ApiQuerySearch.php

Go to the documentation of this file.

<?php
use Wikimedia\ParamValidator\ParamValidator;
use Wikimedia\ParamValidator\TypeDef\EnumDef;
 
class ApiQuerySearch extends ApiQueryGeneratorBase {
    use SearchApi;
 
    private $allowedParams;
 
    private $searchEngineConfig;
 
    private $searchEngineFactory;
 
    public function __construct(
        ApiQuery $query,
        $moduleName,
        SearchEngineConfig $searchEngineConfig,
        SearchEngineFactory $searchEngineFactory
    ) {
        parent::__construct( $query, $moduleName, 'sr' );
        // Services also needed in SearchApi trait
        $this->searchEngineConfig = $searchEngineConfig;
        $this->searchEngineFactory = $searchEngineFactory;
    }
 
    public function execute() {
        $this->run();
    }
 
    public function executeGenerator( $resultPageSet ) {
        $this->run( $resultPageSet );
    }
 
    private function run( $resultPageSet = null ) {
        $params = $this->extractRequestParams();
 
        // Extract parameters
        $query = $params['search'];
        $what = $params['what'];
        $interwiki = $params['interwiki'];
        $searchInfo = array_fill_keys( $params['info'], true );
        $prop = array_fill_keys( $params['prop'], true );
 
        // Create search engine instance and set options
        $search = $this->buildSearchEngine( $params );
        if ( isset( $params['sort'] ) ) {
            $search->setSort( $params['sort'] );
        }
        $search->setFeatureData( 'rewrite', (bool)$params['enablerewrites'] );
        $search->setFeatureData( 'interwiki', (bool)$interwiki );
        // Hint to some SearchEngines about what snippets we would like returned
        $search->setFeatureData( 'snippets', $this->decideSnippets( $prop ) );
 
        $nquery = $search->replacePrefixes( $query );
        if ( $nquery !== $query ) {
            $query = $nquery;
            wfDeprecatedMsg( 'SearchEngine::replacePrefixes() is overridden by ' .
                get_class( $search ) . ', this was deprecated in MediaWiki 1.32',
                '1.32' );
        }
        // Perform the actual search
        if ( $what == 'text' ) {
            $matches = $search->searchText( $query );
        } elseif ( $what == 'title' ) {
            $matches = $search->searchTitle( $query );
        } elseif ( $what == 'nearmatch' ) {
            // near matches must receive the user input as provided, otherwise
            // the near matches within namespaces are lost.
            $matches = $search->getNearMatcher( $this->getConfig() )
                ->getNearMatchResultSet( $params['search'] );
        } else {
            // We default to title searches; this is a terrible legacy
            // of the way we initially set up the MySQL fulltext-based
            // search engine with separate title and text fields.
            // In the future, the default should be for a combined index.
            $what = 'title';
            $matches = $search->searchTitle( $query );
 
            // Not all search engines support a separate title search,
            // for instance the Lucene-based engine we use on Wikipedia.
            // In this case, fall back to full-text search (which will
            // include titles in it!)
            if ( $matches === null ) {
                $what = 'text';
                $matches = $search->searchText( $query );
            }
        }
 
        if ( $matches instanceof Status ) {
            $status = $matches;
            $matches = $status->getValue();
        } else {
            $status = null;
        }
 
        if ( $status ) {
            if ( $status->isOK() ) {
                $this->getMain()->getErrorFormatter()->addMessagesFromStatus(
                    $this->getModuleName(),
                    $status
                );
            } else {
                $this->dieStatus( $status );
            }
        } elseif ( $matches === null ) {
            $this->dieWithError( [ 'apierror-searchdisabled', $what ], "search-{$what}-disabled" );
        }
 
        $apiResult = $this->getResult();
        // Add search meta data to result
        if ( isset( $searchInfo['totalhits'] ) ) {
            $totalhits = $matches->getTotalHits();
            if ( $totalhits !== null ) {
                $apiResult->addValue( [ 'query', 'searchinfo' ],
                    'totalhits', $totalhits );
            }
        }
        if ( isset( $searchInfo['suggestion'] ) && $matches->hasSuggestion() ) {
            $apiResult->addValue( [ 'query', 'searchinfo' ],
                'suggestion', $matches->getSuggestionQuery() );
            $apiResult->addValue( [ 'query', 'searchinfo' ],
                'suggestionsnippet', HtmlArmor::getHtml( $matches->getSuggestionSnippet() ) );
        }
        if ( isset( $searchInfo['rewrittenquery'] ) && $matches->hasRewrittenQuery() ) {
            $apiResult->addValue( [ 'query', 'searchinfo' ],
                'rewrittenquery', $matches->getQueryAfterRewrite() );
            $apiResult->addValue( [ 'query', 'searchinfo' ],
                'rewrittenquerysnippet', HtmlArmor::getHtml( $matches->getQueryAfterRewriteSnippet() ) );
        }
 
