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;
51 * @deprecated since 1.34
53 protected $searchTerms = [];
56 protected $showSuggestion = true;
57 private $sort = self
::DEFAULT_SORT
;
59 /** @var array Feature values */
60 protected $features = [];
62 /** Profile type for completionSearch */
63 const COMPLETION_PROFILE_TYPE
= 'completionSearchProfile';
65 /** Profile type for query independent ranking features */
66 const FT_QUERY_INDEP_PROFILE_TYPE
= 'fulltextQueryIndepProfile';
68 /** Integer flag for legalSearchChars: includes all chars allowed in a search query */
71 /** Integer flag for legalSearchChars: includes all chars allowed in a search term */
72 const CHARS_NO_SYNTAX
= 2;
75 * Perform a full text search query and return a result set.
76 * If full text searches are not supported or disabled, return null.
78 * As of 1.32 overriding this function is deprecated. It will
79 * be converted to final in 1.34. Override self::doSearchText().
81 * @param string $term Raw search term
82 * @return ISearchResultSet|Status|null
84 public function searchText( $term ) {
85 return $this->maybePaginate( function () use ( $term ) {
86 return $this->doSearchText( $term );
91 * Perform a full text search query and return a result set.
93 * @param string $term Raw search term
94 * @return ISearchResultSet|Status|null
97 protected function doSearchText( $term ) {
102 * Perform a title search in the article archive.
103 * NOTE: these results still should be filtered by
104 * matching against PageArchive, permissions checks etc
105 * The results returned by this methods are only sugegstions and
106 * may not end up being shown to the user.
108 * As of 1.32 overriding this function is deprecated. It will
109 * be converted to final in 1.34. Override self::doSearchArchiveTitle().
111 * @param string $term Raw search term
115 public function searchArchiveTitle( $term ) {
116 return $this->doSearchArchiveTitle( $term );
120 * Perform a title search in the article archive.
122 * @param string $term Raw search term
126 protected function doSearchArchiveTitle( $term ) {
127 return Status
::newGood( [] );
131 * Perform a title-only search query and return a result set.
132 * If title searches are not supported or disabled, return null.
135 * As of 1.32 overriding this function is deprecated. It will
136 * be converted to final in 1.34. Override self::doSearchTitle().
138 * @param string $term Raw search term
139 * @return ISearchResultSet|null
141 public function searchTitle( $term ) {
142 return $this->maybePaginate( function () use ( $term ) {
143 return $this->doSearchTitle( $term );
148 * Perform a title-only search query and return a result set.
150 * @param string $term Raw search term
151 * @return ISearchResultSet|null
154 protected function doSearchTitle( $term ) {
159 * Performs an overfetch and shrink operation to determine if
160 * the next page is available for search engines that do not
161 * explicitly implement their own pagination.
163 * @param Closure $fn Takes no arguments
164 * @return ISearchResultSet|Status<ISearchResultSet>|null Result of calling $fn
166 private function maybePaginate( Closure
$fn ) {
167 if ( $this instanceof PaginatingSearchEngine
) {
172 $resultSetOrStatus = $fn();
178 if ( $resultSetOrStatus instanceof ISearchResultSet
) {
179 $resultSet = $resultSetOrStatus;
180 } elseif ( $resultSetOrStatus instanceof Status
&&
181 $resultSetOrStatus->getValue() instanceof ISearchResultSet
183 $resultSet = $resultSetOrStatus->getValue();
186 $resultSet->shrink( $this->limit
);
189 return $resultSetOrStatus;
194 * @param string $feature
197 public function supports( $feature ) {
198 switch ( $feature ) {
199 case 'search-update':
201 case 'title-suffix-filter':
208 * Way to pass custom data for engines
210 * @param string $feature
213 public function setFeatureData( $feature, $data ) {
214 $this->features
[$feature] = $data;
218 * Way to retrieve custom data set by setFeatureData
219 * or by the engine itself.
