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
{
38 /** @var int[]|null */
39 public $namespaces = [ NS_MAIN
];
42 protected $limit = 10;
45 protected $offset = 0;
47 /** @var array|string */
48 protected $searchTerms = [];
51 protected $showSuggestion = true;
52 private $sort = 'relevance';
54 /** @var array Feature values */
55 protected $features = [];
57 /** @const string profile type for completionSearch */
58 const COMPLETION_PROFILE_TYPE
= 'completionSearchProfile';
60 /** @const string profile type for query independent ranking features */
61 const FT_QUERY_INDEP_PROFILE_TYPE
= 'fulltextQueryIndepProfile';
63 /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
66 /** @const int flag for legalSearchChars: includes all chars allowed in a search term */
67 const CHARS_NO_SYNTAX
= 2;
70 * Perform a full text search query and return a result set.
71 * If full text searches are not supported or disabled, return null.
73 * As of 1.32 overriding this function is deprecated. It will
74 * be converted to final in 1.34. Override self::doSearchText().
76 * @param string $term Raw search term
77 * @return SearchResultSet|Status|null
79 public function searchText( $term ) {
80 return $this->maybePaginate( function () use ( $term ) {
81 return $this->doSearchText( $term );
86 * Perform a full text search query and return a result set.
88 * @param string $term Raw search term
89 * @return SearchResultSet|Status|null
92 protected function doSearchText( $term ) {
97 * Perform a title search in the article archive.
98 * NOTE: these results still should be filtered by
99 * matching against PageArchive, permissions checks etc
100 * The results returned by this methods are only sugegstions and
101 * may not end up being shown to the user.
103 * As of 1.32 overriding this function is deprecated. It will
104 * be converted to final in 1.34. Override self::doSearchArchiveTitle().
106 * @param string $term Raw search term
107 * @return Status<Title[]>
110 public function searchArchiveTitle( $term ) {
111 return $this->doSearchArchiveTitle( $term );
115 * Perform a title search in the article archive.
117 * @param string $term Raw search term
118 * @return Status<Title[]>
121 protected function doSearchArchiveTitle( $term ) {
122 return Status
::newGood( [] );
126 * Perform a title-only search query and return a result set.
127 * If title searches are not supported or disabled, return null.
130 * As of 1.32 overriding this function is deprecated. It will
131 * be converted to final in 1.34. Override self::doSearchTitle().
133 * @param string $term Raw search term
134 * @return SearchResultSet|null
136 public function searchTitle( $term ) {
137 return $this->maybePaginate( function () use ( $term ) {
138 return $this->doSearchTitle( $term );
143 * Perform a title-only search query and return a result set.
145 * @param string $term Raw search term
146 * @return SearchResultSet|null
149 protected function doSearchTitle( $term ) {
154 * Performs an overfetch and shrink operation to determine if
155 * the next page is available for search engines that do not
156 * explicitly implement their own pagination.
158 * @param Closure $fn Takes no arguments
159 * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
161 private function maybePaginate( Closure
$fn ) {
162 if ( $this instanceof PaginatingSearchEngine
) {
167 $resultSetOrStatus = $fn();
173 if ( $resultSetOrStatus instanceof SearchResultSet
) {
174 $resultSet = $resultSetOrStatus;
175 } elseif ( $resultSetOrStatus instanceof Status
&&
176 $resultSetOrStatus->getValue() instanceof SearchResultSet
178 $resultSet = $resultSetOrStatus->getValue();
181 $resultSet->shrink( $this->limit
);
184 return $resultSetOrStatus;
189 * @param string $feature
192 public function supports( $feature ) {
193 switch ( $feature ) {
194 case 'search-update':
196 case 'title-suffix-filter':
203 * Way to pass custom data for engines
205 * @param string $feature
208 public function setFeatureData( $feature, $data ) {
209 $this->features
[$feature] = $data;
213 * Way to retrieve custom data set by setFeatureData
214 * or by the engine itself.
216 * @param string $feature feature name
217 * @return mixed the feature value or null if unset
219 public function getFeatureData( $feature ) {
220 if ( isset( $this->features
[$feature] ) ) {
221 return $this->features
[$feature];
227 * When overridden in derived class, performs database-specific conversions
228 * on text to be used for searching or updating search index.
