Merge "Improve docs for Title::getInternalURL/getCanonicalURL"
[lhc/web/wiklou.git] / includes / search / SearchEngine.php
1 <?php
2 /**
3 * Basic search engine
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Search
22 */
23
24 /**
25 * @defgroup Search Search
26 */
27
28 use MediaWiki\MediaWikiServices;
29
30 /**
31 * Contain a class for special pages
32 * @ingroup Search
33 */
34 abstract class SearchEngine {
35 const DEFAULT_SORT = 'relevance';
36
37 /** @var string */
38 public $prefix = '';
39
40 /** @var int[]|null */
41 public $namespaces = [ NS_MAIN ];
42
43 /** @var int */
44 protected $limit = 10;
45
46 /** @var int */
47 protected $offset = 0;
48
49 /** @var array|string */
50 protected $searchTerms = [];
51
52 /** @var bool */
53 protected $showSuggestion = true;
54 private $sort = self::DEFAULT_SORT;
55
56 /** @var array Feature values */
57 protected $features = [];
58
59 /** @const string profile type for completionSearch */
60 const COMPLETION_PROFILE_TYPE = 'completionSearchProfile';
61
62 /** @const string profile type for query independent ranking features */
63 const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile';
64
65 /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
66 const CHARS_ALL = 1;
67
68 /** @const int flag for legalSearchChars: includes all chars allowed in a search term */
69 const CHARS_NO_SYNTAX = 2;
70
71 /**
72 * Perform a full text search query and return a result set.
73 * If full text searches are not supported or disabled, return null.
74 *
75 * As of 1.32 overriding this function is deprecated. It will
76 * be converted to final in 1.34. Override self::doSearchText().
77 *
78 * @param string $term Raw search term
79 * @return SearchResultSet|Status|null
80 */
81 public function searchText( $term ) {
82 return $this->maybePaginate( function () use ( $term ) {
83 return $this->doSearchText( $term );
84 } );
85 }
86
87 /**
88 * Perform a full text search query and return a result set.
89 *
90 * @param string $term Raw search term
91 * @return SearchResultSet|Status|null
92 * @since 1.32
93 */
94 protected function doSearchText( $term ) {
95 return null;
96 }
97
98 /**
99 * Perform a title search in the article archive.
100 * NOTE: these results still should be filtered by
101 * matching against PageArchive, permissions checks etc
102 * The results returned by this methods are only sugegstions and
103 * may not end up being shown to the user.
104 *
105 * As of 1.32 overriding this function is deprecated. It will
106 * be converted to final in 1.34. Override self::doSearchArchiveTitle().
107 *
108 * @param string $term Raw search term
109 * @return Status<Title[]>
110 * @since 1.29
111 */
112 public function searchArchiveTitle( $term ) {
113 return $this->doSearchArchiveTitle( $term );
114 }
115
116 /**
117 * Perform a title search in the article archive.
118 *
119 * @param string $term Raw search term
120 * @return Status<Title[]>
121 * @since 1.32
122 */
123 protected function doSearchArchiveTitle( $term ) {
124 return Status::newGood( [] );
125 }
126
127 /**
128 * Perform a title-only search query and return a result set.
129 * If title searches are not supported or disabled, return null.
130 * STUB
131 *
132 * As of 1.32 overriding this function is deprecated. It will
133 * be converted to final in 1.34. Override self::doSearchTitle().
134 *
135 * @param string $term Raw search term
136 * @return SearchResultSet|null
137 */
138 public function searchTitle( $term ) {
139 return $this->maybePaginate( function () use ( $term ) {
140 return $this->doSearchTitle( $term );
141 } );
142 }
143
144 /**
145 * Perform a title-only search query and return a result set.
146 *
147 * @param string $term Raw search term
148 * @return SearchResultSet|null
149 * @since 1.32
150 */
151 protected function doSearchTitle( $term ) {
152 return null;
153 }
154
155 /**
156 * Performs an overfetch and shrink operation to determine if
157 * the next page is available for search engines that do not
158 * explicitly implement their own pagination.