221 * @param string $feature feature name
222 * @return mixed the feature value or null if unset
224 public function getFeatureData( $feature ) {
225 return $this->features
[$feature] ??
null;
229 * When overridden in derived class, performs database-specific conversions
230 * on text to be used for searching or updating search index.
231 * Default implementation does nothing (simply returns $string).
233 * @param string $string String to process
236 public function normalizeText( $string ) {
237 // Some languages such as Chinese require word segmentation
238 return MediaWikiServices
::getInstance()->getContentLanguage()->segmentByWord( $string );
242 * Get service class to finding near matches.
243 * @param Config $config Configuration to use for the matcher.
244 * @return SearchNearMatcher
246 public function getNearMatcher( Config
$config ) {
247 return new SearchNearMatcher( $config,
248 MediaWikiServices
::getInstance()->getContentLanguage() );
252 * Get near matcher for default SearchEngine.
253 * @return SearchNearMatcher
255 protected static function defaultNearMatcher() {
256 $services = MediaWikiServices
::getInstance();
257 $config = $services->getMainConfig();
258 return $services->newSearchEngine()->getNearMatcher( $config );
262 * Get chars legal for search
263 * @param int $type type of search chars (see self::CHARS_ALL
264 * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
267 public function legalSearchChars( $type = self
::CHARS_ALL
) {
268 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
272 * Set the maximum number of results to return
273 * and how many to skip before returning the first.
278 function setLimitOffset( $limit, $offset = 0 ) {
279 $this->limit
= intval( $limit );
280 $this->offset
= intval( $offset );
284 * Set which namespaces the search should include.
285 * Give an array of namespace index numbers.
287 * @param int[]|null $namespaces
289 function setNamespaces( $namespaces ) {
291 // Filter namespaces to only keep valid ones
292 $validNs = MediaWikiServices
::getInstance()->getSearchEngineConfig()->searchableNamespaces();
293 $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
294 return $ns < 0 ||
isset( $validNs[$ns] );
299 $this->namespaces
= $namespaces;
303 * Set whether the searcher should try to build a suggestion. Note: some searchers
304 * don't support building a suggestion in the first place and others don't respect
307 * @param bool $showSuggestion Should the searcher try to build suggestions
309 function setShowSuggestion( $showSuggestion ) {
310 $this->showSuggestion
= $showSuggestion;
314 * Get the valid sort directions. All search engines support 'relevance' but others
315 * might support more. The default in all implementations must be 'relevance.'
318 * @return string[] the valid sort directions for setSort
320 public function getValidSorts() {
321 return [ self
::DEFAULT_SORT
];
325 * Set the sort direction of the search results. Must be one returned by
326 * SearchEngine::getValidSorts()
329 * @throws InvalidArgumentException
330 * @param string $sort sort direction for query result
332 public function setSort( $sort ) {
333 if ( !in_array( $sort, $this->getValidSorts() ) ) {
334 throw new InvalidArgumentException( "Invalid sort: $sort. " .
335 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
341 * Get the sort direction of the search results
346 public function getSort() {
351 * Parse some common prefixes: all (search everything)
352 * or namespace names and set the list of namespaces
353 * of this class accordingly.
355 * @deprecated since 1.32; should be handled internally by the search engine
356 * @param string $query
359 function replacePrefixes( $query ) {
364 * Parse some common prefixes: all (search everything)
367 * @param string $query
368 * @param bool $withAllKeyword activate support of the "all:" keyword and its
369 * translations to activate searching on all namespaces.
370 * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
371 * if classic namespace identification did not match.
372 * @return false|array false if no namespace was extracted, an array
373 * with the parsed query at index 0 and an array of namespaces at index
374 * 1 (or null for all namespaces).
