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 ) {
235 // Some languages such as Chinese require word segmentation
236 return MediaWikiServices
::getInstance()->getContentLanguage()->segmentByWord( $string );
240 * Transform search term in cases when parts of the query came as different
241 * GET params (when supported), e.g. for prefix queries:
242 * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
243 * @param string $term
245 * @deprecated since 1.32 this should now be handled internally by the
248 public function transformSearchTerm( $term ) {
253 * Get service class to finding near matches.
254 * @param Config $config Configuration to use for the matcher.
255 * @return SearchNearMatcher
257 public function getNearMatcher( Config
$config ) {
258 return new SearchNearMatcher( $config,
259 MediaWikiServices
::getInstance()->getContentLanguage() );
263 * Get near matcher for default SearchEngine.
264 * @return SearchNearMatcher
266 protected static function defaultNearMatcher() {
267 $services = MediaWikiServices
::getInstance();
268 $config = $services->getMainConfig();
269 return $services->newSearchEngine()->getNearMatcher( $config );
273 * If an exact title match can be found, or a very slightly close match,
274 * return the title. If no match, returns NULL.
275 * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
276 * @param string $searchterm
279 public static function getNearMatch( $searchterm ) {
280 return static::defaultNearMatcher()->getNearMatch( $searchterm );
284 * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
286 * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
287 * @param string $searchterm
288 * @return SearchResultSet
290 public static function getNearMatchResultSet( $searchterm ) {
291 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
295 * Get chars legal for search
296 * NOTE: usage as static is deprecated and preserved only as BC measure
297 * @param int $type type of search chars (see self::CHARS_ALL
298 * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
301 public static function legalSearchChars( $type = self
::CHARS_ALL
) {
302 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
306 * Set the maximum number of results to return
307 * and how many to skip before returning the first.
312 function setLimitOffset( $limit, $offset = 0 ) {
313 $this->limit
= intval( $limit );
314 $this->offset
= intval( $offset );
318 * Set which namespaces the search should include.
319 * Give an array of namespace index numbers.
321 * @param int[]|null $namespaces
323 function setNamespaces( $namespaces ) {
325 // Filter namespaces to only keep valid ones
326 $validNs = $this->searchableNamespaces();
327 $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
328 return $ns < 0 ||
isset( $validNs[$ns] );
333 $this->namespaces
= $namespaces;
337 * Set whether the searcher should try to build a suggestion. Note: some searchers
338 * don't support building a suggestion in the first place and others don't respect
341 * @param bool $showSuggestion Should the searcher try to build suggestions
343 function setShowSuggestion( $showSuggestion ) {
344 $this->showSuggestion
= $showSuggestion;
348 * Get the valid sort directions. All search engines support 'relevance' but others
349 * might support more. The default in all implementations should be 'relevance.'
352 * @return string[] the valid sort directions for setSort
354 public function getValidSorts() {
355 return [ 'relevance' ];
359 * Set the sort direction of the search results. Must be one returned by
360 * SearchEngine::getValidSorts()
363 * @throws InvalidArgumentException
364 * @param string $sort sort direction for query result
366 public function setSort( $sort ) {
367 if ( !in_array( $sort, $this->getValidSorts() ) ) {
368 throw new InvalidArgumentException( "Invalid sort: $sort. " .
369 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
375 * Get the sort direction of the search results
380 public function getSort() {
385 * Parse some common prefixes: all (search everything)
386 * or namespace names and set the list of namespaces
387 * of this class accordingly.
389 * @deprecated since 1.32; should be handled internally by the search engine
390 * @param string $query
393 function replacePrefixes( $query ) {
398 * Parse some common prefixes: all (search everything)
401 * @param string $query
402 * @param bool $withAllKeyword activate support of the "all:" keyword and its
403 * translations to activate searching on all namespaces.
404 * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
405 * if classic namespace identification did not match.
406 * @return false|array false if no namespace was extracted, an array
407 * with the parsed query at index 0 and an array of namespaces at index
408 * 1 (or null for all namespaces).
