search MDL-25153 fixed performance problem
[moodle.git] / search / query.php
CommitLineData
682d4032 1<?php
3319ef85 2 /**
3 * Global Search Engine for Moodle
4 *
5 * @package search
6 * @category core
7 * @subpackage search_engine
8 * @author Michael Champanis (mchampan) [cynnical@gmail.com], Valery Fremaux [valery.fremaux@club-internet.fr] > 1.8
9 * @date 2008/03/31
10 * @license http://www.gnu.org/copyleft/gpl.html GNU Public License
11 *
12 * The query page - accepts a user-entered query string and returns results.
13 *
14 * Queries are boolean-aware, e.g.:
15 *
16 * '+' term required
17 * '-' term must not be present
18 * '' (no modifier) term's presence increases rank, but isn't required
19 * 'field:' search this field
20 *
21 * Examples:
22 *
23 * 'earthquake +author:michael'
24 * Searches for documents written by 'michael' that contain 'earthquake'
25 *
26 * 'earthquake +doctype:wiki'
27 * Search all wiki pages for 'earthquake'
28 *
29 * '+author:helen +author:foster'
30 * All articles written by Helen Foster
31 *
32 */
70f52c70 33
3319ef85 34 /**
35 * includes and requires
36 */
37 require_once('../config.php');
964a5e92 38 require_once($CFG->dirroot.'/search/lib.php');
eef868d1 39
3319ef85 40 if ($CFG->forcelogin) {
41 require_login();
42 }
70f52c70 43
3319ef85 44 if (empty($CFG->enableglobalsearch)) {
32487831 45 print_error('globalsearchdisabled', 'search');
3319ef85 46 }
70f52c70 47
a1abd74d 48 $adv = new stdClass();
70f52c70 49
3319ef85 50/// check for php5, but don't die yet (see line 52)
eef868d1 51
964a5e92 52 require_once($CFG->dirroot.'/search/querylib.php');
32487831 53
964a5e92 54 $page_number = optional_param('page', -1, PARAM_INT);
55 $pages = ($page_number == -1) ? false : true;
56 $advanced = (optional_param('a', '0', PARAM_INT) == '1') ? true : false;
00521032 57 $query_string = optional_param('query_string', '', PARAM_CLEAN);
32487831 58
a6855934 59 $url = new moodle_url('/search/query.php');
c86bdd5d 60 if ($page_number !== -1) {
61 $url->param('page', $page_number);
62 }
63 if ($advanced) {
64 $url->param('a', '1');
65 }
66 $PAGE->set_url($url);
67
70f52c70 68/// discard harmfull searches
32487831 69
964a5e92 70 if (!isset($CFG->block_search_utf8dir)){
71 set_config('block_search_utf8dir', 1);
72 }
70f52c70 73
74/// discard harmfull searches
32487831 75
964a5e92 76 if (preg_match("/^[\*\?]+$/", $query_string)){
77 $query_string = '';
78 $error = get_string('fullwildcardquery','search');
79 }
70f52c70 80
964a5e92 81
82 if ($pages && isset($_SESSION['search_advanced_query'])) {
83 // if both are set, then we are busy browsing through the result pages of an advanced query
84 $adv = unserialize($_SESSION['search_advanced_query']);
85 } elseif ($advanced) {
86 // otherwise we are dealing with a new advanced query
87 unset($_SESSION['search_advanced_query']);
88 session_unregister('search_advanced_query');
70f52c70 89
964a5e92 90 // chars to strip from strings (whitespace)
91 $chars = " \t\n\r\0\x0B,-+";
70f52c70 92
964a5e92 93 // retrieve advanced query variables
94 $adv->mustappear = trim(optional_param('mustappear', '', PARAM_CLEAN), $chars);
95 $adv->notappear = trim(optional_param('notappear', '', PARAM_CLEAN), $chars);
96 $adv->canappear = trim(optional_param('canappear', '', PARAM_CLEAN), $chars);
97 $adv->module = optional_param('module', '', PARAM_CLEAN);
98 $adv->title = trim(optional_param('title', '', PARAM_CLEAN), $chars);
99 $adv->author = trim(optional_param('author', '', PARAM_CLEAN), $chars);
70f52c70 100 }
964a5e92 101
102 if ($advanced) {
103 //parse the advanced variables into a query string
104 //TODO: move out to external query class (QueryParse?)
