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 /** @const string profile type for completionSearch */
60 const COMPLETION_PROFILE_TYPE
= 'completionSearchProfile';
62 /** @const string profile type for query independent ranking features */
63 const FT_QUERY_INDEP_PROFILE_TYPE
= 'fulltextQueryIndepProfile';
65 /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
68 /** @const int 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 * 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
244 * @deprecated since 1.32 this should now be handled internally by the
247 public function transformSearchTerm( $term ) {
252 * Get service class to finding near matches.
253 * @param Config $config Configuration to use for the matcher.
254 * @return SearchNearMatcher
256 public function getNearMatcher( Config
$config ) {
257 return new SearchNearMatcher( $config,
258 MediaWikiServices
::getInstance()->getContentLanguage() );
262 * Get near matcher for default SearchEngine.
263 * @return SearchNearMatcher
265 protected static function defaultNearMatcher() {
266 $services = MediaWikiServices
::getInstance();
267 $config = $services->getMainConfig();
268 return $services->newSearchEngine()->getNearMatcher( $config );
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
278 public static function getNearMatch( $searchterm ) {
279 return static::defaultNearMatcher()->getNearMatch( $searchterm );
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
289 public static function legalSearchChars( $type = self
::CHARS_ALL
) {
290 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
294 * Set the maximum number of results to return
295 * and how many to skip before returning the first.
300 function setLimitOffset( $limit, $offset = 0 ) {
301 $this->limit
= intval( $limit );
302 $this->offset
= intval( $offset );
306 * Set which namespaces the search should include.
307 * Give an array of namespace index numbers.
309 * @param int[]|null $namespaces
311 function setNamespaces( $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] );
321 $this->namespaces
= $namespaces;
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
329 * @param bool $showSuggestion Should the searcher try to build suggestions
331 function setShowSuggestion( $showSuggestion ) {
332 $this->showSuggestion
= $showSuggestion;
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.'
340 * @return string[] the valid sort directions for setSort
342 public function getValidSorts() {
343 return [ self
::DEFAULT_SORT
];
347 * Set the sort direction of the search results. Must be one returned by
348 * SearchEngine::getValidSorts()
351 * @throws InvalidArgumentException
352 * @param string $sort sort direction for query result
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() ) );
363 * Get the sort direction of the search results
368 public function getSort() {
373 * Parse some common prefixes: all (search everything)
374 * or namespace names and set the list of namespaces
375 * of this class accordingly.
377 * @deprecated since 1.32; should be handled internally by the search engine
378 * @param string $query
381 function replacePrefixes( $query ) {
386 * Parse some common prefixes: all (search everything)
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).
398 * @throws MWException
400 public static function parseNamespacePrefixes(
402 $withAllKeyword = true,
403 $withPrefixSearchExtractNamespaceHook = false
406 if ( strpos( $query, ':' ) === false ) { // nothing to do
409 $extractedNamespace = null;
412 if ( $withAllKeyword ) {
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:';
421 foreach ( $allkeywords as $kw ) {
422 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
423 $extractedNamespace = null;
424 $parsed = substr( $query, strlen( $kw ) );
431 if ( !$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 $hookNamespaces = [ NS_MAIN
];
440 Hooks
::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
441 if ( $hookQuery !== $query ) {
442 $parsed = $hookQuery;
443 $extractedNamespace = $hookNamespaces;
452 return [ $parsed, $extractedNamespace ];
456 * Find snippet highlight settings for all users
457 * @return array Contextlines, contextchars
459 public static function userHighlightPrefs() {
460 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
461 $contextchars = 75; // same as above.... :P
462 return [ $contextlines, $contextchars ];
466 * Create or update the search index record for the given page.
467 * Title and text should be pre-processed.
471 * @param string $title
472 * @param string $text
474 function update( $id, $title, $text ) {
479 * Update a search index record's title only.
480 * Title should be pre-processed.
484 * @param string $title
486 function updateTitle( $id, $title ) {
491 * Delete an indexed page
492 * Title should be pre-processed.
495 * @param int $id Page id that was deleted
496 * @param string $title Title of page that was deleted
498 function delete( $id, $title ) {
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
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
512 public function getTextFromContent( Title
$t, Content
$c = null ) {
513 return $c ?
$c->getTextForSearchIndex() : '';
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.
523 public function textAlreadyUpdatedForIndex() {
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
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];
543 * Perform an overfetch of completion search results. This allows
544 * determining if another page of results is available.
546 * @param string $search
547 * @return SearchSuggestionSet
549 protected function completionSearchBackendOverfetch( $search ) {
552 return $this->completionSearchBackend( $search );
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
565 protected function completionSearchBackend( $search ) {
568 $search = trim( $search );
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
]
574 // False means hook worked.
575 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
577 return SearchSuggestionSet
::fromStrings( $results );
579 // Hook did not do the job, use default simple search
580 $results = $this->simplePrefixSearch( $search );
581 return SearchSuggestionSet
::fromTitles( $results );
586 * Perform a completion search.
587 * @param string $search
588 * @return SearchSuggestionSet
590 public function completionSearch( $search ) {
591 if ( trim( $search ) === '' ) {
592 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
594 $search = $this->normalizeNamespaces( $search );
595 $suggestions = $this->completionSearchBackendOverfetch( $search );
596 return $this->processCompletionResults( $search, $suggestions );
600 * Perform a completion search with variants.
