SearchHandler.php

Go to the documentation of this file.

<?php
 
namespace MediaWiki\Rest\Handler;
 
use InvalidArgumentException;
use ISearchResultSet;
use MediaWiki\Cache\CacheKeyHelper;
use MediaWiki\Config\Config;
use MediaWiki\MainConfigNames;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Page\PageStore;
use MediaWiki\Page\RedirectLookup;
use MediaWiki\Permissions\PermissionManager;
use MediaWiki\Rest\Handler;
use MediaWiki\Rest\Handler\Helper\RestStatusTrait;
use MediaWiki\Rest\LocalizedHttpException;
use MediaWiki\Rest\Response;
use MediaWiki\Search\Entity\SearchResultThumbnail;
use MediaWiki\Search\SearchResultThumbnailProvider;
use MediaWiki\Title\TitleFormatter;
use SearchEngine;
use SearchEngineConfig;
use SearchEngineFactory;
use SearchResult;
use SearchSuggestion;
use StatusValue;
use Wikimedia\Message\MessageValue;
use Wikimedia\ParamValidator\ParamValidator;
use Wikimedia\ParamValidator\TypeDef\IntegerDef;
 
class SearchHandler extends Handler {
    use RestStatusTrait;
 
    private SearchEngineFactory $searchEngineFactory;
    private SearchEngineConfig $searchEngineConfig;
    private SearchResultThumbnailProvider $searchResultThumbnailProvider;
    private PermissionManager $permissionManager;
    private RedirectLookup $redirectLookup;
    private PageStore $pageStore;
    private TitleFormatter $titleFormatter;
 
    public const FULLTEXT_MODE = 'fulltext';
 
    public const COMPLETION_MODE = 'completion';
 
    private const SUPPORTED_MODES = [ self::FULLTEXT_MODE, self::COMPLETION_MODE ];
 
    private $mode = null;
 
    private const LIMIT = 50;
 
    private const MAX_LIMIT = 100;
 
    private const OFFSET = 0;
 
    private $completionCacheExpiry;
 
    public function __construct(
        Config $config,
        SearchEngineFactory $searchEngineFactory,
        SearchEngineConfig $searchEngineConfig,
        SearchResultThumbnailProvider $searchResultThumbnailProvider,
        PermissionManager $permissionManager,
        RedirectLookup $redirectLookup,
        PageStore $pageStore,
        TitleFormatter $titleFormatter
    ) {
        $this->searchEngineFactory = $searchEngineFactory;
        $this->searchEngineConfig = $searchEngineConfig;
        $this->searchResultThumbnailProvider = $searchResultThumbnailProvider;
        $this->permissionManager = $permissionManager;
        $this->redirectLookup = $redirectLookup;
        $this->pageStore = $pageStore;
        $this->titleFormatter = $titleFormatter;
 
        // @todo Avoid injecting the entire config, see T246377
        $this->completionCacheExpiry = $config->get( MainConfigNames::SearchSuggestCacheExpiry );
    }
 
    protected function postInitSetup() {
        $this->mode = $this->getConfig()['mode'] ?? self::FULLTEXT_MODE;
 
        if ( !in_array( $this->mode, self::SUPPORTED_MODES ) ) {
            throw new InvalidArgumentException(
                "Unsupported search mode `{$this->mode}` configured. Supported modes: " .
                implode( ', ', self::SUPPORTED_MODES )
            );
        }
    }
 
    private function createSearchEngine() {
        $limit = $this->getValidatedParams()['limit'];
 
        $searchEngine = $this->searchEngineFactory->create();
        $searchEngine->setNamespaces( $this->searchEngineConfig->defaultNamespaces() );
        $searchEngine->setLimitOffset( $limit, self::OFFSET );
        return $searchEngine;
    }
 
    public function needsWriteAccess() {
        return false;
    }
 
    private function getSearchResultsOrThrow( $results ) {
        if ( $results ) {
            if ( $results instanceof StatusValue ) {
                $status = $results;
                if ( !$status->isOK() ) {
                    if ( $status->getMessages( 'error' ) ) { // Only throw for errors, suppress warnings (for now)
                        $this->throwExceptionForStatus( $status, 'rest-search-error', 500 );
                    }
                }
                $statusValue = $status->getValue();
                if ( $statusValue instanceof ISearchResultSet ) {
                    return $statusValue->extractResults();
                }
            } else {
                return $results->extractResults();
            }
        }
        return [];
    }
 