159 *
160 * @param Closure $fn Takes no arguments
161 * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
162 */
163 private function maybePaginate( Closure $fn ) {
164 if ( $this instanceof PaginatingSearchEngine ) {
165 return $fn();
166 }
167 $this->limit++;
168 try {
169 $resultSetOrStatus = $fn();
170 } finally {
171 $this->limit--;
172 }
173
174 $resultSet = null;
175 if ( $resultSetOrStatus instanceof SearchResultSet ) {
176 $resultSet = $resultSetOrStatus;
177 } elseif ( $resultSetOrStatus instanceof Status &&
178 $resultSetOrStatus->getValue() instanceof SearchResultSet
179 ) {
180 $resultSet = $resultSetOrStatus->getValue();
181 }
182 if ( $resultSet ) {
183 $resultSet->shrink( $this->limit );
184 }
185
186 return $resultSetOrStatus;
187 }
188
189 /**
190 * @since 1.18
191 * @param string $feature
192 * @return bool
193 */
194 public function supports( $feature ) {
195 switch ( $feature ) {
196 case 'search-update':
197 return true;
198 case 'title-suffix-filter':
199 default:
200 return false;
201 }
202 }
203
204 /**
205 * Way to pass custom data for engines
206 * @since 1.18
207 * @param string $feature
208 * @param mixed $data
209 */
210 public function setFeatureData( $feature, $data ) {
211 $this->features[$feature] = $data;
212 }
213
214 /**
215 * Way to retrieve custom data set by setFeatureData
216 * or by the engine itself.
217 * @since 1.29
218 * @param string $feature feature name
219 * @return mixed the feature value or null if unset
220 */
221 public function getFeatureData( $feature ) {
222 return $this->features[$feature] ?? null;
223 }
224
225 /**
226 * When overridden in derived class, performs database-specific conversions
227 * on text to be used for searching or updating search index.
228 * Default implementation does nothing (simply returns $string).
229 *
230 * @param string $string String to process
231 * @return string
232 */
233 public function normalizeText( $string ) {
234 // Some languages such as Chinese require word segmentation
235 return MediaWikiServices::getInstance()->getContentLanguage()->segmentByWord( $string );
236 }
237
238 /**
239 * Transform search term in cases when parts of the query came as different
240 * GET params (when supported), e.g. for prefix queries:
241 * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
242 * @param string $term
243 * @return string
244 * @deprecated since 1.32 this should now be handled internally by the
245 * search engine
246 */
247 public function transformSearchTerm( $term ) {
248 return $term;
249 }
250
251 /**
252 * Get service class to finding near matches.
253 * @param Config $config Configuration to use for the matcher.
254 * @return SearchNearMatcher
255 */
256 public function getNearMatcher( Config $config ) {
257 return new SearchNearMatcher( $config,
258 MediaWikiServices::getInstance()->getContentLanguage() );
259 }
260
261 /**
262 * Get near matcher for default SearchEngine.
263 * @return SearchNearMatcher
264 */
265 protected static function defaultNearMatcher() {
266 $services = MediaWikiServices::getInstance();
267 $config = $services->getMainConfig();
268 return $services->newSearchEngine()->getNearMatcher( $config );
269 }
270
271 /**
272 * If an exact title match can be found, or a very slightly close match,
273 * return the title. If no match, returns NULL.
274 * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
275 * @param string $searchterm
276 * @return Title
277 */
278 public static function getNearMatch( $searchterm ) {
279 return static::defaultNearMatcher()->getNearMatch( $searchterm );
280 }
281
282 /**
283 * Get chars legal for search
284 * NOTE: usage as static is deprecated and preserved only as BC measure
285 * @param int $type type of search chars (see self::CHARS_ALL
286 * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
287 * @return string
288 */
289 public static function legalSearchChars( $type = self::CHARS_ALL ) {
290 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
291 }
292
293 /**
294 * Set the maximum number of results to return
295 * and how many to skip before returning the first.
296 *
297 * @param int $limit
298 * @param int $offset
299 */
300 function setLimitOffset( $limit, $offset = 0 ) {
301 $this->limit = intval( $limit );
302 $this->offset = intval( $offset );
303 }
304
305 /**
306 * Set which namespaces the search should include.
307 * Give an array of namespace index numbers.