229 * Default implementation does nothing (simply returns $string).
231 * @param string $string String to process
234 public function normalizeText( $string ) {
237 // Some languages such as Chinese require word segmentation
238 return $wgContLang->segmentByWord( $string );
242 * Transform search term in cases when parts of the query came as different
243 * GET params (when supported), e.g. for prefix queries:
244 * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
245 * @param string $term
247 * @deprecated since 1.32 this should now be handled internally by the
250 public function transformSearchTerm( $term ) {
255 * Get service class to finding near matches.
256 * @param Config $config Configuration to use for the matcher.
257 * @return SearchNearMatcher
259 public function getNearMatcher( Config
$config ) {
261 return new SearchNearMatcher( $config, $wgContLang );
265 * Get near matcher for default SearchEngine.
266 * @return SearchNearMatcher
268 protected static function defaultNearMatcher() {
269 $config = MediaWikiServices
::getInstance()->getMainConfig();
270 return MediaWikiServices
::getInstance()->newSearchEngine()->getNearMatcher( $config );
274 * If an exact title match can be found, or a very slightly close match,
275 * return the title. If no match, returns NULL.
276 * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
277 * @param string $searchterm
280 public static function getNearMatch( $searchterm ) {
281 return static::defaultNearMatcher()->getNearMatch( $searchterm );
285 * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
287 * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
288 * @param string $searchterm
289 * @return SearchResultSet
291 public static function getNearMatchResultSet( $searchterm ) {
292 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
296 * Get chars legal for search
297 * NOTE: usage as static is deprecated and preserved only as BC measure
298 * @param int $type type of search chars (see self::CHARS_ALL
299 * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
302 public static function legalSearchChars( $type = self
::CHARS_ALL
) {
303 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
307 * Set the maximum number of results to return
308 * and how many to skip before returning the first.
313 function setLimitOffset( $limit, $offset = 0 ) {
314 $this->limit
= intval( $limit );
315 $this->offset
= intval( $offset );
319 * Set which namespaces the search should include.
320 * Give an array of namespace index numbers.
322 * @param int[]|null $namespaces
324 function setNamespaces( $namespaces ) {
326 // Filter namespaces to only keep valid ones
327 $validNs = $this->searchableNamespaces();
328 $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
329 return $ns < 0 ||
isset( $validNs[$ns] );
334 $this->namespaces
= $namespaces;
338 * Set whether the searcher should try to build a suggestion. Note: some searchers
339 * don't support building a suggestion in the first place and others don't respect
342 * @param bool $showSuggestion Should the searcher try to build suggestions
344 function setShowSuggestion( $showSuggestion ) {
345 $this->showSuggestion
= $showSuggestion;
349 * Get the valid sort directions. All search engines support 'relevance' but others
350 * might support more. The default in all implementations should be 'relevance.'
353 * @return string[] the valid sort directions for setSort
355 public function getValidSorts() {
356 return [ 'relevance' ];
360 * Set the sort direction of the search results. Must be one returned by
361 * SearchEngine::getValidSorts()
364 * @throws InvalidArgumentException
365 * @param string $sort sort direction for query result
367 public function setSort( $sort ) {
368 if ( !in_array( $sort, $this->getValidSorts() ) ) {
369 throw new InvalidArgumentException( "Invalid sort: $sort. " .
370 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
376 * Get the sort direction of the search results
381 public function getSort() {
386 * Parse some common prefixes: all (search everything)
387 * or namespace names and set the list of namespaces
388 * of this class accordingly.
390 * @deprecated since 1.32; should be handled internally by the search engine
391 * @param string $query
394 function replacePrefixes( $query ) {
399 * Parse some common prefixes: all (search everything)
402 * @param string $query
403 * @param bool $withAllKeyword activate support of the "all:" keyword and its
404 * translations to activate searching on all namespaces.
405 * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
406 * if classic namespace identification did not match.
407 * @return false|array false if no namespace was extracted, an array
408 * with the parsed query at index 0 and an array of namespaces at index
409 * 1 (or null for all namespaces).