    private function doSearch( $searchEngine ) {
        $query = $this->getValidatedParams()['q'];
 
        if ( $this->mode == self::COMPLETION_MODE ) {
            $completionSearch = $searchEngine->completionSearchWithVariants( $query );
            return $this->buildPageObjects( $completionSearch->getSuggestions() );
        } else {
            $titleSearch = $searchEngine->searchTitle( $query );
            $textSearch = $searchEngine->searchText( $query );
 
            $titleSearchResults = $this->getSearchResultsOrThrow( $titleSearch );
            $textSearchResults = $this->getSearchResultsOrThrow( $textSearch );
 
            $mergedResults = array_merge( $titleSearchResults, $textSearchResults );
            return $this->buildPageObjects( $mergedResults );
        }
    }
 
    private function buildPageObjects( array $searchResponse ): array {
        $pageInfos = [];
        foreach ( $searchResponse as $response ) {
            $isSearchResult = $response instanceof SearchResult;
            if ( $isSearchResult ) {
                if ( $response->isBrokenTitle() || $response->isMissingRevision() ) {
                    continue;
                }
                $title = $response->getTitle();
            } else {
                $title = $response->getSuggestedTitle();
            }
            $pageObj = $this->buildSinglePage( $title, $response );
            if ( $pageObj ) {
                $pageNsAndID = CacheKeyHelper::getKeyForPage( $pageObj['pageIdentity'] );
                // This handles the edge case where we have both the redirect source and redirect target page come back
                // in our search results. In such event, we prefer (and thus replace) with  the redirect target page.
                if ( isset( $pageInfos[$pageNsAndID] ) ) {
                    if ( $pageInfos[$pageNsAndID]['redirect'] !== null ) {
                        $pageInfos[$pageNsAndID]['result'] = $isSearchResult ? $response : null;
                        $pageInfos[$pageNsAndID]['suggestion'] = $isSearchResult ? null : $response;
                    }
                    continue;
                }
                $pageInfos[$pageNsAndID] = $pageObj;
            }
        }
        return $pageInfos;
    }
 
    private function buildSinglePage( $title, $result ) {
        $redirectTarget = $title->canExist() ? $this->redirectLookup->getRedirectTarget( $title ) : null;
        // Our page has a redirect that is not in a virtual namespace and is not an interwiki link.
        // See T301346, T303352
        if ( $redirectTarget && $redirectTarget->getNamespace() > -1 && !$redirectTarget->isExternal() ) {
            $redirectSource = $title;
            $anchor = $redirectTarget->getFragment();
            $title = $this->pageStore->getPageForLink( $redirectTarget );
        } else {
            $redirectSource = null;
            $anchor = null;
        }
        if ( !$title || !$this->getAuthority()->probablyCan( 'read', $title ) ) {
            return false;
        }
        return [
            'pageIdentity' => $title,
            'suggestion' => $result instanceof SearchSuggestion ? $result : null,
            'result' => $result instanceof SearchResult ? $result : null,
            'redirect' => $redirectSource,
            'anchor' => $anchor
        ];
    }
 
    private function buildResultFromPageInfos( array $pageInfos, array $thumbsAndDesc ): array {
        $pages = [];
        foreach ( $pageInfos as $pageInfo ) {
            [
                'pageIdentity' => $page,
                'suggestion' => $sugg,
                'result' => $result,
                'redirect' => $redirect,
                'anchor' => $anchor
            ] = $pageInfo;
            $excerpt = $sugg ? $sugg->getText() : $result->getTextSnippet();
            $id = ( $page instanceof PageIdentity && $page->canExist() ) ? $page->getId() : 0;
            $pages[] = [
                'id' => $id,
                'key' => $this->titleFormatter->getPrefixedDBkey( $page ),
                'title' => $this->titleFormatter->getPrefixedText( $page ),
                'excerpt' => $excerpt ?: null,
                'matched_title' => $redirect ? $this->titleFormatter->getPrefixedText( $redirect ) : null,
                'anchor' => $anchor ?: null,
                'description' => $id > 0 ? $thumbsAndDesc[$id]['description'] : null,
                'thumbnail' => $id > 0 ? $thumbsAndDesc[$id]['thumbnail'] : null,
            ];
        }
        return $pages;
    }
 
