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.
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.
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
25 * @defgroup Search Search
28 use MediaWiki\MediaWikiServices
;
31 * Contain a class for special pages
34 abstract class SearchEngine
{
35 const DEFAULT_SORT
= 'relevance';
40 /** @var int[]|null */
41 public $namespaces = [ NS_MAIN
];
44 protected $limit = 10;
47 protected $offset = 0;
49 /** @var array|string */
50 protected $searchTerms = [];
53 protected $showSuggestion = true;
54 private $sort = self
::DEFAULT_SORT
;
56 /** @var array Feature values */
57 protected $features = [];
59 /** Profile type for completionSearch */
60 const COMPLETION_PROFILE_TYPE
= 'completionSearchProfile';
62 /** Profile type for query independent ranking features */
63 const FT_QUERY_INDEP_PROFILE_TYPE
= 'fulltextQueryIndepProfile';
65 /** Integer flag for legalSearchChars: includes all chars allowed in a search query */
68 /** Integer flag for legalSearchChars: includes all chars allowed in a search term */
69 const CHARS_NO_SYNTAX
= 2;
72 * Perform a full text search query and return a result set.
73 * If full text searches are not supported or disabled, return null.
75 * As of 1.32 overriding this function is deprecated. It will
76 * be converted to final in 1.34. Override self::doSearchText().
78 * @param string $term Raw search term
79 * @return SearchResultSet|Status|null
81 public function searchText( $term ) {
82 return $this->maybePaginate( function () use ( $term ) {
83 return $this->doSearchText( $term );
88 * Perform a full text search query and return a result set.
90 * @param string $term Raw search term
91 * @return SearchResultSet|Status|null
94 protected function doSearchText( $term ) {
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.
105 * As of 1.32 overriding this function is deprecated. It will
106 * be converted to final in 1.34. Override self::doSearchArchiveTitle().
108 * @param string $term Raw search term
109 * @return Status<Title[]>
112 public function searchArchiveTitle( $term ) {
113 return $this->doSearchArchiveTitle( $term );
117 * Perform a title search in the article archive.
119 * @param string $term Raw search term
120 * @return Status<Title[]>
123 protected function doSearchArchiveTitle( $term ) {
124 return Status
::newGood( [] );
128 * Perform a title-only search query and return a result set.
129 * If title searches are not supported or disabled, return null.
132 * As of 1.32 overriding this function is deprecated. It will
133 * be converted to final in 1.34. Override self::doSearchTitle().
135 * @param string $term Raw search term
136 * @return SearchResultSet|null
138 public function searchTitle( $term ) {
139 return $this->maybePaginate( function () use ( $term ) {
140 return $this->doSearchTitle( $term );
145 * Perform a title-only search query and return a result set.
147 * @param string $term Raw search term
148 * @return SearchResultSet|null
151 protected function doSearchTitle( $term ) {
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.
160 * @param Closure $fn Takes no arguments
161 * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
163 private function maybePaginate( Closure
$fn ) {
164 if ( $this instanceof PaginatingSearchEngine
) {
169 $resultSetOrStatus = $fn();
175 if ( $resultSetOrStatus instanceof SearchResultSet
) {
176 $resultSet = $resultSetOrStatus;
177 } elseif ( $resultSetOrStatus instanceof Status
&&
178 $resultSetOrStatus->getValue() instanceof SearchResultSet
180 $resultSet = $resultSetOrStatus->getValue();
183 $resultSet->shrink( $this->limit
);
186 return $resultSetOrStatus;
191 * @param string $feature
194 public function supports( $feature ) {
195 switch ( $feature ) {
196 case 'search-update':
198 case 'title-suffix-filter':
205 * Way to pass custom data for engines
207 * @param string $feature
210 public function setFeatureData( $feature, $data ) {
211 $this->features
[$feature] = $data;
215 * Way to retrieve custom data set by setFeatureData
216 * or by the engine itself.
218 * @param string $feature feature name
219 * @return mixed the feature value or null if unset
221 public function getFeatureData( $feature ) {
222 return $this->features
[$feature] ??
null;
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).
230 * @param string $string String to process
233 public function normalizeText( $string ) {
234 // Some languages such as Chinese require word segmentation
235 return MediaWikiServices
::getInstance()->getContentLanguage()->segmentByWord( $string );
239 * Get service class to finding near matches.
240 * @param Config $config Configuration to use for the matcher.
241 * @return SearchNearMatcher
243 public function getNearMatcher( Config
$config ) {
244 return new SearchNearMatcher( $config,
245 MediaWikiServices
::getInstance()->getContentLanguage() );
249 * Get near matcher for default SearchEngine.