411 * @throws MWException
413 public static function parseNamespacePrefixes(
415 $withAllKeyword = true,
416 $withPrefixSearchExtractNamespaceHook = false
421 if ( strpos( $query, ':' ) === false ) { // nothing to do
424 $extractedNamespace = null;
427 if ( $withAllKeyword ) {
430 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
431 // force all: so that we have a common syntax for all the wikis
432 if ( !in_array( 'all:', $allkeywords ) ) {
433 $allkeywords[] = 'all:';
436 foreach ( $allkeywords as $kw ) {
437 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
438 $extractedNamespace = null;
439 $parsed = substr( $query, strlen( $kw ) );
446 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
447 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
448 $index = $wgContLang->getNsIndex( $prefix );
449 if ( $index !== false ) {
450 $extractedNamespace = [ $index ];
451 $parsed = substr( $query, strlen( $prefix ) +
1 );
452 } elseif ( $withPrefixSearchExtractNamespaceHook ) {
453 $hookNamespaces = [ NS_MAIN
];
455 Hooks
::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
456 if ( $hookQuery !== $query ) {
457 $parsed = $hookQuery;
458 $extractedNamespace = $hookNamespaces;
467 return [ $parsed, $extractedNamespace ];
471 * Find snippet highlight settings for all users
472 * @return array Contextlines, contextchars
474 public static function userHighlightPrefs() {
475 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
476 $contextchars = 75; // same as above.... :P
477 return [ $contextlines, $contextchars ];
481 * Create or update the search index record for the given page.
482 * Title and text should be pre-processed.
486 * @param string $title
487 * @param string $text
489 function update( $id, $title, $text ) {
494 * Update a search index record's title only.
495 * Title should be pre-processed.
499 * @param string $title
501 function updateTitle( $id, $title ) {
506 * Delete an indexed page
507 * Title should be pre-processed.
510 * @param int $id Page id that was deleted
511 * @param string $title Title of page that was deleted
513 function delete( $id, $title ) {
518 * Get the raw text for updating the index from a content object
519 * Nicer search backends could possibly do something cooler than
520 * just returning raw text
522 * @todo This isn't ideal, we'd really like to have content-specific handling here
523 * @param Title $t Title we're indexing
524 * @param Content|null $c Content of the page to index
527 public function getTextFromContent( Title
$t, Content
$c = null ) {
528 return $c ?
$c->getTextForSearchIndex() : '';
532 * If an implementation of SearchEngine handles all of its own text processing
533 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
534 * rather silly handling, it should return true here instead.
538 public function textAlreadyUpdatedForIndex() {
543 * Makes search simple string if it was namespaced.
544 * Sets namespaces of the search to namespaces extracted from string.
545 * @param string $search
546 * @return string Simplified search string
548 protected function normalizeNamespaces( $search ) {
549 $queryAndNs = self
::parseNamespacePrefixes( $search, false, true );
550 if ( $queryAndNs !== false ) {
551 $this->setNamespaces( $queryAndNs[1] );
552 return $queryAndNs[0];
558 * Perform an overfetch of completion search results. This allows
559 * determining if another page of results is available.
561 * @param string $search
562 * @return SearchSuggestionSet
564 protected function completionSearchBackendOverfetch( $search ) {
567 return $this->completionSearchBackend( $search );
574 * Perform a completion search.
575 * Does not resolve namespaces and does not check variants.
576 * Search engine implementations may want to override this function.
577 * @param string $search
578 * @return SearchSuggestionSet
580 protected function completionSearchBackend( $search ) {
583 $search = trim( $search );
585 if ( !in_array( NS_SPECIAL
, $this->namespaces
) && // We do not run hook on Special: search
586 !Hooks
::run( 'PrefixSearchBackend',
587 [ $this->namespaces
, $search, $this->limit
, &$results, $this->offset
]
589 // False means hook worked.
590 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
592 return SearchSuggestionSet
::fromStrings( $results );
594 // Hook did not do the job, use default simple search
595 $results = $this->simplePrefixSearch( $search );
596 return SearchSuggestionSet
::fromTitles( $results );
601 * Perform a completion search.