    private function serializeThumbnail( ?SearchResultThumbnail $thumbnail ): ?array {
        if ( $thumbnail == null ) {
            return null;
        }
 
        return [
            'mimetype' => $thumbnail->getMimeType(),
            'width' => $thumbnail->getWidth(),
            'height' => $thumbnail->getHeight(),
            'duration' => $thumbnail->getDuration(),
            'url' => $thumbnail->getUrl(),
        ];
    }
 
    private function buildDescriptionsFromPageIdentities( array $pageIdentities ) {
        $descriptions = array_fill_keys( array_keys( $pageIdentities ), null );
 
        $this->getHookRunner()->onSearchResultProvideDescription( $pageIdentities, $descriptions );
 
        return array_map( static function ( $description ) {
            return [ 'description' => $description ];
        }, $descriptions );
    }
 
    private function buildThumbnailsFromPageIdentities( array $pageIdentities ) {
        $thumbnails = $this->searchResultThumbnailProvider->getThumbnails( $pageIdentities );
        $thumbnails += array_fill_keys( array_keys( $pageIdentities ), null );
 
        return array_map( function ( $thumbnail ) {
            return [ 'thumbnail' => $this->serializeThumbnail( $thumbnail ) ];
        }, $thumbnails );
    }
 
    public function execute() {
        $searchEngine = $this->createSearchEngine();
        $pageInfos = $this->doSearch( $searchEngine );
 
        // We can only pass validated "real" PageIdentities to our hook handlers below
        $pageIdentities = array_reduce(
            array_values( $pageInfos ),
            static function ( $realPages, $item ) {
                $page = $item['pageIdentity'];
                if ( $page instanceof PageIdentity && $page->exists() ) {
                    $realPages[$item['pageIdentity']->getId()] = $item['pageIdentity'];
                }
                return $realPages;
            }, []
        );
 
        $descriptions = $this->buildDescriptionsFromPageIdentities( $pageIdentities );
        $thumbs = $this->buildThumbnailsFromPageIdentities( $pageIdentities );
 
        $thumbsAndDescriptions = [];
        foreach ( $descriptions as $pageId => $description ) {
            $thumbsAndDescriptions[$pageId] = $description + $thumbs[$pageId];
        }
 
        $result = $this->buildResultFromPageInfos( $pageInfos, $thumbsAndDescriptions );
 
        $response = $this->getResponseFactory()->createJson( [ 'pages' => $result ] );
 
        if ( $this->mode === self::COMPLETION_MODE && $this->completionCacheExpiry ) {
            // Type-ahead completion matches should be cached by the client and
            // in the CDN, especially for short prefixes.
            // See also $wgSearchSuggestCacheExpiry and ApiOpenSearch
            if ( $this->permissionManager->isEveryoneAllowed( 'read' ) ) {
                $response->setHeader( 'Cache-Control', 'public, max-age=' . $this->completionCacheExpiry );
            } else {
                $response->setHeader( 'Cache-Control', 'no-store, max-age=0' );
            }
        }
 
        return $response;
    }
 
    public function getParamSettings() {
        return [
            'q' => [
                self::PARAM_SOURCE => 'query',
                ParamValidator::PARAM_TYPE => 'string',
                ParamValidator::PARAM_REQUIRED => true,
                Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-search-q' ),
            ],
            'limit' => [
                self::PARAM_SOURCE => 'query',
                ParamValidator::PARAM_TYPE => 'integer',
                ParamValidator::PARAM_REQUIRED => false,
                ParamValidator::PARAM_DEFAULT => self::LIMIT,
                IntegerDef::PARAM_MIN => 1,
                IntegerDef::PARAM_MAX => self::MAX_LIMIT,
                Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-search-limit' ),
            ],
        ];
    }
 
    public function getResponseBodySchemaFileName( string $method ): ?string {
        return 'includes/Rest/Handler/Schema/SearchResults.json';
    }
}