Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 372 |
|
0.00% |
0 / 33 |
CRAP | |
0.00% |
0 / 1 |
| ElasticSearchTtmServer | |
0.00% |
0 / 371 |
|
0.00% |
0 / 33 |
7310 | |
0.00% |
0 / 1 |
| isLocalSuggestion | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| expandLocation | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| query | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| doQuery | |
0.00% |
0 / 78 |
|
0.00% |
0 / 1 |
110 | |||
| update | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
42 | |||
| createDocument | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
2 | |||
| createIndex | |
0.00% |
0 / 30 |
|
0.00% |
0 / 1 |
6 | |||
| beginBootstrap | |
0.00% |
0 / 41 |
|
0.00% |
0 / 1 |
12 | |||
| beginBatch | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| batchInsertDefinitions | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| batchInsertTranslations | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
6 | |||
| endBatch | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| endBootstrap | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| getClient | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| useWikimediaExtraPlugin | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| getIndexName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getIndex | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getShardCount | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getReplicaCount | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| waitUntilReady | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| setLogger | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| logOutput | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| setDoReIndex | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| parseQueryString | |
0.00% |
0 / 39 |
|
0.00% |
0 / 1 |
110 | |||
| createSearch | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
30 | |||
| search | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getFacets | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
12 | |||
| getTotalHits | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getDocuments | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
30 | |||
| deleteByQuery | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| getElasticsearchVersion | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| checkElasticsearchVersion | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| assertResultSetInstance | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\TtmServer; |
| 5 | |
| 6 | use Elastica\Aggregation\Terms; |
| 7 | use Elastica\Client; |
| 8 | use Elastica\Document; |
| 9 | use Elastica\Exception\ExceptionInterface; |
| 10 | use Elastica\Index; |
| 11 | use Elastica\Mapping; |
| 12 | use Elastica\Query; |
| 13 | use Elastica\Query\BoolQuery; |
| 14 | use Elastica\Query\FunctionScore; |
| 15 | use Elastica\Query\MatchQuery; |
| 16 | use Elastica\Query\Term; |
| 17 | use Elastica\ResultSet; |
| 18 | use Elastica\Search; |
| 19 | use Exception; |
| 20 | use MediaWiki\Extension\Elastica\MWElasticUtils; |
| 21 | use MediaWiki\Extension\Translate\LogNames; |
| 22 | use MediaWiki\Extension\Translate\MessageLoading\MessageHandle; |
| 23 | use MediaWiki\Extension\Translate\TranslatorInterface\TranslationHelperException; |
| 24 | use MediaWiki\Logger\LoggerFactory; |
| 25 | use MediaWiki\MediaWikiServices; |
| 26 | use MediaWiki\Title\Title; |
| 27 | use MediaWiki\WikiMap\WikiMap; |
| 28 | use RuntimeException; |
| 29 | use TTMServerBootstrap; |
| 30 | |
| 31 | /** |
| 32 | * TtmServer backend based on ElasticSearch. Depends on Elastica. |
| 33 | * @author Niklas Laxström |
| 34 | * @license GPL-2.0-or-later |
| 35 | * @since 2014.04 |
| 36 | * @ingroup TTMServer |
| 37 | */ |
| 38 | class ElasticSearchTtmServer |
| 39 | extends TtmServer |
| 40 | implements ReadableTtmServer, WritableTtmServer, SearchableTtmServer |