602 * @param string $search
603 * @return SearchSuggestionSet
605 public function completionSearch( $search ) {
606 if ( trim( $search ) === '' ) {
607 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
609 $search = $this->normalizeNamespaces( $search );
610 $suggestions = $this->completionSearchBackendOverfetch( $search );
611 return $this->processCompletionResults( $search, $suggestions );
615 * Perform a completion search with variants.
616 * @param string $search
617 * @return SearchSuggestionSet
619 public function completionSearchWithVariants( $search ) {
620 if ( trim( $search ) === '' ) {
621 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
623 $search = $this->normalizeNamespaces( $search );
625 $results = $this->completionSearchBackendOverfetch( $search );
626 $fallbackLimit = 1 +
$this->limit
- $results->getSize();
627 if ( $fallbackLimit > 0 ) {
630 $fallbackSearches = $wgContLang->autoConvertToAllVariants( $search );
631 $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
633 foreach ( $fallbackSearches as $fbs ) {
634 $this->setLimitOffset( $fallbackLimit );
635 $fallbackSearchResult = $this->completionSearch( $fbs );
636 $results->appendAll( $fallbackSearchResult );
637 $fallbackLimit -= $fallbackSearchResult->getSize();
638 if ( $fallbackLimit <= 0 ) {
643 return $this->processCompletionResults( $search, $results );
647 * Extract titles from completion results
648 * @param SearchSuggestionSet $completionResults
651 public function extractTitles( SearchSuggestionSet
$completionResults ) {
652 return $completionResults->map( function ( SearchSuggestion
$sugg ) {
653 return $sugg->getSuggestedTitle();
658 * Process completion search results.
659 * Resolves the titles and rescores.
660 * @param string $search
661 * @param SearchSuggestionSet $suggestions
662 * @return SearchSuggestionSet
664 protected function processCompletionResults( $search, SearchSuggestionSet
$suggestions ) {
665 // We over-fetched to determine pagination. Shrink back down if we have extra results
666 // and mark if pagination is possible
667 $suggestions->shrink( $this->limit
);
669 $search = trim( $search );
670 // preload the titles with LinkBatch
671 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion
$sugg ) {
672 return $sugg->getSuggestedTitle();
674 $lb->setCaller( __METHOD__
);
677 $diff = $suggestions->filter( function ( SearchSuggestion
$sugg ) {
678 return $sugg->getSuggestedTitle()->isKnown();
681 MediaWikiServices
::getInstance()->getStatsdDataFactory()
682 ->updateCount( 'search.completion.missing', $diff );
685 $results = $suggestions->map( function ( SearchSuggestion
$sugg ) {
686 return $sugg->getSuggestedTitle()->getPrefixedText();
689 if ( $this->offset
=== 0 ) {
690 // Rescore results with an exact title match
691 // NOTE: in some cases like cross-namespace redirects
692 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
693 // backends like Cirrus will return no results. We should still
694 // try an exact title match to workaround this limitation
695 $rescorer = new SearchExactMatchRescorer();
696 $rescoredResults = $rescorer->rescore( $search, $this->namespaces
, $results, $this->limit
);
698 // No need to rescore if offset is not 0
699 // The exact match must have been returned at position 0
701 $rescoredResults = $results;
704 if ( count( $rescoredResults ) > 0 ) {
705 $found = array_search( $rescoredResults[0], $results );
706 if ( $found === false ) {
707 // If the first result is not in the previous array it
708 // means that we found a new exact match
709 $exactMatch = SearchSuggestion
::fromTitle( 0, Title
::newFromText( $rescoredResults[0] ) );
710 $suggestions->prepend( $exactMatch );
711 $suggestions->shrink( $this->limit
);
713 // if the first result is not the same we need to rescore
715 $suggestions->rescore( $found );
724 * Simple prefix search for subpages.
725 * @param string $search
728 public function defaultPrefixSearch( $search ) {
729 if ( trim( $search ) === '' ) {
733 $search = $this->normalizeNamespaces( $search );
734 return $this->simplePrefixSearch( $search );
738 * Call out to simple search backend.
739 * Defaults to TitlePrefixSearch.
740 * @param string $search
743 protected function simplePrefixSearch( $search ) {
744 // Use default database prefix search
745 $backend = new TitlePrefixSearch
;
746 return $backend->defaultSearchBackend( $this->namespaces
, $search, $this->limit
, $this->offset
);
750 * Make a list of searchable namespaces and their canonical names.