250 * @return SearchNearMatcher
252 protected static function defaultNearMatcher() {
253 $services = MediaWikiServices
::getInstance();
254 $config = $services->getMainConfig();
255 return $services->newSearchEngine()->getNearMatcher( $config );
259 * Get chars legal for search
260 * @param int $type type of search chars (see self::CHARS_ALL
261 * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
264 public function legalSearchChars( $type = self
::CHARS_ALL
) {
265 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
269 * Set the maximum number of results to return
270 * and how many to skip before returning the first.
275 function setLimitOffset( $limit, $offset = 0 ) {
276 $this->limit
= intval( $limit );
277 $this->offset
= intval( $offset );
281 * Set which namespaces the search should include.
282 * Give an array of namespace index numbers.
284 * @param int[]|null $namespaces
286 function setNamespaces( $namespaces ) {
288 // Filter namespaces to only keep valid ones
289 $validNs = MediaWikiServices
::getInstance()->getSearchEngineConfig()->searchableNamespaces();
290 $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
291 return $ns < 0 ||
isset( $validNs[$ns] );
296 $this->namespaces
= $namespaces;
300 * Set whether the searcher should try to build a suggestion. Note: some searchers
301 * don't support building a suggestion in the first place and others don't respect
304 * @param bool $showSuggestion Should the searcher try to build suggestions
306 function setShowSuggestion( $showSuggestion ) {
307 $this->showSuggestion
= $showSuggestion;
311 * Get the valid sort directions. All search engines support 'relevance' but others
312 * might support more. The default in all implementations must be 'relevance.'
315 * @return string[] the valid sort directions for setSort
317 public function getValidSorts() {
318 return [ self
::DEFAULT_SORT
];
322 * Set the sort direction of the search results. Must be one returned by
323 * SearchEngine::getValidSorts()
326 * @throws InvalidArgumentException
327 * @param string $sort sort direction for query result
329 public function setSort( $sort ) {
330 if ( !in_array( $sort, $this->getValidSorts() ) ) {
331 throw new InvalidArgumentException( "Invalid sort: $sort. " .
332 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
338 * Get the sort direction of the search results
343 public function getSort() {
348 * Parse some common prefixes: all (search everything)
349 * or namespace names and set the list of namespaces
350 * of this class accordingly.
352 * @deprecated since 1.32; should be handled internally by the search engine
353 * @param string $query
356 function replacePrefixes( $query ) {
361 * Parse some common prefixes: all (search everything)
364 * @param string $query
365 * @param bool $withAllKeyword activate support of the "all:" keyword and its
366 * translations to activate searching on all namespaces.
367 * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
368 * if classic namespace identification did not match.
369 * @return false|array false if no namespace was extracted, an array
370 * with the parsed query at index 0 and an array of namespaces at index
371 * 1 (or null for all namespaces).
373 * @throws MWException
375 public static function parseNamespacePrefixes(
377 $withAllKeyword = true,
378 $withPrefixSearchExtractNamespaceHook = false
381 if ( strpos( $query, ':' ) === false ) { // nothing to do
384 $extractedNamespace = null;
387 if ( $withAllKeyword ) {
390 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
391 // force all: so that we have a common syntax for all the wikis
392 if ( !in_array( 'all:', $allkeywords ) ) {
393 $allkeywords[] = 'all:';
396 foreach ( $allkeywords as $kw ) {
397 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
398 $extractedNamespace = null;
399 $parsed = substr( $query, strlen( $kw ) );
406 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
407 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
408 $index = MediaWikiServices
::getInstance()->getContentLanguage()->getNsIndex( $prefix );
409 if ( $index !== false ) {
410 $extractedNamespace = [ $index ];
411 $parsed = substr( $query, strlen( $prefix ) +
1 );
412 } elseif ( $withPrefixSearchExtractNamespaceHook ) {
413 $hookNamespaces = [ NS_MAIN
];
415 Hooks
::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
416 if ( $hookQuery !== $query ) {
417 $parsed = $hookQuery;
418 $extractedNamespace = $hookNamespaces;
427 return [ $parsed, $extractedNamespace ];
431 * Find snippet highlight settings for all users
432 * @return array Contextlines, contextchars
434 public static function userHighlightPrefs() {
435 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
436 $contextchars = 75; // same as above.... :P
437 return [ $contextlines, $contextchars ];
441 * Create or update the search index record for the given page.
442 * Title and text should be pre-processed.
446 * @param string $title
447 * @param string $text
449 function update( $id, $title, $text ) {
454 * Update a search index record's title only.
455 * Title should be pre-processed.
459 * @param string $title
461 function updateTitle( $id, $title ) {
466 * Delete an indexed page
467 * Title should be pre-processed.