| 41 | { |
| 42 | /** |
| 43 | * @const int in case a write operation fails during a batch process |
| 44 | * this constant controls the number of times we will retry the same |
| 45 | * operation. |
| 46 | */ |
| 47 | private const BULK_INDEX_RETRY_ATTEMPTS = 5; |
| 48 | |
| 49 | /** |
| 50 | * @const int time (seconds) to wait for the index to ready before |
| 51 | * starting to index. Since we wait for index status it can be relatively |
| 52 | * long especially if some nodes are restarted. |
| 53 | */ |
| 54 | private const WAIT_UNTIL_READY_TIMEOUT = 3600; |
| 55 | |
| 56 | private ?Client $client = null; |
| 57 | /** Reference to the maintenance script to relay logging output. */ |
| 58 | private ?TTMServerBootstrap $logger = null; |
| 59 | /** Used for reindex */ |
| 60 | private bool $updateMapping = false; |
| 61 | |
| 62 | public function isLocalSuggestion( array $suggestion ): bool { |
| 63 | return $suggestion['wiki'] === WikiMap::getCurrentWikiId(); |
| 64 | } |
| 65 | |
| 66 | public function expandLocation( array $suggestion ): string { |
| 67 | return $suggestion['uri']; |
| 68 | } |
| 69 | |
| 70 | public function query( string $sourceLanguage, string $targetLanguage, string $text ): array { |
| 71 | try { |
| 72 | return $this->doQuery( $sourceLanguage, $targetLanguage, $text ); |
| 73 | } catch ( Exception $e ) { |
| 74 | throw new TranslationHelperException( 'Elastica exception: ' . $e ); |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | private function doQuery( string $sourceLanguage, string $targetLanguage, string $text ): array { |
| 79 | if ( !$this->useWikimediaExtraPlugin() ) { |
| 80 | // ElasticTTM is currently not compatible with elasticsearch 2.x/5.x |
| 81 | // It needs FuzzyLikeThis ported via the wmf extra plugin |
| 82 | throw new RuntimeException( 'The wikimedia extra plugin is mandatory.' ); |
| 83 | } |
| 84 | /* Two query system: |
| 85 | * 1) Find all strings in source language that match text |
| 86 | * 2) Do another query for translations for those strings |
| 87 | */ |
| 88 | $connection = $this->getClient()->getConnection(); |
| 89 | $oldTimeout = $connection->getTimeout(); |
| 90 | $connection->setTimeout( 10 ); |
| 91 | |
| 92 | $fuzzyQuery = new FuzzyLikeThis(); |
| 93 | $fuzzyQuery->setLikeText( $text ); |
| 94 | $fuzzyQuery->addFieldNames( [ 'content' ] ); |
| 95 | |
| 96 | $boostQuery = new FunctionScore(); |
| 97 | $boostQuery->addFunction( |
| 98 | 'levenshtein_distance_score', |
| 99 | [ |
| 100 | 'text' => $text, |
| 101 | 'field' => 'content' |
| 102 | ] |
| 103 | ); |
| 104 | $boostQuery->setBoostMode( FunctionScore::BOOST_MODE_REPLACE ); |
| 105 | |
| 106 | // Wrap the fuzzy query, so it can be used as a filter. |
| 107 | // This is slightly faster, as ES can throw away the scores by this query. |
| 108 | $bool = new BoolQuery(); |
| 109 | $bool->addFilter( $fuzzyQuery ); |
| 110 | $bool->addMust( $boostQuery ); |
| 111 | |
| 112 | $languageFilter = new Term(); |
| 113 | $languageFilter->setTerm( 'language', $sourceLanguage ); |
| 114 | $bool->addFilter( $languageFilter ); |
| 115 | |
| 116 | // The whole query |
| 117 | $query = new Query(); |
| 118 | $query->setQuery( $bool ); |
| 119 | |
| 120 | // The interface usually displays three best candidates. These might |
| 121 | // come from more than three source things, if the translations are |
| 122 | // the same. In other words suggestions are grouped by the suggested |
| 123 | // translation. This algorithm might not find all suggestions, if the |
| 124 | // top N best matching source texts don't have equivalent translations |
| 125 | // in the target language, but worse matches which we did not fetch do. |
| 126 | // This code tries to balance between doing too many or too big queries |
| 127 | // and not fetching enough results to show all possible suggestions. |