70f52c70 105
964a5e92 106 $query_string = '';
70f52c70 107
964a5e92 108 // get all available module types adding third party modules
109 $module_types = array_merge(array('all'), array_values(search_get_document_types()));
110 $module_types = array_merge($module_types, array_values(search_get_document_types('X_SEARCH_TYPE')));
111 $adv->module = in_array($adv->module, $module_types) ? $adv->module : 'all';
70f52c70 112
964a5e92 113 // convert '1 2' into '+1 +2' for required words field
114 if (strlen(trim($adv->mustappear)) > 0) {
115 $query_string = ' +'.implode(' +', preg_split("/[\s,;]+/", $adv->mustappear));
70f52c70 116 }
117
964a5e92 118 // convert '1 2' into '-1 -2' for not wanted words field
119 if (strlen(trim($adv->notappear)) > 0) {
120 $query_string .= ' -'.implode(' -', preg_split("/[\s,;]+/", $adv->notappear));
70f52c70 121 }
122
964a5e92 123 // this field is left untouched, apart from whitespace being stripped
124 if (strlen(trim($adv->canappear)) > 0) {
125 $query_string .= ' '.implode(' ', preg_split("/[\s,;]+/", $adv->canappear));
70f52c70 126 }
127
964a5e92 128 // add module restriction
4021d539 129 $doctypestr = 'doctype';
130 $titlestr = 'title';
131 $authorstr = 'author';
964a5e92 132 if ($adv->module != 'all') {
133 $query_string .= " +{$doctypestr}:".$adv->module;
70f52c70 134 }
135
964a5e92 136 // create title search string
137 if (strlen(trim($adv->title)) > 0) {
138 $query_string .= " +{$titlestr}:".implode(" +{$titlestr}:", preg_split("/[\s,;]+/", $adv->title));
70f52c70 139 }
140
964a5e92 141 // create author search string
142 if (strlen(trim($adv->author)) > 0) {
143 $query_string .= " +{$authorstr}:".implode(" +{$authorstr}:", preg_split("/[\s,;]+/", $adv->author));
70f52c70 144 }
145
964a5e92 146 // save our options if the query is valid
147 if (!empty($query_string)) {
148 $_SESSION['search_advanced_query'] = serialize($adv);
70f52c70 149 }
150 }
964a5e92 151
152 // normalise page number
153 if ($page_number < 1) {
154 $page_number = 1;
70f52c70 155 }
964a5e92 156
157 //run the query against the index ensuring internal coding works in UTF-8
158 Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive());
159 $sq = new SearchQuery($query_string, $page_number, 10, false);
70f52c70 160
2a250a0b 161 $site = get_site();
70f52c70 162
3319ef85 163 $strsearch = get_string('search', 'search');
164 $strquery = get_string('enteryoursearchquery', 'search');
70f52c70 165
964a5e92 166 // print the header
964a5e92 167 $site = get_site();
4c4d78fb 168 $PAGE->set_context(get_context_instance(CONTEXT_SYSTEM));
a6855934
PS
169 $PAGE->navbar->add($strsearch, new moodle_url('/search/index.php'));
170 $PAGE->navbar->add($strquery, new moodle_url('/search/stats.php'));
766ccfbd 171 $PAGE->set_title($strsearch);
172 $PAGE->set_heading($site->fullname);
173 echo $OUTPUT->header();
70f52c70 174
e3c7f155 175 if (!empty($error)){
176 notice ($error);
177 }
70f52c70 178
179 echo $OUTPUT->box_start();
f7a1496e 180 echo $OUTPUT->heading($strquery);
70f52c70 181
182 echo $OUTPUT->box_start();
183
3319ef85 184 $vars = get_object_vars($adv);
70f52c70 185
3319ef85 186 if (isset($vars)) {
187 foreach ($vars as $key => $value) {
964a5e92 188 // htmlentities breaks non-ascii chars ??