376 * @throws MWException
378 public static function parseNamespacePrefixes(
380 $withAllKeyword = true,
381 $withPrefixSearchExtractNamespaceHook = false
384 if ( strpos( $query, ':' ) === false ) { // nothing to do
387 $extractedNamespace = null;
390 if ( $withAllKeyword ) {
393 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
394 // force all: so that we have a common syntax for all the wikis
395 if ( !in_array( 'all:', $allkeywords ) ) {
396 $allkeywords[] = 'all:';
399 foreach ( $allkeywords as $kw ) {
400 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
401 $extractedNamespace = null;
402 $parsed = substr( $query, strlen( $kw ) );
409 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
410 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
411 $index = MediaWikiServices
::getInstance()->getContentLanguage()->getNsIndex( $prefix );
412 if ( $index !== false ) {
413 $extractedNamespace = [ $index ];
414 $parsed = substr( $query, strlen( $prefix ) +
1 );
415 } elseif ( $withPrefixSearchExtractNamespaceHook ) {
416 $hookNamespaces = [ NS_MAIN
];
418 Hooks
::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
419 if ( $hookQuery !== $query ) {
420 $parsed = $hookQuery;
421 $extractedNamespace = $hookNamespaces;
430 return [ $parsed, $extractedNamespace ];
434 * Find snippet highlight settings for all users
435 * @return array Contextlines, contextchars
436 * @deprecated in 1.34 use the SearchHighlighter constants directly
437 * @see SearchHighlighter::DEFAULT_CONTEXT_CHARS
438 * @see SearchHighlighter::DEFAULT_CONTEXT_LINES
440 public static function userHighlightPrefs() {
441 $contextlines = SearchHighlighter
::DEFAULT_CONTEXT_LINES
;
442 $contextchars = SearchHighlighter
::DEFAULT_CONTEXT_CHARS
;
443 return [ $contextlines, $contextchars ];
447 * Create or update the search index record for the given page.
448 * Title and text should be pre-processed.
452 * @param string $title
453 * @param string $text
455 function update( $id, $title, $text ) {
460 * Update a search index record's title only.
461 * Title should be pre-processed.
465 * @param string $title
467 function updateTitle( $id, $title ) {
472 * Delete an indexed page
473 * Title should be pre-processed.
476 * @param int $id Page id that was deleted
477 * @param string $title Title of page that was deleted
479 function delete( $id, $title ) {
484 * Get the raw text for updating the index from a content object
485 * Nicer search backends could possibly do something cooler than
486 * just returning raw text
488 * @todo This isn't ideal, we'd really like to have content-specific handling here
489 * @param Title $t Title we're indexing
490 * @param Content|null $c Content of the page to index
492 * @deprecated since 1.34 use Content::getTextForSearchIndex directly
494 public function getTextFromContent( Title
$t, Content
$c = null ) {
495 return $c ?
$c->getTextForSearchIndex() : '';
499 * If an implementation of SearchEngine handles all of its own text processing
500 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
501 * rather silly handling, it should return true here instead.
504 * @deprecated since 1.34 no longer needed since getTextFromContent is being deprecated
506 public function textAlreadyUpdatedForIndex() {
511 * Makes search simple string if it was namespaced.
512 * Sets namespaces of the search to namespaces extracted from string.
513 * @param string $search
514 * @return string Simplified search string
516 protected function normalizeNamespaces( $search ) {
517 $queryAndNs = self
::parseNamespacePrefixes( $search, false, true );
518 if ( $queryAndNs !== false ) {
519 $this->setNamespaces( $queryAndNs[1] );
520 return $queryAndNs[0];
526 * Perform an overfetch of completion search results. This allows
527 * determining if another page of results is available.
529 * @param string $search
530 * @return SearchSuggestionSet
532 protected function completionSearchBackendOverfetch( $search ) {
535 return $this->completionSearchBackend( $search );
542 * Perform a completion search.
543 * Does not resolve namespaces and does not check variants.
544 * Search engine implementations may want to override this function.