470 * @param int $id Page id that was deleted
471 * @param string $title Title of page that was deleted
473 function delete( $id, $title ) {
478 * Get the raw text for updating the index from a content object
479 * Nicer search backends could possibly do something cooler than
480 * just returning raw text
482 * @todo This isn't ideal, we'd really like to have content-specific handling here
483 * @param Title $t Title we're indexing
484 * @param Content|null $c Content of the page to index
487 public function getTextFromContent( Title
$t, Content
$c = null ) {
488 return $c ?
$c->getTextForSearchIndex() : '';
492 * If an implementation of SearchEngine handles all of its own text processing
493 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
494 * rather silly handling, it should return true here instead.
498 public function textAlreadyUpdatedForIndex() {
503 * Makes search simple string if it was namespaced.
504 * Sets namespaces of the search to namespaces extracted from string.
505 * @param string $search
506 * @return string Simplified search string
508 protected function normalizeNamespaces( $search ) {
509 $queryAndNs = self
::parseNamespacePrefixes( $search, false, true );
510 if ( $queryAndNs !== false ) {
511 $this->setNamespaces( $queryAndNs[1] );
512 return $queryAndNs[0];
518 * Perform an overfetch of completion search results. This allows
519 * determining if another page of results is available.
521 * @param string $search
522 * @return SearchSuggestionSet
524 protected function completionSearchBackendOverfetch( $search ) {
527 return $this->completionSearchBackend( $search );
534 * Perform a completion search.
535 * Does not resolve namespaces and does not check variants.
536 * Search engine implementations may want to override this function.
537 * @param string $search
538 * @return SearchSuggestionSet
540 protected function completionSearchBackend( $search ) {
543 $search = trim( $search );
545 if ( !in_array( NS_SPECIAL
, $this->namespaces
) && // We do not run hook on Special: search
546 !Hooks
::run( 'PrefixSearchBackend',
547 [ $this->namespaces
, $search, $this->limit
, &$results, $this->offset
]
549 // False means hook worked.
550 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
552 return SearchSuggestionSet
::fromStrings( $results );
554 // Hook did not do the job, use default simple search
555 $results = $this->simplePrefixSearch( $search );
556 return SearchSuggestionSet
::fromTitles( $results );
561 * Perform a completion search.
562 * @param string $search
563 * @return SearchSuggestionSet
565 public function completionSearch( $search ) {
566 if ( trim( $search ) === '' ) {
567 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
569 $search = $this->normalizeNamespaces( $search );
570 $suggestions = $this->completionSearchBackendOverfetch( $search );
571 return $this->processCompletionResults( $search, $suggestions );
575 * Perform a completion search with variants.
576 * @param string $search
577 * @return SearchSuggestionSet
579 public function completionSearchWithVariants( $search ) {
580 if ( trim( $search ) === '' ) {
581 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
583 $search = $this->normalizeNamespaces( $search );
585 $results = $this->completionSearchBackendOverfetch( $search );
586 $fallbackLimit = 1 +
$this->limit
- $results->getSize();
587 if ( $fallbackLimit > 0 ) {
588 $fallbackSearches = MediaWikiServices
::getInstance()->getContentLanguage()->
589 autoConvertToAllVariants( $search );
590 $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
592 foreach ( $fallbackSearches as $fbs ) {
593 $this->setLimitOffset( $fallbackLimit );
594 $fallbackSearchResult = $this->completionSearch( $fbs );
595 $results->appendAll( $fallbackSearchResult );
596 $fallbackLimit -= $fallbackSearchResult->getSize();
597 if ( $fallbackLimit <= 0 ) {
602 return $this->processCompletionResults( $search, $results );
606 * Extract titles from completion results
607 * @param SearchSuggestionSet $completionResults
610 public function extractTitles( SearchSuggestionSet
$completionResults ) {
611 return $completionResults->map( function ( SearchSuggestion
$sugg ) {
612 return $sugg->getSuggestedTitle();
617 * Process completion search results.
618 * Resolves the titles and rescores.
