
207  //only search txt files
208  if(!preg_match('#\.txt$#',$file)) return true;;
209
210  //absolute search id
211  $sid = cleanID($opts['ns'].':'.$opts['name']);
212
213  //current id and namespace
214  $cid = pathID($file);
215  $cns = getNS($cid);
216
217  //check ACL
218  if(auth_quickaclcheck($cid) < AUTH_READ){
219    return false;
220  }
221
222  //fetch instructions
223  require_once(DOKU_INC.'inc/parserutils.php');
224  $instructions = p_cached_instructions($base.$file,true);
225  if(is_null($instructions)) return false;
226
227  //check all links for match
228  foreach($instructions as $ins){
229    if($ins[0] == 'internallink' || ($conf['camelcase'] && $ins[0] == 'camelcaselink') ){
230      $mid = $ins[1][0];
231      resolve_pageid($cns,$mid,$exists); //exists is not used
232      if($mid == $sid){
233        //we have a match - finish
234        $data[]['id'] = $cid;
235        break;
236      }
237    }
238  }
239
240  return false;
241}
242
243/**
244 * Fulltextsearch
245 *
246 * $opts['query'] is the search query
247 *
248 * @author  Andreas Gohr <andi@splitbrain.org>
249 */
250function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
251  //we do nothing with directories
252  if($type == 'd') return true;;
253  //only search txt files
254  if(!preg_match('#\.txt$#',$file)) return true;;
255
256  //check ACL
257  $id = pathID($file);
258  if(auth_quickaclcheck($id) < AUTH_READ){
259    return false;
260  }
261
262  //get text
263  $text = io_readfile($base.'/'.$file);
264  //lowercase text (u modifier does not help with case)
265  $lctext = utf8_strtolower($text);
266
267  //create regexp from queries
268  $qpreg = preg_split('/\s+/',preg_quote($opts['query'],'#'));
269  $qpreg = '('.join('|',$qpreg).')';
270
271  //do the fulltext search
272  $matches = array();
273  if($cnt = preg_match_all('#'.$qpreg.'#usi',$lctext,$matches)){
274    //this is not the best way for snippet generation but the fastest I could find
275    //split query and only use the first token
276    $q = preg_split('/\s+/',$opts['query'],2);
277    $q = $q[0];
278    $p = utf8_strpos($lctext,$q);
279    $f = $p - 100;
280    $l = utf8_strlen($q) + 200;
281    if($f < 0) $f = 0;
282    $snippet = '<span class="search_sep"> ... </span>'.
283               htmlspecialchars(utf8_substr($text,$f,$l)).
284               '<span class="search_sep"> ... </span>';
285    $snippet = preg_replace('#'.$qpreg.'#si','<span class="search_hit">\\1</span>',$snippet);
286
287    $data[] = array(
288      'id'      => $id,
289      'count'   => $cnt,
290      'snippet' => $snippet,
291    );
292  }
293
294  return true;
295}
296
297/**
298 * fulltext sort
299 *
300 * Callback sort function for use with usort to sort the data
301 * structure created by search_fulltext. Sorts descending by count
302 *
303 * @author  Andreas Gohr <andi@splitbrain.org>
304 */
305function sort_search_fulltext($a,$b){
306  if($a['count'] > $b['count']){
307    return -1;
308  }elseif($a['count'] < $b['count']){
309    return 1;
310  }else{
311    return strcmp($a['id'],$b['id']);
312  }
313}
314
315/**
316 * translates a document path to an ID
317 *
318 * @author  Andreas Gohr <andi@splitbrain.org>
319 * @todo    move to pageutils
320 */
321function pathID($path){
322  $id = utf8_decodeFN($path);
323  $id = str_replace('/',':',$id);
324  $id = preg_replace('#\.txt$#','',$id);
325  $id = preg_replace('#^:+#','',$id);
326  $id = preg_replace('#:+$#','',$id);
327  return $id;
328}
329
330
331//Setup VIM: ex: et ts=2 enc=utf-8 :
332