410 * @throws MWException
412 public static function parseNamespacePrefixes(
414 $withAllKeyword = true,
415 $withPrefixSearchExtractNamespaceHook = false
418 if ( strpos( $query, ':' ) === false ) { // nothing to do
421 $extractedNamespace = null;
424 if ( $withAllKeyword ) {
427 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
428 // force all: so that we have a common syntax for all the wikis
429 if ( !in_array( 'all:', $allkeywords ) ) {
430 $allkeywords[] = 'all:';
433 foreach ( $allkeywords as $kw ) {
434 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
435 $extractedNamespace = null;
436 $parsed = substr( $query, strlen( $kw ) );
443 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
444 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
445 $index = MediaWikiServices
::getInstance()->getContentLanguage()->getNsIndex( $prefix );
446 if ( $index !== false ) {
447 $extractedNamespace = [ $index ];
448 $parsed = substr( $query, strlen( $prefix ) +
1 );
449 } elseif ( $withPrefixSearchExtractNamespaceHook ) {
450 $hookNamespaces = [ NS_MAIN
];
452 Hooks
::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
453 if ( $hookQuery !== $query ) {
454 $parsed = $hookQuery;
455 $extractedNamespace = $hookNamespaces;
464 return [ $parsed, $extractedNamespace ];
468 * Find snippet highlight settings for all users
469 * @return array Contextlines, contextchars
471 public static function userHighlightPrefs() {
472 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
473 $contextchars = 75; // same as above.... :P
474 return [ $contextlines, $contextchars ];
478 * Create or update the search index record for the given page.
479 * Title and text should be pre-processed.
483 * @param string $title
484 * @param string $text
486 function update( $id, $title, $text ) {
491 * Update a search index record's title only.
492 * Title should be pre-processed.
496 * @param string $title
498 function updateTitle( $id, $title ) {
503 * Delete an indexed page
504 * Title should be pre-processed.
507 * @param int $id Page id that was deleted
508 * @param string $title Title of page that was deleted
510 function delete( $id, $title ) {
515 * Get the raw text for updating the index from a content object
516 * Nicer search backends could possibly do something cooler than
517 * just returning raw text
519 * @todo This isn't ideal, we'd really like to have content-specific handling here
520 * @param Title $t Title we're indexing
521 * @param Content|null $c Content of the page to index
524 public function getTextFromContent( Title
$t, Content
$c = null ) {
525 return $c ?
$c->getTextForSearchIndex() : '';
529 * If an implementation of SearchEngine handles all of its own text processing
530 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
531 * rather silly handling, it should return true here instead.
535 public function textAlreadyUpdatedForIndex() {
540 * Makes search simple string if it was namespaced.
541 * Sets namespaces of the search to namespaces extracted from string.
542 * @param string $search
543 * @return string Simplified search string
545 protected function normalizeNamespaces( $search ) {
546 $queryAndNs = self
::parseNamespacePrefixes( $search, false, true );
547 if ( $queryAndNs !== false ) {
548 $this->setNamespaces( $queryAndNs[1] );
549 return $queryAndNs[0];
555 * Perform an overfetch of completion search results. This allows
556 * determining if another page of results is available.
558 * @param string $search
559 * @return SearchSuggestionSet
561 protected function completionSearchBackendOverfetch( $search ) {
564 return $this->completionSearchBackend( $search );
571 * Perform a completion search.
572 * Does not resolve namespaces and does not check variants.
573 * Search engine implementations may want to override this function.
574 * @param string $search
575 * @return SearchSuggestionSet
577 protected function completionSearchBackend( $search ) {
580 $search = trim( $search );
582 if ( !in_array( NS_SPECIAL
, $this->namespaces
) && // We do not run hook on Special: search
583 !Hooks
::run( 'PrefixSearchBackend',
584 [ $this->namespaces
, $search, $this->limit
, &$results, $this->offset
]
586 // False means hook worked.
587 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
589 return SearchSuggestionSet
::fromStrings( $results );
591 // Hook did not do the job, use default simple search
592 $results = $this->simplePrefixSearch( $search );
593 return SearchSuggestionSet
::fromTitles( $results );
598 * Perform a completion search.
599 * @param string $search
600 * @return SearchSuggestionSet
602 public function completionSearch( $search ) {
603 if ( trim( $search ) === '' ) {
604 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
606 $search = $this->normalizeNamespaces( $search );
607 $suggestions = $this->completionSearchBackendOverfetch( $search );
608 return $this->processCompletionResults( $search, $suggestions );
612 * Perform a completion search with variants.