545 * @param string $search
546 * @return SearchSuggestionSet
548 protected function completionSearchBackend( $search ) {
551 $search = trim( $search );
553 if ( !in_array( NS_SPECIAL
, $this->namespaces
) && // We do not run hook on Special: search
554 !Hooks
::run( 'PrefixSearchBackend',
555 [ $this->namespaces
, $search, $this->limit
, &$results, $this->offset
]
557 // False means hook worked.
558 // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
560 return SearchSuggestionSet
::fromStrings( $results );
562 // Hook did not do the job, use default simple search
563 $results = $this->simplePrefixSearch( $search );
564 return SearchSuggestionSet
::fromTitles( $results );
569 * Perform a completion search.
570 * @param string $search
571 * @return SearchSuggestionSet
573 public function completionSearch( $search ) {
574 if ( trim( $search ) === '' ) {
575 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
577 $search = $this->normalizeNamespaces( $search );
578 $suggestions = $this->completionSearchBackendOverfetch( $search );
579 return $this->processCompletionResults( $search, $suggestions );
583 * Perform a completion search with variants.
584 * @param string $search
585 * @return SearchSuggestionSet
587 public function completionSearchWithVariants( $search ) {
588 if ( trim( $search ) === '' ) {
589 return SearchSuggestionSet
::emptySuggestionSet(); // Return empty result
591 $search = $this->normalizeNamespaces( $search );
593 $results = $this->completionSearchBackendOverfetch( $search );
594 $fallbackLimit = 1 +
$this->limit
- $results->getSize();
595 if ( $fallbackLimit > 0 ) {
596 $fallbackSearches = MediaWikiServices
::getInstance()->getContentLanguage()->
597 autoConvertToAllVariants( $search );
598 $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
600 foreach ( $fallbackSearches as $fbs ) {
601 $this->setLimitOffset( $fallbackLimit );
602 $fallbackSearchResult = $this->completionSearch( $fbs );
603 $results->appendAll( $fallbackSearchResult );
604 $fallbackLimit -= $fallbackSearchResult->getSize();
605 if ( $fallbackLimit <= 0 ) {
610 return $this->processCompletionResults( $search, $results );
614 * Extract titles from completion results
615 * @param SearchSuggestionSet $completionResults
618 public function extractTitles( SearchSuggestionSet
$completionResults ) {
619 return $completionResults->map( function ( SearchSuggestion
$sugg ) {
620 return $sugg->getSuggestedTitle();
625 * Process completion search results.
626 * Resolves the titles and rescores.
627 * @param string $search
628 * @param SearchSuggestionSet $suggestions
629 * @return SearchSuggestionSet
631 protected function processCompletionResults( $search, SearchSuggestionSet
$suggestions ) {
632 // We over-fetched to determine pagination. Shrink back down if we have extra results
633 // and mark if pagination is possible
634 $suggestions->shrink( $this->limit
);
636 $search = trim( $search );
637 // preload the titles with LinkBatch
638 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion
$sugg ) {
639 return $sugg->getSuggestedTitle();
641 $lb->setCaller( __METHOD__
);
644 $diff = $suggestions->filter( function ( SearchSuggestion
$sugg ) {
645 return $sugg->getSuggestedTitle()->isKnown();
648 MediaWikiServices
::getInstance()->getStatsdDataFactory()
649 ->updateCount( 'search.completion.missing', $diff );
652 $results = $suggestions->map( function ( SearchSuggestion
$sugg ) {
653 return $sugg->getSuggestedTitle()->getPrefixedText();
656 if ( $this->offset
=== 0 ) {
657 // Rescore results with an exact title match
658 // NOTE: in some cases like cross-namespace redirects
659 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
660 // backends like Cirrus will return no results. We should still
661 // try an exact title match to workaround this limitation
662 $rescorer = new SearchExactMatchRescorer();
663 $rescoredResults = $rescorer->rescore( $search, $this->namespaces
, $results, $this->limit
);
665 // No need to rescore if offset is not 0
666 // The exact match must have been returned at position 0
668 $rescoredResults = $results;
671 if ( count( $rescoredResults ) > 0 ) {
672 $found = array_search( $rescoredResults[0], $results );
673 if ( $found === false ) {
674 // If the first result is not in the previous array it
675 // means that we found a new exact match
676 $exactMatch = SearchSuggestion
::fromTitle( 0, Title
::newFromText( $rescoredResults[0] ) );
677 $suggestions->prepend( $exactMatch );
678 $suggestions->shrink( $this->limit
);
680 // if the first result is not the same we need to rescore
682 $suggestions->rescore( $found );
691 * Simple prefix search for subpages.
