/** * merge top level multi-queries and resolve returned pageIds into Title objects. * * WARNING: experimental API * * @param string $query the user query * @param \Elastica\Response $response Response from elasticsearch _suggest api * @param array $profiles the suggestion profiles * @param int $limit Maximum suggestions to return, -1 for unlimited * @return SearchSuggestionSet a set of Suggestions */ protected function postProcessSuggest(\Elastica\Response $response, $profiles, $limit = -1) { $this->logContext['elasticTookMs'] = intval($response->getQueryTime() * 1000); $data = $response->getData(); unset($data['_shards']); $suggestions = array(); foreach ($data as $name => $results) { $discount = $profiles[$name]['discount']; foreach ($results as $suggested) { foreach ($suggested['options'] as $suggest) { $output = SuggestBuilder::decodeOutput($suggest['text']); if ($output === null) { // Ignore broken output continue; } $pageId = $output['id']; $type = $output['type']; $score = $discount * $suggest['score']; if (!isset($suggestions[$pageId]) || $score > $suggestions[$pageId]->getScore()) { $suggestion = new SearchSuggestion($score, null, null, $pageId); // If it's a title suggestion we have the text if ($type === SuggestBuilder::TITLE_SUGGESTION) { $suggestion->setText($output['text']); } $suggestions[$pageId] = $suggestion; } } } } // simply sort by existing scores uasort($suggestions, function ($a, $b) { return $b->getScore() - $a->getScore(); }); $this->logContext['hitsTotal'] = count($suggestions); if ($limit > 0) { $suggestions = array_slice($suggestions, 0, $limit, true); } $this->logContext['hitsReturned'] = count($suggestions); $this->logContext['hitsOffset'] = 0; // we must fetch redirect data for redirect suggestions $missingText = array(); foreach ($suggestions as $id => $suggestion) { if ($suggestion->getText() === null) { $missingText[] = $id; } } if (!empty($missingText)) { // Experimental. // // Second pass query to fetch redirects. // It's not clear if it's the best option, this will slowdown the whole query // when we hit a redirect suggestion. // Other option would be to encode redirects as a payload resulting in a // very big index... // XXX: we support only the content index $type = $this->connection->getPageType($this->indexBaseName, Connection::CONTENT_INDEX_TYPE); // NOTE: we are already in a poolCounterWork // Multi get is not supported by elastica $redirResponse = null; try { $redirResponse = $type->request('_mget', 'GET', array('ids' => $missingText), array('_source_include' => 'redirect')); if ($redirResponse->isOk()) { $this->logContext['elasticTook2PassMs'] = intval($redirResponse->getQueryTime() * 1000); $docs = $redirResponse->getData(); foreach ($docs['docs'] as $doc) { if (empty($doc['_source']['redirect'])) { continue; } // We use the original query, we should maybe use the variant that generated this result? $text = Util::chooseBestRedirect($this->term, $doc['_source']['redirect']); if (!empty($suggestions[$doc['_id']])) { $suggestions[$doc['_id']]->setText($text); } } } else { LoggerFactory::getInstance('CirrusSearch')->warning('Unable to fetch redirects for suggestion {query} with results {ids} : {error}', array('query' => $this->term, 'ids' => serialize($missingText), 'error' => $redirResponse->getError())); } } catch (\Elastica\Exception\ExceptionInterface $e) { LoggerFactory::getInstance('CirrusSearch')->warning('Unable to fetch redirects for suggestion {query} with results {ids} : {error}', array('query' => $this->term, 'ids' => serialize($missingText), 'error' => $this->extractMessage($e))); } } return new SearchSuggestionSet(array_filter($suggestions, function ($suggestion) { // text should be not empty for suggestions return $suggestion->getText() != null; })); }
/** * This implementation will run the completion suggester if it's enabled and if the * query is for NS_MAIN. Fallback to SearchEngine default implemention otherwise. * * @param string $search the user query * @return SearchSuggestionSet the suggestions */ public function searchSuggestions($search) { $config = ConfigFactory::getDefaultInstance()->makeConfig('CirrusSearch'); $useCompletionSuggester = $config->getElement('CirrusSearchUseCompletionSuggester'); $context = RequestContext::getMain(); $request = $context->getRequest(); // Allow experimentation with query parameters if ($request && $request->getVal('cirrusUseCompletionSuggester') === 'yes') { $useCompletionSuggester = true; } if (!$useCompletionSuggester) { // Completion suggester is not enabled, fallback to // default implementation return $this->searchSuggestionsPrefixSearchFallback($search); } // We use Title to extract namespace from a Title string // We append a random letter behind just in case the search // string ends with ':'. $title = Title::newFromText($search . "A"); if ($title->getNamespace() != NS_MAIN || count($this->namespaces) != 1 || reset($this->namespaces) != NS_MAIN) { // Fallback to prefix search if we are not on content namespace return $this->searchSuggestionsPrefixSearchFallback($search); } $user = $context->getUser(); // offset is omitted, searchSuggestion does not support // scrolling results $suggester = new CompletionSuggester($this->connection, $this->limit, $config, $this->namespaces, $user, $this->indexBaseName); // Not really useful, mostly for testing purpose $variants = $request->getArray('cirrusCompletionSuggesterVariant'); if (empty($variants)) { global $wgContLang; $variants = $wgContLang->autoConvertToAllVariants($search); } else { if (count($variants) > 3) { // We should not allow too many variants $variants = array_slice($variants, 0, 3); } } $response = $suggester->suggest($search, $variants); if ($response->isOK()) { // Errors will be logged, let's try the exact db match $suggestions = $response->getValue(); } else { $suggestions = SearchSuggestionSet::emptySuggestionSet(); } // preload the titles with LinkBatch $titles = $suggestions->map(function ($sugg) { return $sugg->getSuggestedTitle(); }); $lb = new LinkBatch($titles); $lb->setCaller(__METHOD__); $lb->execute(); $results = $suggestions->map(function ($sugg) { return $sugg->getSuggestedTitle()->getPrefixedText(); }); // now we can trim $search = trim($search); // Rescore results with an exact title match $rescorer = new SearchExactMatchRescorer(); $rescoredResults = $rescorer->rescore($search, $this->namespaces, $results, $this->limit); if (count($rescoredResults) > 0) { $found = array_search($rescoredResults[0], $results); if ($found === false) { // If the first result is not in the previous array it // means that we found a new exact match $exactMatch = SearchSuggestion::fromTitle(0, Title::newFromText($rescoredResults[0])); $suggestions->prepend($exactMatch); $suggestions->shrink($this->limit); } else { // if the first result is not the same we need to rescore if ($found > 0) { $suggestions->rescore($found); } } } return $suggestions; }