        $titles = [];
        $data = [];
        $count = 0;
 
        if ( $matches->hasMoreResults() ) {
            $this->setContinueEnumParameter( 'offset', $params['offset'] + $params['limit'] );
        }
 
        foreach ( $matches as $result ) {
            $count++;
            // Silently skip broken and missing titles
            if ( $result->isBrokenTitle() || $result->isMissingRevision() ) {
                continue;
            }
 
            $vals = $this->getSearchResultData( $result, $prop );
 
            if ( $resultPageSet === null ) {
                if ( $vals ) {
                    // Add item to results and see whether it fits
                    $fit = $apiResult->addValue( [ 'query', $this->getModuleName() ], null, $vals );
                    if ( !$fit ) {
                        $this->setContinueEnumParameter( 'offset', $params['offset'] + $count - 1 );
                        break;
                    }
                }
            } else {
                $titles[] = $result->getTitle();
                $data[] = $vals ?: [];
            }
        }
 
        // Here we assume interwiki results do not count with
        // regular search results. We may want to reconsider this
        // if we ever return a lot of interwiki results or want pagination
        // for them.
        // Interwiki results inside main result set
        $canAddInterwiki = (bool)$params['enablerewrites'] && ( $resultPageSet === null );
        if ( $canAddInterwiki ) {
            $this->addInterwikiResults( $matches, $apiResult, $prop, 'additional',
                ISearchResultSet::INLINE_RESULTS );
        }
 
        // Interwiki results outside main result set
        if ( $interwiki && $resultPageSet === null ) {
            $this->addInterwikiResults( $matches, $apiResult, $prop, 'interwiki',
                ISearchResultSet::SECONDARY_RESULTS );
        }
 
        if ( $resultPageSet === null ) {
            $apiResult->addIndexedTagName( [
                'query', $this->getModuleName()
            ], 'p' );
        } else {
            $resultPageSet->setRedirectMergePolicy( static function ( $current, $new ) {
                if ( !isset( $current['index'] ) || $new['index'] < $current['index'] ) {
                    $current['index'] = $new['index'];
                }
                return $current;
            } );
            $resultPageSet->populateFromTitles( $titles );
            $offset = $params['offset'] + 1;
            foreach ( $titles as $index => $title ) {
                $resultPageSet->setGeneratorData(
                    $title,
                    $data[ $index ] + [ 'index' => $index + $offset ]
                );
            }
        }
    }
 
    private function getSearchResultData( SearchResult $result, $prop ) {
        // Silently skip broken and missing titles
        if ( $result->isBrokenTitle() || $result->isMissingRevision() ) {
            return null;
        }
 
        $vals = [];
 
        $title = $result->getTitle();
        ApiQueryBase::addTitleInfo( $vals, $title );
        $vals['pageid'] = $title->getArticleID();
 
        if ( isset( $prop['size'] ) ) {
            $vals['size'] = $result->getByteSize();
        }
        if ( isset( $prop['wordcount'] ) ) {
            $vals['wordcount'] = $result->getWordCount();
        }
        if ( isset( $prop['snippet'] ) ) {
            $vals['snippet'] = $result->getTextSnippet();
        }
        if ( isset( $prop['timestamp'] ) ) {
            $vals['timestamp'] = wfTimestamp( TS_ISO_8601, $result->getTimestamp() );
        }
        if ( isset( $prop['titlesnippet'] ) ) {
            $vals['titlesnippet'] = $result->getTitleSnippet();
        }
        if ( isset( $prop['categorysnippet'] ) ) {
            $vals['categorysnippet'] = $result->getCategorySnippet();
        }
        if ( $result->getRedirectTitle() !== null ) {
            if ( isset( $prop['redirecttitle'] ) ) {
                $vals['redirecttitle'] = $result->getRedirectTitle()->getPrefixedText();
            }
            if ( isset( $prop['redirectsnippet'] ) ) {
                $vals['redirectsnippet'] = $result->getRedirectSnippet();
            }
        }
        if ( $result->getSectionTitle() !== null ) {
            if ( isset( $prop['sectiontitle'] ) ) {
                $vals['sectiontitle'] = $result->getSectionTitle()->getFragment();
            }
            if ( isset( $prop['sectionsnippet'] ) ) {
                $vals['sectionsnippet'] = $result->getSectionSnippet();
            }
        }
        if ( isset( $prop['isfilematch'] ) ) {
            $vals['isfilematch'] = $result->isFileMatch();
        }
 
        if ( isset( $prop['extensiondata'] ) ) {
            $extra = $result->getExtensionData();
            // Add augmented data to the result. The data would be organized as a map:
            // augmentorName => data
            if ( $extra ) {
                $vals['extensiondata'] = ApiResult::addMetadataToResultVars( $extra );
            }
        }
 
        return $vals;
    }
 
    private function addInterwikiResults(
        ISearchResultSet $matches, ApiResult $apiResult, $prop,
        $section, $type
    ) {
        $totalhits = null;
        if ( $matches->hasInterwikiResults( $type ) ) {
            foreach ( $matches->getInterwikiResults( $type ) as $interwikiMatches ) {
                // Include number of results if requested
                $totalhits += $interwikiMatches->getTotalHits();
 
                foreach ( $interwikiMatches as $result ) {
                    $title = $result->getTitle();
                    $vals = $this->getSearchResultData( $result, $prop );
 