601 * @param string $search
602 * @return SearchSuggestionSet
604 public function completionSearchWithVariants( $search ) {
605 if ( trim( $search ) === '' ) {
606 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
608 $search = $this->normalizeNamespaces( $search );
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 ] );
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 ) {
627 return $this->processCompletionResults( $search, $results );
631 * Extract titles from completion results
632 * @param SearchSuggestionSet $completionResults
635 public function extractTitles( SearchSuggestionSet
$completionResults ) {
636 return $completionResults->map( function ( SearchSuggestion
$sugg ) {
637 return $sugg->getSuggestedTitle();
642 * Process completion search results.
643 * Resolves the titles and rescores.
644 * @param string $search
645 * @param SearchSuggestionSet $suggestions
646 * @return SearchSuggestionSet
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
);
653 $search = trim( $search );
654 // preload the titles with LinkBatch
655 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion
$sugg ) {
656 return $sugg->getSuggestedTitle();
658 $lb->setCaller( __METHOD__
);
661 $diff = $suggestions->filter( function ( SearchSuggestion
$sugg ) {
662 return $sugg->getSuggestedTitle()->isKnown();
665 MediaWikiServices
::getInstance()->getStatsdDataFactory()
666 ->updateCount( 'search.completion.missing', $diff );
669 $results = $suggestions->map( function ( SearchSuggestion
$sugg ) {
670 return $sugg->getSuggestedTitle()->getPrefixedText();
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
);
682 // No need to rescore if offset is not 0
683 // The exact match must have been returned at position 0
685 $rescoredResults = $results;
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
);
697 // if the first result is not the same we need to rescore
699 $suggestions->rescore( $found );
708 * Simple prefix search for subpages.
709 * @param string $search
712 public function defaultPrefixSearch( $search ) {
713 if ( trim( $search ) === '' ) {
717 $search = $this->normalizeNamespaces( $search );
718 return $this->simplePrefixSearch( $search );
722 * Call out to simple search backend.
723 * Defaults to TitlePrefixSearch.
724 * @param string $search
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
);
734 * Make a list of searchable namespaces and their canonical names.
735 * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
738 public static function searchableNamespaces() {
739 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->searchableNamespaces();
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()
749 public static function userNamespaces( $user ) {
750 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
754 * An array of namespaces indexes to be searched by default
755 * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
758 public static function defaultNamespaces() {
759 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->defaultNamespaces();
763 * Get a list of namespace names useful for showing in tooltips
765 * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
766 * @param array $namespaces
769 public static function namespacesAsText( $namespaces ) {
770 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
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
780 public static function create( $type = null ) {
781 return MediaWikiServices
::getInstance()->getSearchEngineFactory()->create( $type );
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()
790 public static function getSearchTypes() {
791 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->getSearchTypes();
795 * Get a list of supported profiles.
796 * Some search engine implementations may expose specific profiles to fine-tune
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
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
809 public function getProfiles( $profileType, User
$user = null ) {
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
821 public function makeSearchFieldMapping( $name, $type ) {
822 return new NullIndexField();
826 * Get fields for search index
828 * @return SearchIndexField[] Index field definitions for all content handlers
830 public function getSearchIndexFields() {
831 $models = ContentHandler
::getContentModels();
833 $seenHandlers = new SplObjectStorage();
834 foreach ( $models as $model ) {
836 $handler = ContentHandler
::getForModelID( $model );
838 catch ( MWUnknownContentModelException
$e ) {
839 // If we can find no handler, ignore it
842 // Several models can have the same handler, so avoid processing it repeatedly
843 if ( $seenHandlers->contains( $handler ) ) {
844 // We already did this one
847 $seenHandlers->attach( $handler );
848 $handlerFields = $handler->getFieldsForSearchIndex( $this );
849 foreach ( $handlerFields as $fieldName => $fieldData ) {
850 if ( empty( $fields[$fieldName] ) ) {
851 $fields[$fieldName] = $fieldData;
853 // TODO: do we allow some clashes with the same type or reject all of them?
854 $mergeDef = $fields[$fieldName]->merge( $fieldData );
856 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
858 $fields[$fieldName] = $mergeDef;
862 // Hook to allow extensions to produce search mapping fields
863 Hooks
::run( 'SearchIndexFields', [ &$fields, $this ] );
868 * Augment search results with extra data.
870 * @param SearchResultSet $resultSet
872 public function augmentSearchResults( SearchResultSet
$resultSet ) {
875 Hooks
::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
876 if ( !$setAugmentors && !$rowAugmentors ) {
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" );
886 $setAugmentors[$name] = new PerRowAugmentor( $row );
889 foreach ( $setAugmentors as $name => $augmentor ) {
890 $data = $augmentor->augmentAll( $resultSet );
892 $resultSet->setAugmentedData( $name, $data );
899 * Dummy class to be used when non-supported Database engine is present.
900 * @todo FIXME: Dummy class should probably try something at least mildly useful,
901 * such as a LIKE search through titles.
904 class SearchEngineDummy
extends SearchEngine
{