308 *
309 * @param int[]|null $namespaces
310 */
311 function setNamespaces( $namespaces ) {
312 if ( $namespaces ) {
313 // Filter namespaces to only keep valid ones
314 $validNs = $this->searchableNamespaces();
315 $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
316 return $ns < 0 || isset( $validNs[$ns] );
317 } );
318 } else {
319 $namespaces = [];
320 }
321 $this->namespaces = $namespaces;
322 }
323
324 /**
325 * Set whether the searcher should try to build a suggestion. Note: some searchers
326 * don't support building a suggestion in the first place and others don't respect
327 * this flag.
328 *
329 * @param bool $showSuggestion Should the searcher try to build suggestions
330 */
331 function setShowSuggestion( $showSuggestion ) {
332 $this->showSuggestion = $showSuggestion;
333 }
334
335 /**
336 * Get the valid sort directions. All search engines support 'relevance' but others
337 * might support more. The default in all implementations must be 'relevance.'
338 *
339 * @since 1.25
340 * @return string[] the valid sort directions for setSort
341 */
342 public function getValidSorts() {
343 return [ self::DEFAULT_SORT ];
344 }
345
346 /**
347 * Set the sort direction of the search results. Must be one returned by
348 * SearchEngine::getValidSorts()
349 *
350 * @since 1.25
351 * @throws InvalidArgumentException
352 * @param string $sort sort direction for query result
353 */
354 public function setSort( $sort ) {
355 if ( !in_array( $sort, $this->getValidSorts() ) ) {
356 throw new InvalidArgumentException( "Invalid sort: $sort. " .
357 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
358 }
359 $this->sort = $sort;
360 }
361
362 /**
363 * Get the sort direction of the search results
364 *
365 * @since 1.25
366 * @return string
367 */
368 public function getSort() {
369 return $this->sort;
370 }
371
372 /**
373 * Parse some common prefixes: all (search everything)
374 * or namespace names and set the list of namespaces
375 * of this class accordingly.
376 *
377 * @deprecated since 1.32; should be handled internally by the search engine
378 * @param string $query
379 * @return string
380 */
381 function replacePrefixes( $query ) {
382 return $query;
383 }
384
385 /**
386 * Parse some common prefixes: all (search everything)
387 * or namespace names
388 *
389 * @param string $query
390 * @param bool $withAllKeyword activate support of the "all:" keyword and its
391 * translations to activate searching on all namespaces.
392 * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
393 * if classic namespace identification did not match.
394 * @return false|array false if no namespace was extracted, an array
395 * with the parsed query at index 0 and an array of namespaces at index
396 * 1 (or null for all namespaces).
397 * @throws FatalError
398 * @throws MWException
399 */
400 public static function parseNamespacePrefixes(
401 $query,
402 $withAllKeyword = true,
403 $withPrefixSearchExtractNamespaceHook = false
404 ) {
405 $parsed = $query;
406 if ( strpos( $query, ':' ) === false ) { // nothing to do
407 return false;
408 }
409 $extractedNamespace = null;
410
411 $allQuery = false;
412 if ( $withAllKeyword ) {
413 $allkeywords = [];
414
415 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
416 // force all: so that we have a common syntax for all the wikis
417 if ( !in_array( 'all:', $allkeywords ) ) {
418 $allkeywords[] = 'all:';
419 }
420
421 foreach ( $allkeywords as $kw ) {
422 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
423 $extractedNamespace = null;
424 $parsed = substr( $query, strlen( $kw ) );
425 $allQuery = true;
426 break;
427 }
428 }
429 }
430
431 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
432 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
433 $index = MediaWikiServices::getInstance()->getContentLanguage()->getNsIndex( $prefix );
434 if ( $index !== false ) {
435 $extractedNamespace = [ $index ];
436 $parsed = substr( $query, strlen( $prefix ) + 1 );
437 } elseif ( $withPrefixSearchExtractNamespaceHook ) {
438 $hookNamespaces = [ NS_MAIN ];
439 $hookQuery = $query;
440 Hooks::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
441 if ( $hookQuery !== $query ) {
442 $parsed = $hookQuery;
443 $extractedNamespace = $hookNamespaces;
444 } else {
445 return false;
446 }
447 } else {
448 return false;
449 }
450 }
451
452 return [ $parsed, $extractedNamespace ];
453 }
454
455 /**
456 * Find snippet highlight settings for all users
457 * @return array Contextlines, contextchars
458 */
459 public static function userHighlightPrefs() {
460 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
461 $contextchars = 75; // same as above.... :P
462 return [ $contextlines, $contextchars ];
463 }
464
465 /**
466 * Create or update the search index record for the given page.