619 * @param string $search
620 * @param SearchSuggestionSet $suggestions
621 * @return SearchSuggestionSet
623 protected function processCompletionResults( $search, SearchSuggestionSet
$suggestions ) {
624 // We over-fetched to determine pagination. Shrink back down if we have extra results
625 // and mark if pagination is possible
626 $suggestions->shrink( $this->limit
);
628 $search = trim( $search );
629 // preload the titles with LinkBatch
630 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion
$sugg ) {
631 return $sugg->getSuggestedTitle();
633 $lb->setCaller( __METHOD__
);
636 $diff = $suggestions->filter( function ( SearchSuggestion
$sugg ) {
637 return $sugg->getSuggestedTitle()->isKnown();
640 MediaWikiServices
::getInstance()->getStatsdDataFactory()
641 ->updateCount( 'search.completion.missing', $diff );
644 $results = $suggestions->map( function ( SearchSuggestion
$sugg ) {
645 return $sugg->getSuggestedTitle()->getPrefixedText();
648 if ( $this->offset
=== 0 ) {
649 // Rescore results with an exact title match
650 // NOTE: in some cases like cross-namespace redirects
651 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
652 // backends like Cirrus will return no results. We should still
653 // try an exact title match to workaround this limitation
654 $rescorer = new SearchExactMatchRescorer();
655 $rescoredResults = $rescorer->rescore( $search, $this->namespaces
, $results, $this->limit
);
657 // No need to rescore if offset is not 0
658 // The exact match must have been returned at position 0
660 $rescoredResults = $results;
663 if ( count( $rescoredResults ) > 0 ) {
664 $found = array_search( $rescoredResults[0], $results );
665 if ( $found === false ) {
666 // If the first result is not in the previous array it
667 // means that we found a new exact match
668 $exactMatch = SearchSuggestion
::fromTitle( 0, Title
::newFromText( $rescoredResults[0] ) );
669 $suggestions->prepend( $exactMatch );
670 $suggestions->shrink( $this->limit
);
672 // if the first result is not the same we need to rescore
674 $suggestions->rescore( $found );
683 * Simple prefix search for subpages.
684 * @param string $search
687 public function defaultPrefixSearch( $search ) {
688 if ( trim( $search ) === '' ) {
692 $search = $this->normalizeNamespaces( $search );
693 return $this->simplePrefixSearch( $search );
697 * Call out to simple search backend.
698 * Defaults to TitlePrefixSearch.
699 * @param string $search
702 protected function simplePrefixSearch( $search ) {
703 // Use default database prefix search
704 $backend = new TitlePrefixSearch
;
705 return $backend->defaultSearchBackend( $this->namespaces
, $search, $this->limit
, $this->offset
);
709 * Get a list of supported profiles.
710 * Some search engine implementations may expose specific profiles to fine-tune
712 * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
713 * The array returned by this function contains the following keys:
714 * - name: the profile name to use with setFeatureData
715 * - desc-message: the i18n description
716 * - default: set to true if this profile is the default
719 * @param string $profileType the type of profiles
720 * @param User|null $user the user requesting the list of profiles
721 * @return array|null the list of profiles or null if none available
723 public function getProfiles( $profileType, User
$user = null ) {
728 * Create a search field definition.
729 * Specific search engines should override this method to create search fields.
730 * @param string $name
731 * @param string $type One of the types in SearchIndexField::INDEX_TYPE_*
732 * @return SearchIndexField
735 public function makeSearchFieldMapping( $name, $type ) {
736 return new NullIndexField();
740 * Get fields for search index
742 * @return SearchIndexField[] Index field definitions for all content handlers
744 public function getSearchIndexFields() {
745 $models = ContentHandler
::getContentModels();
747 $seenHandlers = new SplObjectStorage();
748 foreach ( $models as $model ) {
750 $handler = ContentHandler
::getForModelID( $model );
752 catch ( MWUnknownContentModelException
$e ) {
753 // If we can find no handler, ignore it
756 // Several models can have the same handler, so avoid processing it repeatedly
757 if ( $seenHandlers->contains( $handler ) ) {
758 // We already did this one
761 $seenHandlers->attach( $handler );
762 $handlerFields = $handler->getFieldsForSearchIndex( $this );
763 foreach ( $handlerFields as $fieldName => $fieldData ) {
764 if ( empty( $fields[$fieldName] ) ) {
765 $fields[$fieldName] = $fieldData;
767 // TODO: do we allow some clashes with the same type or reject all of them?
768 $mergeDef = $fields[$fieldName]->merge( $fieldData );
770 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
772 $fields[$fieldName] = $mergeDef;
776 // Hook to allow extensions to produce search mapping fields
777 Hooks
::run( 'SearchIndexFields', [ &$fields, $this ] );
782 * Augment search results with extra data.
784 * @param SearchResultSet $resultSet
786 public function augmentSearchResults( SearchResultSet
$resultSet ) {
789 Hooks
::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
790 if ( !$setAugmentors && !$rowAugmentors ) {
795 // Convert row augmentors to set augmentor
796 foreach ( $rowAugmentors as $name => $row ) {
797 if ( isset( $setAugmentors[$name] ) ) {
798 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
800 $setAugmentors[$name] = new PerRowAugmentor( $row );
803 foreach ( $setAugmentors as $name => $augmentor ) {
804 $data = $augmentor->augmentAll( $resultSet );
806 $resultSet->setAugmentedData( $name, $data );