| 128 | $sizeFirst = 100; |
| 129 | $sizeSecond = $sizeFirst * 5; |
| 130 | |
| 131 | $query->setFrom( 0 ); |
| 132 | $query->setSize( $sizeFirst ); |
| 133 | $query->setParam( '_source', [ 'content' ] ); |
| 134 | $cutoff = $this->config['cutoff'] ?? 0.65; |
| 135 | $query->setParam( 'min_score', $cutoff ); |
| 136 | $query->setSort( [ '_score', 'wiki', 'localid' ] ); |
| 137 | |
| 138 | /* This query is doing two unrelated things: |
| 139 | * 1) Collect the message contents and scores so that they can |
| 140 | * be accessed later for the translations we found. |
| 141 | * 2) Build the query string for the query that fetches the translations. |
| 142 | */ |
| 143 | $contents = $scores = $terms = []; |
| 144 | do { |
| 145 | $resultset = $this->getIndex()->search( $query ); |
| 146 | |
| 147 | if ( count( $resultset ) === 0 ) { |
| 148 | break; |
| 149 | } |
| 150 | |
| 151 | foreach ( $resultset->getResults() as $result ) { |
| 152 | $data = $result->getData(); |
| 153 | $score = $result->getScore(); |
| 154 | |
| 155 | $sourceId = preg_replace( '~/[^/]+$~', '', $result->getId() ); |
| 156 | $contents[$sourceId] = $data['content']; |
| 157 | $scores[$sourceId] = $score; |
| 158 | $terms[] = "$sourceId/$targetLanguage"; |
| 159 | } |
| 160 | |
| 161 | // Check if it looks like that we are hitting the long tail already. |
| 162 | // Otherwise, we'll do a query to fetch some more to reach a "sane" |
| 163 | // breaking point, i.e. include all suggestions with same content |
| 164 | // for reliable used X times statistics. |
| 165 | if ( count( array_unique( $scores ) ) > 5 ) { |
| 166 | break; |
| 167 | } |
| 168 | |
| 169 | // Okay, We are now in second iteration of the loop. We already got |
| 170 | // lots of suggestions. We will give up for now even if it means we |
| 171 | // return in some sense incomplete results. |
| 172 | if ( count( $resultset ) === $sizeSecond ) { |
| 173 | break; |
| 174 | } |
| 175 | |
| 176 | // After the first query, the smallest score is the new threshold. |
| 177 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable |
| 178 | $query->setParam( 'min_score', $score ); |
| 179 | $query->setFrom( $query->getParam( 'size' ) + $query->getParam( 'from' ) ); |
| 180 | $query->setSize( $sizeSecond ); |
| 181 | |
| 182 | // Break if we already got all hits |
| 183 | } while ( $resultset->getTotalHits() > count( $contents ) ); |
| 184 | |
| 185 | $suggestions = []; |
| 186 | |
| 187 | // Skip second query if first query found nothing. Keeping only one return |
| 188 | // statement in this method to avoid forgetting to reset connection timeout |
| 189 | if ( $terms !== [] ) { |
| 190 | $idQuery = new Query\Terms( '_id', $terms ); |
| 191 | |
| 192 | $query = new Query( $idQuery ); |
| 193 | $query->setSize( 25 ); |
| 194 | $query->setParam( '_source', [ 'wiki', 'uri', 'content', 'localid' ] ); |
| 195 | $resultset = $this->getIndex()->search( $query ); |
| 196 | |
| 197 | foreach ( $resultset->getResults() as $result ) { |
| 198 | $data = $result->getData(); |
| 199 | |
| 200 | // Construct the matching source id |
| 201 | $sourceId = preg_replace( '~/[^/]+$~', '', $result->getId() ); |
| 202 | |
| 203 | $suggestions[] = [ |
| 204 | 'source' => $contents[$sourceId], |
| 205 | 'target' => $data['content'], |
| 206 | 'context' => $data['localid'], |
| 207 | 'quality' => $scores[$sourceId], |
| 208 | 'wiki' => $data['wiki'], |
| 209 | 'location' => $data['localid'] . '/' . $targetLanguage, |
| 210 | 'uri' => $data['uri'], |
| 211 | ]; |
| 212 | } |
| 213 | |
| 214 | // Ensure results are in quality order |
| 215 | uasort( $suggestions, static function ( $a, $b ) { |
| 216 | if ( $a['quality'] === $b['quality'] ) { |
| 217 | return 0; |
| 218 | } |
| 219 | |
| 220 | return ( $a['quality'] < $b['quality'] ) ? 1 : -1; |
| 221 | } ); |
| 222 | } |
| 223 | |
| 224 | $connection->setTimeout( $oldTimeout ); |