467 * Title and text should be pre-processed.
468 * STUB
469 *
470 * @param int $id
471 * @param string $title
472 * @param string $text
473 */
474 function update( $id, $title, $text ) {
475 // no-op
476 }
477
478 /**
479 * Update a search index record's title only.
480 * Title should be pre-processed.
481 * STUB
482 *
483 * @param int $id
484 * @param string $title
485 */
486 function updateTitle( $id, $title ) {
487 // no-op
488 }
489
490 /**
491 * Delete an indexed page
492 * Title should be pre-processed.
493 * STUB
494 *
495 * @param int $id Page id that was deleted
496 * @param string $title Title of page that was deleted
497 */
498 function delete( $id, $title ) {
499 // no-op
500 }
501
502 /**
503 * Get the raw text for updating the index from a content object
504 * Nicer search backends could possibly do something cooler than
505 * just returning raw text
506 *
507 * @todo This isn't ideal, we'd really like to have content-specific handling here
508 * @param Title $t Title we're indexing
509 * @param Content|null $c Content of the page to index
510 * @return string
511 */
512 public function getTextFromContent( Title $t, Content $c = null ) {
513 return $c ? $c->getTextForSearchIndex() : '';
514 }
515
516 /**
517 * If an implementation of SearchEngine handles all of its own text processing
518 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
519 * rather silly handling, it should return true here instead.
520 *
521 * @return bool
522 */
523 public function textAlreadyUpdatedForIndex() {
524 return false;
525 }
526
527 /**
528 * Makes search simple string if it was namespaced.
529 * Sets namespaces of the search to namespaces extracted from string.
530 * @param string $search
531 * @return string Simplified search string
532 */
533 protected function normalizeNamespaces( $search ) {
534 $queryAndNs = self::parseNamespacePrefixes( $search, false, true );
535 if ( $queryAndNs !== false ) {
536 $this->setNamespaces( $queryAndNs[1] );
537 return $queryAndNs[0];
538 }
539 return $search;
540 }
541
542 /**
543 * Perform an overfetch of completion search results. This allows
544 * determining if another page of results is available.
545 *
546 * @param string $search
547 * @return SearchSuggestionSet
548 */
549 protected function completionSearchBackendOverfetch( $search ) {
550 $this->limit++;
551 try {
552 return $this->completionSearchBackend( $search );
553 } finally {
554 $this->limit--;
555 }
556 }
557
558 /**
559 * Perform a completion search.
560 * Does not resolve namespaces and does not check variants.
561 * Search engine implementations may want to override this function.
562 * @param string $search
563 * @return SearchSuggestionSet
564 */
565 protected function completionSearchBackend( $search ) {
566 $results = [];
567
568 $search = trim( $search );
569
570 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
571 !Hooks::run( 'PrefixSearchBackend',
572 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
573 ) ) {
574 // False means hook worked.
575 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
576
577 return SearchSuggestionSet::fromStrings( $results );
578 } else {
579 // Hook did not do the job, use default simple search
580 $results = $this->simplePrefixSearch( $search );
581 return SearchSuggestionSet::fromTitles( $results );
582 }
583 }
584
585 /**
586 * Perform a completion search.
587 * @param string $search
588 * @return SearchSuggestionSet
589 */
590 public function completionSearch( $search ) {
591 if ( trim( $search ) === '' ) {
592 return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
593 }
594 $search = $this->normalizeNamespaces( $search );
595 $suggestions = $this->completionSearchBackendOverfetch( $search );
596 return $this->processCompletionResults( $search, $suggestions );
597 }
598
599 /**
600 * Perform a completion search with variants.