00521032
PS
189 $adv->key = $value;
190 //$adv->$key = htmlentities($value);
70f52c70 191 }
3319ef85 192 }
193 ?>
3319ef85 194 <form id="query" method="get" action="query.php">
70f52c70 195 <?php
196 if (!$advanced) {
3319ef85 197 ?>
964a5e92 198 <input type="text" name="query_string" length="50" value="<?php p($query_string) ?>" />&nbsp;
199 <input type="submit" value="<?php print_string('search', 'search') ?>" /> &nbsp;
3319ef85 200 <a href="query.php?a=1"><?php print_string('advancedsearch', 'search') ?></a> |
201 <a href="stats.php"><?php print_string('statistics', 'search') ?></a>
70f52c70 202 <?php
203 }
e3c7f155 204 else {
70f52c70 205 echo $OUTPUT->box_start();
3319ef85 206 ?>
9a439947 207 <input type="hidden" name="a" value="<?php p($advanced); ?>"/>
70f52c70 208
3319ef85 209 <table border="0" cellpadding="3" cellspacing="3">
70f52c70 210
3319ef85 211 <tr>
212 <td width="240"><?php print_string('thesewordsmustappear', 'search') ?>:</td>
9a439947 213 <td><input type="text" name="mustappear" length="50" value="<?php p($adv->mustappear); ?>" /></td>
3319ef85 214 </tr>
70f52c70 215
3319ef85 216 <tr>
217 <td><?php print_string('thesewordsmustnotappear', 'search') ?>:</td>
9a439947 218 <td><input type="text" name="notappear" length="50" value="<?php p($adv->notappear); ?>" /></td>
3319ef85 219 </tr>
70f52c70 220
3319ef85 221 <tr>
222 <td><?php print_string('thesewordshelpimproverank', 'search') ?>:</td>
9a439947 223 <td><input type="text" name="canappear" length="50" value="<?php p($adv->canappear); ?>" /></td>
3319ef85 224 </tr>
70f52c70 225
3319ef85 226 <tr>
227 <td><?php print_string('whichmodulestosearch?', 'search') ?>:</td>
228 <td>
229 <select name="module">
70f52c70 230 <?php
3319ef85 231 foreach($module_types as $mod) {
232 if ($mod == $adv->module) {
233 if ($mod != 'all'){
234 print "<option value='$mod' selected=\"selected\">".get_string('modulenameplural', $mod)."</option>\n";
235 }
236 else{
237 print "<option value='$mod' selected=\"selected\">".get_string('all', 'search')."</option>\n";
238 }
70f52c70 239 }
3319ef85 240 else {
241 if ($mod != 'all'){
242 print "<option value='$mod'>".get_string('modulenameplural', $mod)."</option>\n";
243 }
244 else{
245 print "<option value='$mod'>".get_string('all', 'search')."</option>\n";
246 }
70f52c70 247 }
248 }
3319ef85 249 ?>
250 </select>
251 </td>
252 </tr>
70f52c70 253
3319ef85 254 <tr>
255 <td><?php print_string('wordsintitle', 'search') ?>:</td>
9a439947 256 <td><input type="text" name="title" length="50" value="<?php p($adv->title); ?>" /></td>
3319ef85 257 </tr>
70f52c70 258
3319ef85 259 <tr>
260 <td><?php print_string('authorname', 'search') ?>:</td>
9a439947 261 <td><input type="text" name="author" length="50" value="<?php p($adv->author); ?>" /></td>
3319ef85 262 </tr>
70f52c70 263
3319ef85 264 <tr>
9a439947 265 <td colspan="3" align="center"><br /><input type="submit" value="<?php p(get_string('search', 'search')) ?>" /></td>
3319ef85 266 </tr>
70f52c70 267
3319ef85 268 <tr>
269 <td colspan="3" align="center">
270 <table border="0" cellpadding="0" cellspacing="0">
271 <tr>
272 <td><a href="query.php"><?php print_string('normalsearch', 'search') ?></a> |</td>
273 <td>&nbsp;<a href="stats.php"><?php print_string('statistics', 'search') ?></a></td>
274 </tr>
275 </table>
276 </td>
277 </tr>
6e780562 278 </table>
3319ef85 279 <?php
70f52c70 280 echo $OUTPUT->box_end();
281 }
3319ef85 282 ?>
283 </form>
284 <br/>
70f52c70 285
4021d539 286 <div align="center">
3319ef85 287 <?php
288 print_string('searching', 'search') . ': ';
70f52c70 289
3319ef85 290 if ($sq->is_valid_index()) {
291 //use cached variable to show up-to-date index size (takes deletions into account)
292 print $CFG->search_index_size;
70f52c70 293 }
e3c7f155 294 else {
3319ef85 295 print "0";
70f52c70 296 }
297
3319ef85 298 print ' ';
299 print_string('documents', 'search');
300 print '.';
70f52c70 301
4f0c2d00 302 if (!$sq->is_valid_index() and has_capability('moodle/site:config', get_context_instance(CONTEXT_SYSTEM))) {