613 * @param string $search
614 * @return SearchSuggestionSet
616 public function completionSearchWithVariants( $search ) {
617 if ( trim( $search ) === '' ) {
618 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
620 $search = $this->normalizeNamespaces( $search );
622 $results = $this->completionSearchBackendOverfetch( $search );
623 $fallbackLimit = 1 +
$this->limit
- $results->getSize();
624 if ( $fallbackLimit > 0 ) {
625 $fallbackSearches = MediaWikiServices
::getInstance()->getContentLanguage()->
626 autoConvertToAllVariants( $search );
627 $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
629 foreach ( $fallbackSearches as $fbs ) {
630 $this->setLimitOffset( $fallbackLimit );
631 $fallbackSearchResult = $this->completionSearch( $fbs );
632 $results->appendAll( $fallbackSearchResult );
633 $fallbackLimit -= $fallbackSearchResult->getSize();
634 if ( $fallbackLimit <= 0 ) {
639 return $this->processCompletionResults( $search, $results );
643 * Extract titles from completion results
644 * @param SearchSuggestionSet $completionResults
647 public function extractTitles( SearchSuggestionSet
$completionResults ) {
648 return $completionResults->map( function ( SearchSuggestion
$sugg ) {
649 return $sugg->getSuggestedTitle();
654 * Process completion search results.
655 * Resolves the titles and rescores.
656 * @param string $search
657 * @param SearchSuggestionSet $suggestions
658 * @return SearchSuggestionSet
660 protected function processCompletionResults( $search, SearchSuggestionSet
$suggestions ) {
661 // We over-fetched to determine pagination. Shrink back down if we have extra results
662 // and mark if pagination is possible
663 $suggestions->shrink( $this->limit
);
665 $search = trim( $search );
666 // preload the titles with LinkBatch
667 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion
$sugg ) {
668 return $sugg->getSuggestedTitle();
670 $lb->setCaller( __METHOD__
);
673 $diff = $suggestions->filter( function ( SearchSuggestion
$sugg ) {
674 return $sugg->getSuggestedTitle()->isKnown();
677 MediaWikiServices
::getInstance()->getStatsdDataFactory()
678 ->updateCount( 'search.completion.missing', $diff );
681 $results = $suggestions->map( function ( SearchSuggestion
$sugg ) {
682 return $sugg->getSuggestedTitle()->getPrefixedText();
685 if ( $this->offset
=== 0 ) {
686 // Rescore results with an exact title match
687 // NOTE: in some cases like cross-namespace redirects
688 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
689 // backends like Cirrus will return no results. We should still
690 // try an exact title match to workaround this limitation
691 $rescorer = new SearchExactMatchRescorer();
692 $rescoredResults = $rescorer->rescore( $search, $this->namespaces
, $results, $this->limit
);
694 // No need to rescore if offset is not 0
695 // The exact match must have been returned at position 0
697 $rescoredResults = $results;
700 if ( count( $rescoredResults ) > 0 ) {
701 $found = array_search( $rescoredResults[0], $results );
702 if ( $found === false ) {
703 // If the first result is not in the previous array it
704 // means that we found a new exact match
705 $exactMatch = SearchSuggestion
::fromTitle( 0, Title
::newFromText( $rescoredResults[0] ) );
706 $suggestions->prepend( $exactMatch );
707 $suggestions->shrink( $this->limit
);
709 // if the first result is not the same we need to rescore
711 $suggestions->rescore( $found );
720 * Simple prefix search for subpages.
721 * @param string $search
724 public function defaultPrefixSearch( $search ) {
725 if ( trim( $search ) === '' ) {
729 $search = $this->normalizeNamespaces( $search );
730 return $this->simplePrefixSearch( $search );
734 * Call out to simple search backend.
735 * Defaults to TitlePrefixSearch.
736 * @param string $search
739 protected function simplePrefixSearch( $search ) {
740 // Use default database prefix search
741 $backend = new TitlePrefixSearch
;
742 return $backend->defaultSearchBackend( $this->namespaces
, $search, $this->limit
, $this->offset
);
746 * Make a list of searchable namespaces and their canonical names.