601 * @param string $search
602 * @return SearchSuggestionSet
603 */
604 public function completionSearchWithVariants( $search ) {
605 if ( trim( $search ) === '' ) {
606 return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
607 }
608 $search = $this->normalizeNamespaces( $search );
609
610 $results = $this->completionSearchBackendOverfetch( $search );
611 $fallbackLimit = 1 + $this->limit - $results->getSize();
612 if ( $fallbackLimit > 0 ) {
613 $fallbackSearches = MediaWikiServices::getInstance()->getContentLanguage()->
614 autoConvertToAllVariants( $search );
615 $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
616
617 foreach ( $fallbackSearches as $fbs ) {
618 $this->setLimitOffset( $fallbackLimit );
619 $fallbackSearchResult = $this->completionSearch( $fbs );
620 $results->appendAll( $fallbackSearchResult );
621 $fallbackLimit -= $fallbackSearchResult->getSize();
622 if ( $fallbackLimit <= 0 ) {
623 break;
624 }
625 }
626 }
627 return $this->processCompletionResults( $search, $results );
628 }
629
630 /**
631 * Extract titles from completion results
632 * @param SearchSuggestionSet $completionResults
633 * @return Title[]
634 */
635 public function extractTitles( SearchSuggestionSet $completionResults ) {
636 return $completionResults->map( function ( SearchSuggestion $sugg ) {
637 return $sugg->getSuggestedTitle();
638 } );
639 }
640
641 /**
642 * Process completion search results.
643 * Resolves the titles and rescores.
644 * @param string $search
645 * @param SearchSuggestionSet $suggestions
646 * @return SearchSuggestionSet
647 */
648 protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
649 // We over-fetched to determine pagination. Shrink back down if we have extra results
650 // and mark if pagination is possible
651 $suggestions->shrink( $this->limit );
652
653 $search = trim( $search );
654 // preload the titles with LinkBatch
655 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
656 return $sugg->getSuggestedTitle();
657 } ) );
658 $lb->setCaller( __METHOD__ );
659 $lb->execute();
660
661 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
662 return $sugg->getSuggestedTitle()->isKnown();
663 } );
664 if ( $diff > 0 ) {
665 MediaWikiServices::getInstance()->getStatsdDataFactory()
666 ->updateCount( 'search.completion.missing', $diff );
667 }
668
669 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
670 return $sugg->getSuggestedTitle()->getPrefixedText();
671 } );
672
673 if ( $this->offset === 0 ) {
674 // Rescore results with an exact title match
675 // NOTE: in some cases like cross-namespace redirects
676 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
677 // backends like Cirrus will return no results. We should still
678 // try an exact title match to workaround this limitation
679 $rescorer = new SearchExactMatchRescorer();
680 $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
681 } else {
682 // No need to rescore if offset is not 0
683 // The exact match must have been returned at position 0
684 // if it existed.
685 $rescoredResults = $results;
686 }
687
688 if ( count( $rescoredResults ) > 0 ) {
689 $found = array_search( $rescoredResults[0], $results );
690 if ( $found === false ) {
691 // If the first result is not in the previous array it
692 // means that we found a new exact match
693 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
694 $suggestions->prepend( $exactMatch );
695 $suggestions->shrink( $this->limit );
696 } else {
697 // if the first result is not the same we need to rescore
698 if ( $found > 0 ) {
699 $suggestions->rescore( $found );
700 }
701 }
702 }
703
704 return $suggestions;
705 }
706
707 /**
708 * Simple prefix search for subpages.
709 * @param string $search
710 * @return Title[]
711 */
712 public function defaultPrefixSearch( $search ) {
713 if ( trim( $search ) === '' ) {
714 return [];
715 }
716
717 $search = $this->normalizeNamespaces( $search );
718 return $this->simplePrefixSearch( $search );
719 }
720
721 /**
722 * Call out to simple search backend.
723 * Defaults to TitlePrefixSearch.
724 * @param string $search
725 * @return Title[]
726 */
727 protected function simplePrefixSearch( $search ) {
728 // Use default database prefix search
729 $backend = new TitlePrefixSearch;
730 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
731 }
732
733 /**
734 * Make a list of searchable namespaces and their canonical names.
735 * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
736 * @return array
737 */
738 public static function searchableNamespaces() {
739 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
740 }
741
742 /**
743 * Extract default namespaces to search from the given user's
744 * settings, returning a list of index numbers.
