xref: /dokuwiki/inc/search.php (revision 340756e4d13bbcffbd9cba79aef7fc7a6ca2ecb8) 
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  //get text
211  $text = io_readfile($base.'/'.$file);
212
213  //absolute search id
214  $sid = cleanID($opts['ns'].':'.$opts['name']);
215
216  //construct current namespace
217  $cid = pathID($file);
218  $cns = getNS($cid);
219
220  //check ACL
221  if(auth_quickaclcheck($cid) < AUTH_READ){
222    return false;
223  }
224
225  //match all links
226  //FIXME may be incorrect because of code blocks
227  //      CamelCase isn't supported, too
228  preg_match_all('#\[\[(.+?)\]\]#si',$text,$matches,PREG_SET_ORDER);
229  foreach($matches as $match){
230    //get ID from link and discard most non wikilinks
231    list($mid) = split('\|',$match[1],2);
232    if(preg_match("#^(https?|telnet|gopher|file|wais|ftp|ed2k|irc)://#",$mid)) continue;
233    if(preg_match("#\w+>#",$mid)) continue;
234    $mns = getNS($mid);
235   	//namespace starting with "." - prepend current namespace
236    if(strpos($mns,'.')===0){
237      $mid = $cns.":".substr($mid,1);
238    }
239    if($mns===false){
240      //no namespace in link? add current
241      $mid = "$cns:$mid";
242    }
243    $mid = cleanID($mid);
244
245    if ($mid == $sid){
246      $data[]['id'] = $cid;
247      break;
248    }
249  }
250}
251
252/**
253 * Fulltextsearch
254 *
255 * $opts['query'] is the search query
256 *
257 * @author  Andreas Gohr <andi@splitbrain.org>
258 */
259function search_fulltext(&$data,$base,$file,$type,$lvl,$opts){
260  //we do nothing with directories
261  if($type == 'd') return true;;
262  //only search txt files
263  if(!preg_match('#\.txt$#',$file)) return true;;
264
265  //check ACL
266  $id = pathID($file);
267  if(auth_quickaclcheck($id) < AUTH_READ){
268    return false;
269  }
270
271  //get text
272  $text = io_readfile($base.'/'.$file);
273  //lowercase text (u modifier does not help with case)
274  $lctext = utf8_strtolower($text);
275
276  //create regexp from queries
277  $qpreg = preg_split('/\s+/',preg_quote($opts['query'],'#'));
278  $qpreg = '('.join('|',$qpreg).')';
279
280  //do the fulltext search
281  $matches = array();
282  if($cnt = preg_match_all('#'.$qpreg.'#usi',$lctext,$matches)){
283    //this is not the best way for snippet generation but the fastest I could find
284    //split query and only use the first token
285    $q = preg_split('/\s+/',$opts['query'],2);
286    $q = $q[0];
287    $p = utf8_strpos($lctext,$q);
288    $f = $p - 100;
289    $l = utf8_strlen($q) + 200;
290    if($f < 0) $f = 0;
291    $snippet = '<span class="search_sep"> ... </span>'.
292               htmlspecialchars(utf8_substr($text,$f,$l)).
293               '<span class="search_sep"> ... </span>';
294    $snippet = preg_replace('#'.$qpreg.'#si','<span class="search_hit">\\1</span>',$snippet);
295
296    $data[] = array(
297      'id'      => $id,
298      'count'   => $cnt,
299      'snippet' => $snippet,
300    );
301  }
302
303  return true;
304}
305
306/**
307 * fulltext sort
308 *
309 * Callback sort function for use with usort to sort the data
310 * structure created by search_fulltext. Sorts descending by count
311 *
312 * @author  Andreas Gohr <andi@splitbrain.org>
313 */
314function sort_search_fulltext($a,$b){
315  if($a['count'] > $b['count']){
316    return -1;
317  }elseif($a['count'] < $b['count']){
318    return 1;
319  }else{
320    return strcmp($a['id'],$b['id']);
321  }
322}
323
324/**
325 * translates a document path to an ID
326 *
327 * @author  Andreas Gohr <andi@splitbrain.org>
328 */
329function pathID($path){
330  $id = utf8_decodeFN($path);
331  $id = str_replace('/',':',$id);
332  $id = preg_replace('#\.txt$#','',$id);
333  $id = preg_replace('#^:+#','',$id);
334  $id = preg_replace('#:+$#','',$id);
335  return $id;
336}
337
338
339//Setup VIM: ex: et ts=2 enc=utf-8 :
340