747 * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
750 public static function searchableNamespaces() {
751 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->searchableNamespaces();
755 * Extract default namespaces to search from the given user's
756 * settings, returning a list of index numbers.
757 * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
761 public static function userNamespaces( $user ) {
762 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
766 * An array of namespaces indexes to be searched by default
767 * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
770 public static function defaultNamespaces() {
771 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->defaultNamespaces();
775 * Get a list of namespace names useful for showing in tooltips
777 * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
778 * @param array $namespaces
781 public static function namespacesAsText( $namespaces ) {
782 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
786 * Load up the appropriate search engine class for the currently
787 * active database backend, and return a configured instance.
788 * @deprecated since 1.27; Use SearchEngineFactory::create
789 * @param string $type Type of search backend, if not the default
790 * @return SearchEngine
792 public static function create( $type = null ) {
793 return MediaWikiServices
::getInstance()->getSearchEngineFactory()->create( $type );
797 * Return the search engines we support. If only $wgSearchType
798 * is set, it'll be an array of just that one item.
799 * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
802 public static function getSearchTypes() {
803 return MediaWikiServices
::getInstance()->getSearchEngineConfig()->getSearchTypes();
807 * Get a list of supported profiles.
808 * Some search engine implementations may expose specific profiles to fine-tune
810 * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
811 * The array returned by this function contains the following keys:
812 * - name: the profile name to use with setFeatureData
813 * - desc-message: the i18n description
814 * - default: set to true if this profile is the default
817 * @param string $profileType the type of profiles
818 * @param User|null $user the user requesting the list of profiles
819 * @return array|null the list of profiles or null if none available
821 public function getProfiles( $profileType, User
$user = null ) {
826 * Create a search field definition.
827 * Specific search engines should override this method to create search fields.
828 * @param string $name
829 * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
830 * @return SearchIndexField
833 public function makeSearchFieldMapping( $name, $type ) {
834 return new NullIndexField();
838 * Get fields for search index
840 * @return SearchIndexField[] Index field definitions for all content handlers
842 public function getSearchIndexFields() {
843 $models = ContentHandler
::getContentModels();
845 $seenHandlers = new SplObjectStorage();
846 foreach ( $models as $model ) {
848 $handler = ContentHandler
::getForModelID( $model );
850 catch ( MWUnknownContentModelException
$e ) {
851 // If we can find no handler, ignore it
854 // Several models can have the same handler, so avoid processing it repeatedly
855 if ( $seenHandlers->contains( $handler ) ) {
856 // We already did this one
859 $seenHandlers->attach( $handler );
860 $handlerFields = $handler->getFieldsForSearchIndex( $this );
861 foreach ( $handlerFields as $fieldName => $fieldData ) {
862 if ( empty( $fields[$fieldName] ) ) {
863 $fields[$fieldName] = $fieldData;
865 // TODO: do we allow some clashes with the same type or reject all of them?
866 $mergeDef = $fields[$fieldName]->merge( $fieldData );
868 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
870 $fields[$fieldName] = $mergeDef;
874 // Hook to allow extensions to produce search mapping fields
875 Hooks
::run( 'SearchIndexFields', [ &$fields, $this ] );
880 * Augment search results with extra data.
882 * @param SearchResultSet $resultSet
884 public function augmentSearchResults( SearchResultSet
$resultSet ) {
887 Hooks
::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
888 if ( !$setAugmentors && !$rowAugmentors ) {
893 // Convert row augmentors to set augmentor
894 foreach ( $rowAugmentors as $name => $row ) {
895 if ( isset( $setAugmentors[$name] ) ) {
896 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
898 $setAugmentors[$name] = new PerRowAugmentor( $row );
901 foreach ( $setAugmentors as $name => $augmentor ) {
902 $data = $augmentor->augmentAll( $resultSet );
904 $resultSet->setAugmentedData( $name, $data );
911 * Dummy class to be used when non-supported Database engine is present.
912 * @todo FIXME: Dummy class should probably try something at least mildly useful,
913 * such as a LIKE search through titles.
916 class SearchEngineDummy
extends SearchEngine
{