3319ef85 303 print '<p>' . get_string('noindexmessage', 'search') . '<a href="indexersplash.php">' . get_string('createanindex', 'search')."</a></p>\n";
70f52c70 304 }
305
3319ef85 306 ?>
307 </div>
308 <?php
70f52c70 309 echo $OUTPUT->box_end();
310
3efa38a4 311/// prints all the results in a box
312
3319ef85 313 if ($sq->is_valid()) {
70f52c70 314 echo $OUTPUT->box_start();
315
3319ef85 316 search_stopwatch();
317 $hit_count = $sq->count();
70f52c70 318
3319ef85 319 print "<br />";
70f52c70 320
decf9db1 321 print $hit_count.' '.get_string('resultsreturnedfor', 'search') . " '".s($query_string)."'.";
3319ef85 322 print "<br />";
70f52c70 323
3319ef85 324 if ($hit_count > 0) {
325 $page_links = $sq->page_numbers();
326 $hits = $sq->results();
70f52c70 327
3319ef85 328 if ($advanced) {
329 // if in advanced mode, search options are saved in the session, so
330 // we can remove the query string var from the page links, and replace
331 // it with a=1 (Advanced = on) instead
332 $page_links = preg_replace("/query_string=[^&]+/", 'a=1', $page_links);
70f52c70 333 }
334
3319ef85 335 print "<ol>";
70f52c70 336
3319ef85 337 $typestr = get_string('type', 'search');
338 $scorestr = get_string('score', 'search');
339 $authorstr = get_string('author', 'search');
3efa38a4 340
341 $searchables = search_collect_searchables(false, false);
22820396
AB
342
343 //build a list of distinct user objects needed for results listing.
344 $hitusers = array();
345 foreach ($hits as $listing) {
346 if ($listing->doctype == 'user' and !isset($hitusers[$listing->userid])) {
347 $hitusers[$listing->userid] = $DB->get_record('user', array('id' => $listing->userid));
348 }
349 }
350
70f52c70 351 foreach ($hits as $listing) {
352
22820396
AB
353 if ($listing->doctype == 'user') { // A special handle for users
354 $icon = $OUTPUT->user_picture($hitusers[$listing->userid]);
964a5e92 355 } else {
b5d0cafc 356 $iconpath = $OUTPUT->pix_url('icon', $listing->doctype);
964a5e92 357 $icon = "<img align=\"top\" src=\"".$iconpath."\" class=\"activityicon\" alt=\"\"/>";
358 }
78946b9b 359 $coursename = $DB->get_field('course', 'fullname', array('id' => $listing->courseid));
3efa38a4 360 $courseword = mb_convert_case(get_string('course', 'moodle'), MB_CASE_LOWER, 'UTF-8');
964a5e92 361 $course = ($listing->doctype != 'user') ? '<strong> ('.$courseword.': \''.$coursename.'\')</strong>' : '' ;
362
3319ef85 363 $title_post_processing_function = $listing->doctype.'_link_post_processing';
3efa38a4 364 $searchable_instance = $searchables[$listing->doctype];
365 if ($searchable_instance->location == 'internal'){
366 require_once "{$CFG->dirroot}/search/documents/{$listing->doctype}_document.php";
367 } else {
368 require_once "{$CFG->dirroot}/{$searchable_instance->location}/{$listing->doctype}/search_document.php";
369 }
3319ef85 370 if (function_exists($title_post_processing_function)) {
371 $listing->title = $title_post_processing_function($listing->title);
372 }
70f52c70 373
964a5e92 374 echo "<li value='".($listing->number + 1)."'><a href='"
3efa38a4 375 .str_replace('DEFAULT_POPUP_SETTINGS', DEFAULT_POPUP_SETTINGS ,$listing->url)
964a5e92 376 ."'>$icon $listing->title</a> $course<br />\n";
377 echo "{$typestr}: " . $listing->doctype . ", {$scorestr}: " . round($listing->score, 3);
378 if (!empty($listing->author) && !is_numeric($listing->author)){
3efa38a4 379 echo ", {$authorstr}: ".$listing->author."\n"
b93b987d 380 ."</li>\n";
381 }
70f52c70 382 }
3efa38a4 383 echo "</ol>";
384 echo $page_links;
70f52c70 385 }
386 echo $OUTPUT->box_end();
3319ef85 387 ?>
4021d539 388 <div align="center">
70f52c70 389 <?php
3319ef85 390 print_string('ittook', 'search');
70f52c70 391 search_stopwatch();
3319ef85 392 print_string('tofetchtheseresults', 'search');
393 ?>.
394 </div>
70f52c70 395
3319ef85 396 <?php
397 }
70f52c70 398 echo $OUTPUT->box_end();
daa2cd33 399 echo $OUTPUT->footer();
f7a1496e 400?>