                    $vals['namespace'] = $result->getInterwikiNamespaceText();
                    $vals['title'] = $title->getText();
                    $vals['url'] = $title->getFullURL();
 
                    // Add item to results and see whether it fits
                    $fit = $apiResult->addValue( [
                            'query',
                            $section . $this->getModuleName(),
                            $result->getInterwikiPrefix()
                        ], null, $vals );
 
                    if ( !$fit ) {
                        // We hit the limit. We can't really provide any meaningful
                        // pagination info so just bail out
                        break;
                    }
                }
            }
            if ( $totalhits !== null ) {
                $apiResult->addValue( [ 'query', $section . 'searchinfo' ], 'totalhits', $totalhits );
                $apiResult->addIndexedTagName( [
                    'query', $section . $this->getModuleName()
                ], 'p' );
            }
        }
        return $totalhits;
    }
 
    private function decideSnippets( array $prop ): array {
        // Field names align with definitions in ContentHandler::getFieldsForSearchIndex.
        // Except `redirect` which isn't explicitly created, but refers to the title of
        // pages that redirect to the result page.
        $fields = [];
        if ( isset( $prop['titlesnippet'] ) ) {
            $fields[] = 'title';
        }
        // checking snippet and title variants is a bit special cased, but some search
        // engines generate the title variant from the snippet and thus must have the
        // snippet requested to provide the title.
        if ( isset( $prop['redirectsnippet'] ) || isset( $prop['redirecttitle'] ) ) {
            $fields[] = 'redirect';
        }
        if ( isset( $prop['categorysnippet'] ) ) {
            $fields[] = 'category';
        }
        if ( isset( $prop['sectionsnippet'] ) || isset( $prop['sectiontitle'] ) ) {
            $fields[] = 'heading';
        }
        return $fields;
    }
 
    public function getCacheMode( $params ) {
        return 'public';
    }
 
    public function getAllowedParams() {
        if ( $this->allowedParams !== null ) {
            return $this->allowedParams;
        }
 
        $this->allowedParams = $this->buildCommonApiParams() + [
            'what' => [
                ParamValidator::PARAM_TYPE => [
                    'title',
                    'text',
                    'nearmatch',
                ]
            ],
            'info' => [
                ParamValidator::PARAM_DEFAULT => 'totalhits|suggestion|rewrittenquery',
                ParamValidator::PARAM_TYPE => [
                    'totalhits',
                    'suggestion',
                    'rewrittenquery',
                ],
                ParamValidator::PARAM_ISMULTI => true,
            ],
            'prop' => [
                ParamValidator::PARAM_DEFAULT => 'size|wordcount|timestamp|snippet',
                ParamValidator::PARAM_TYPE => [
                    'size',
                    'wordcount',
                    'timestamp',
                    'snippet',
                    'titlesnippet',
                    'redirecttitle',
                    'redirectsnippet',
                    'sectiontitle',
                    'sectionsnippet',
                    'isfilematch',
                    'categorysnippet',
                    'score', // deprecated
                    'hasrelated', // deprecated
                    'extensiondata',
                ],
                ParamValidator::PARAM_ISMULTI => true,
                ApiBase::PARAM_HELP_MSG_PER_VALUE => [],
                EnumDef::PARAM_DEPRECATED_VALUES => [
                    'score' => true,
                    'hasrelated' => true
                ],
            ],
            'interwiki' => false,
            'enablerewrites' => false,
        ];
 
        // Generators only add info/properties if explicitly requested. T263841
        if ( $this->isInGeneratorMode() ) {
            $this->allowedParams['prop'][ParamValidator::PARAM_DEFAULT] = '';
            $this->allowedParams['info'][ParamValidator::PARAM_DEFAULT] = '';
        }
 
        // If we have more than one engine the list of available sorts is
        // difficult to represent. For now don't expose it.
        $alternatives = $this->searchEngineConfig->getSearchTypes();
        if ( count( $alternatives ) == 1 ) {
            $this->allowedParams['sort'] = [
                ParamValidator::PARAM_DEFAULT => SearchEngine::DEFAULT_SORT,
                ParamValidator::PARAM_TYPE => $this->searchEngineFactory->create()->getValidSorts(),
            ];
        }
 
        return $this->allowedParams;
    }
 
    public function getSearchProfileParams() {
        return [
            'qiprofile' => [
                'profile-type' => SearchEngine::FT_QUERY_INDEP_PROFILE_TYPE,
                'help-message' => 'apihelp-query+search-param-qiprofile',
            ],
        ];
    }
 
    protected function getExamplesMessages() {
        return [
            'action=query&list=search&srsearch=meaning'
                => 'apihelp-query+search-example-simple',
            'action=query&list=search&srwhat=text&srsearch=meaning'
                => 'apihelp-query+search-example-text',
            'action=query&generator=search&gsrsearch=meaning&prop=info'
                => 'apihelp-query+search-example-generator',
        ];
    }
 
    public function getHelpUrls() {
        return 'https://www.mediawiki.org/wiki/Special:MyLanguage/API:Search';
    }
}