751 * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
754 public static function searchableNamespaces() {
755 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->searchableNamespaces();
759 * Extract default namespaces to search from the given user's
760 * settings, returning a list of index numbers.
761 * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
765 public static function userNamespaces( $user ) {
766 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
770 * An array of namespaces indexes to be searched by default
771 * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
774 public static function defaultNamespaces() {
775 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->defaultNamespaces();
779 * Get a list of namespace names useful for showing in tooltips
781 * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
782 * @param array $namespaces
785 public static function namespacesAsText( $namespaces ) {
786 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
790 * Load up the appropriate search engine class for the currently
791 * active database backend, and return a configured instance.
792 * @deprecated since 1.27; Use SearchEngineFactory::create
793 * @param string $type Type of search backend, if not the default
794 * @return SearchEngine
796 public static function create( $type = null ) {
797 return MediaWikiServices
::getInstance()->getSearchEngineFactory()->create( $type );
801 * Return the search engines we support. If only $wgSearchType
802 * is set, it'll be an array of just that one item.
803 * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
806 public static function getSearchTypes() {
807 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->getSearchTypes();
811 * Get a list of supported profiles.
812 * Some search engine implementations may expose specific profiles to fine-tune
814 * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
815 * The array returned by this function contains the following keys:
816 * - name: the profile name to use with setFeatureData
817 * - desc-message: the i18n description
818 * - default: set to true if this profile is the default
821 * @param string $profileType the type of profiles
822 * @param User|null $user the user requesting the list of profiles
823 * @return array|null the list of profiles or null if none available
825 public function getProfiles( $profileType, User
$user = null ) {
830 * Create a search field definition.
831 * Specific search engines should override this method to create search fields.
832 * @param string $name
833 * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
834 * @return SearchIndexField
837 public function makeSearchFieldMapping( $name, $type ) {
838 return new NullIndexField();
842 * Get fields for search index
844 * @return SearchIndexField[] Index field definitions for all content handlers
846 public function getSearchIndexFields() {
847 $models = ContentHandler
::getContentModels();
849 $seenHandlers = new SplObjectStorage();
850 foreach ( $models as $model ) {
852 $handler = ContentHandler
::getForModelID( $model );
854 catch ( MWUnknownContentModelException
$e ) {
855 // If we can find no handler, ignore it
858 // Several models can have the same handler, so avoid processing it repeatedly
859 if ( $seenHandlers->contains( $handler ) ) {
860 // We already did this one
863 $seenHandlers->attach( $handler );
864 $handlerFields = $handler->getFieldsForSearchIndex( $this );
865 foreach ( $handlerFields as $fieldName => $fieldData ) {
866 if ( empty( $fields[$fieldName] ) ) {
867 $fields[$fieldName] = $fieldData;
869 // TODO: do we allow some clashes with the same type or reject all of them?
870 $mergeDef = $fields[$fieldName]->merge( $fieldData );
872 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
874 $fields[$fieldName] = $mergeDef;
878 // Hook to allow extensions to produce search mapping fields
879 Hooks
::run( 'SearchIndexFields', [ &$fields, $this ] );
884 * Augment search results with extra data.
886 * @param SearchResultSet $resultSet
888 public function augmentSearchResults( SearchResultSet
$resultSet ) {
891 Hooks
::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
892 if ( !$setAugmentors && !$rowAugmentors ) {
897 // Convert row augmentors to set augmentor
898 foreach ( $rowAugmentors as $name => $row ) {
899 if ( isset( $setAugmentors[$name] ) ) {
900 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
902 $setAugmentors[$name] = new PerRowAugmentor( $row );
905 foreach ( $setAugmentors as $name => $augmentor ) {
906 $data = $augmentor->augmentAll( $resultSet );
908 $resultSet->setAugmentedData( $name, $data );
915 * Dummy class to be used when non-supported Database engine is present.
916 * @todo FIXME: Dummy class should probably try something at least mildly useful,
917 * such as a LIKE search through titles.
920 class SearchEngineDummy
extends SearchEngine
{