692 * @param string $search
695 public function defaultPrefixSearch( $search ) {
696 if ( trim( $search ) === '' ) {
700 $search = $this->normalizeNamespaces( $search );
701 return $this->simplePrefixSearch( $search );
705 * Call out to simple search backend.
706 * Defaults to TitlePrefixSearch.
707 * @param string $search
710 protected function simplePrefixSearch( $search ) {
711 // Use default database prefix search
712 $backend = new TitlePrefixSearch
;
713 return $backend->defaultSearchBackend( $this->namespaces
, $search, $this->limit
, $this->offset
);
717 * Get a list of supported profiles.
718 * Some search engine implementations may expose specific profiles to fine-tune
720 * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
721 * The array returned by this function contains the following keys:
722 * - name: the profile name to use with setFeatureData
723 * - desc-message: the i18n description
724 * - default: set to true if this profile is the default
727 * @param string $profileType the type of profiles
728 * @param User|null $user the user requesting the list of profiles
729 * @return array|null the list of profiles or null if none available
731 public function getProfiles( $profileType, User
$user = null ) {
736 * Create a search field definition.
737 * Specific search engines should override this method to create search fields.
738 * @param string $name
739 * @param string $type One of the types in SearchIndexField::INDEX_TYPE_*
740 * @return SearchIndexField
743 public function makeSearchFieldMapping( $name, $type ) {
744 return new NullIndexField();
748 * Get fields for search index
750 * @return SearchIndexField[] Index field definitions for all content handlers
752 public function getSearchIndexFields() {
753 $models = ContentHandler
::getContentModels();
755 $seenHandlers = new SplObjectStorage();
756 foreach ( $models as $model ) {
758 $handler = ContentHandler
::getForModelID( $model );
760 catch ( MWUnknownContentModelException
$e ) {
761 // If we can find no handler, ignore it
764 // Several models can have the same handler, so avoid processing it repeatedly
765 if ( $seenHandlers->contains( $handler ) ) {
766 // We already did this one
769 $seenHandlers->attach( $handler );
770 $handlerFields = $handler->getFieldsForSearchIndex( $this );
771 foreach ( $handlerFields as $fieldName => $fieldData ) {
772 if ( empty( $fields[$fieldName] ) ) {
773 $fields[$fieldName] = $fieldData;
775 // TODO: do we allow some clashes with the same type or reject all of them?
776 $mergeDef = $fields[$fieldName]->merge( $fieldData );
778 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
780 $fields[$fieldName] = $mergeDef;
784 // Hook to allow extensions to produce search mapping fields
785 Hooks
::run( 'SearchIndexFields', [ &$fields, $this ] );
790 * Augment search results with extra data.
792 * @param ISearchResultSet $resultSet
794 public function augmentSearchResults( ISearchResultSet
$resultSet ) {
797 Hooks
::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
798 if ( !$setAugmentors && !$rowAugmentors ) {
803 // Convert row augmentors to set augmentor
804 foreach ( $rowAugmentors as $name => $row ) {
805 if ( isset( $setAugmentors[$name] ) ) {
806 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
808 $setAugmentors[$name] = new PerRowAugmentor( $row );
813 * @var ResultSetAugmentor $augmentor
815 foreach ( $setAugmentors as $name => $augmentor ) {
816 $data = $augmentor->augmentAll( $resultSet );
818 $resultSet->setAugmentedData( $name, $data );