745 * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
746 * @param user $user
747 * @return array
748 */
749 public static function userNamespaces( $user ) {
750 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
751 }
752
753 /**
754 * An array of namespaces indexes to be searched by default
755 * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
756 * @return array
757 */
758 public static function defaultNamespaces() {
759 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
760 }
761
762 /**
763 * Get a list of namespace names useful for showing in tooltips
764 * and preferences
765 * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
766 * @param array $namespaces
767 * @return array
768 */
769 public static function namespacesAsText( $namespaces ) {
770 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
771 }
772
773 /**
774 * Load up the appropriate search engine class for the currently
775 * active database backend, and return a configured instance.
776 * @deprecated since 1.27; Use SearchEngineFactory::create
777 * @param string $type Type of search backend, if not the default
778 * @return SearchEngine
779 */
780 public static function create( $type = null ) {
781 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
782 }
783
784 /**
785 * Return the search engines we support. If only $wgSearchType
786 * is set, it'll be an array of just that one item.
787 * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
788 * @return array
789 */
790 public static function getSearchTypes() {
791 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
792 }
793
794 /**
795 * Get a list of supported profiles.
796 * Some search engine implementations may expose specific profiles to fine-tune
797 * its behaviors.
798 * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
799 * The array returned by this function contains the following keys:
800 * - name: the profile name to use with setFeatureData
801 * - desc-message: the i18n description
802 * - default: set to true if this profile is the default
803 *
804 * @since 1.28
805 * @param string $profileType the type of profiles
806 * @param User|null $user the user requesting the list of profiles
807 * @return array|null the list of profiles or null if none available
808 */
809 public function getProfiles( $profileType, User $user = null ) {
810 return null;
811 }
812
813 /**
814 * Create a search field definition.
815 * Specific search engines should override this method to create search fields.
816 * @param string $name
817 * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
818 * @return SearchIndexField
819 * @since 1.28
820 */
821 public function makeSearchFieldMapping( $name, $type ) {
822 return new NullIndexField();
823 }
824
825 /**
826 * Get fields for search index
827 * @since 1.28
828 * @return SearchIndexField[] Index field definitions for all content handlers
829 */
830 public function getSearchIndexFields() {
831 $models = ContentHandler::getContentModels();
832 $fields = [];
833 $seenHandlers = new SplObjectStorage();
834 foreach ( $models as $model ) {
835 try {
836 $handler = ContentHandler::getForModelID( $model );
837 }
838 catch ( MWUnknownContentModelException $e ) {
839 // If we can find no handler, ignore it
840 continue;
841 }
842 // Several models can have the same handler, so avoid processing it repeatedly
843 if ( $seenHandlers->contains( $handler ) ) {
844 // We already did this one
845 continue;
846 }
847 $seenHandlers->attach( $handler );
848 $handlerFields = $handler->getFieldsForSearchIndex( $this );
849 foreach ( $handlerFields as $fieldName => $fieldData ) {
850 if ( empty( $fields[$fieldName] ) ) {
851 $fields[$fieldName] = $fieldData;
852 } else {
853 // TODO: do we allow some clashes with the same type or reject all of them?
854 $mergeDef = $fields[$fieldName]->merge( $fieldData );
855 if ( !$mergeDef ) {
856 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
857 }
858 $fields[$fieldName] = $mergeDef;
859 }
860 }
861 }
862 // Hook to allow extensions to produce search mapping fields
863 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
864 return $fields;
865 }
866
867 /**
868 * Augment search results with extra data.
869 *
870 * @param SearchResultSet $resultSet
871 */
872 public function augmentSearchResults( SearchResultSet $resultSet ) {
873 $setAugmentors = [];
874 $rowAugmentors = [];
875 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
876 if ( !$setAugmentors && !$rowAugmentors ) {
877 // We're done here
878 return;
879 }
880
881 // Convert row augmentors to set augmentor
882 foreach ( $rowAugmentors as $name => $row ) {
883 if ( isset( $setAugmentors[$name] ) ) {
884 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
885 }
886 $setAugmentors[$name] = new PerRowAugmentor( $row );
887 }
888
889 foreach ( $setAugmentors as $name => $augmentor ) {
890 $data = $augmentor->augmentAll( $resultSet );
891 if ( $data ) {
892 $resultSet->setAugmentedData( $name, $data );
893 }
894 }
895 }
896 }