| 225 | |
| 226 | return $suggestions; |
| 227 | } |
| 228 | |
| 229 | /* Write functions */ |
| 230 | |
| 231 | public function update( MessageHandle $handle, ?string $targetText ): bool { |
| 232 | if ( !$handle->isValid() || $handle->getCode() === '' ) { |
| 233 | return false; |
| 234 | } |
| 235 | |
| 236 | /* There are various different cases here: |
| 237 | * [new or updated] [fuzzy|non-fuzzy] [translation|definition] |
| 238 | * 1) We don't distinguish between new or updated here. |
| 239 | * 2) Delete old translation, but not definition |
| 240 | * 3) Insert new translation or definition, if non-fuzzy |
| 241 | * The definition should never be fuzzied anyway. |
| 242 | * |
| 243 | * These only apply to known messages. |
| 244 | */ |
| 245 | |
| 246 | $sourceLanguage = $handle->getGroup()->getSourceLanguage(); |
| 247 | |
| 248 | // Do not delete definitions, because the translations are attached to that |
| 249 | if ( $handle->getCode() !== $sourceLanguage ) { |
| 250 | $localid = $handle->getTitleForBase()->getPrefixedText(); |
| 251 | $this->deleteByQuery( $this->getIndex(), Query::create( |
| 252 | ( new BoolQuery() ) |
| 253 | ->addFilter( new Term( [ 'wiki' => WikiMap::getCurrentWikiId() ] ) ) |
| 254 | ->addFilter( new Term( [ 'language' => $handle->getCode() ] ) ) |
| 255 | ->addFilter( new Term( [ 'localid' => $localid ] ) ) ) ); |
| 256 | } |
| 257 | |
| 258 | // If translation was made fuzzy, we do not need to add anything |
| 259 | if ( $targetText === null ) { |
| 260 | return true; |
| 261 | } |
| 262 | |
| 263 | // source language is null, skip doing rest of the stuff |
| 264 | if ( $sourceLanguage === null ) { |
| 265 | return true; |
| 266 | } |
| 267 | |
| 268 | $revId = $handle->getTitleForLanguage( $sourceLanguage )->getLatestRevID(); |
| 269 | $doc = $this->createDocument( $handle, $targetText, $revId ); |
| 270 | $fname = __METHOD__; |
| 271 | |
| 272 | MWElasticUtils::withRetry( self::BULK_INDEX_RETRY_ATTEMPTS, |
| 273 | function () use ( $doc ) { |
| 274 | $this->getIndex()->addDocuments( [ $doc ] ); |
| 275 | }, |
| 276 | static function ( $e, $errors ) use ( $fname ) { |
| 277 | $c = get_class( $e ); |
| 278 | $msg = $e->getMessage(); |
| 279 | error_log( $fname . ": update failed ($c: $msg); retrying." ); |
| 280 | sleep( 10 ); |
| 281 | } |
| 282 | ); |
| 283 | |
| 284 | return true; |
| 285 | } |
| 286 | |
| 287 | private function createDocument( MessageHandle $handle, string $text, int $revId ): Document { |
| 288 | $language = $handle->getCode(); |
| 289 | |
| 290 | $localid = $handle->getTitleForBase()->getPrefixedText(); |
| 291 | $wiki = WikiMap::getCurrentWikiId(); |
| 292 | $globalid = "$wiki-$localid-$revId/$language"; |
| 293 | |
| 294 | $data = [ |
| 295 | 'wiki' => $wiki, |
| 296 | 'uri' => $handle->getTitle()->getCanonicalURL(), |
| 297 | 'localid' => $localid, |
| 298 | 'language' => $language, |
| 299 | 'content' => $text, |
| 300 | 'group' => $handle->getGroupIds(), |
| 301 | ]; |
| 302 | |
| 303 | return new Document( $globalid, $data, '_doc' ); |
| 304 | } |
| 305 | |
| 306 | /** @param bool $rebuild Deletes index first if already exists */ |
| 307 | private function createIndex( bool $rebuild ): void { |
| 308 | $indexSettings = [ |
| 309 | 'settings' => [ |
| 310 | 'index' => [ |
| 311 | 'number_of_shards' => $this->getShardCount(), |
| 312 | 'analysis' => [ |
| 313 | 'filter' => [ |
| 314 | 'prefix_filter' => [ |
| 315 | 'type' => 'edge_ngram', |
| 316 | 'min_gram' => 2, |
| 317 | 'max_gram' => 20 |
| 318 | ] |
| 319 | ], |
| 320 | 'analyzer' => [ |
| 321 | 'prefix' => [ |
| 322 | 'type' => 'custom', |
| 323 | 'tokenizer' => 'standard', |
| 324 | 'filter' => [ 'lowercase', 'prefix_filter' ] |
| 325 | ], |
| 326 | 'casesensitive' => [ |
| 327 | 'tokenizer' => 'standard' |
| 328 | ] |
| 329 | ] |
| 330 | ] |
| 331 | ], |
| 332 | ], |
| 333 | ]; |
| 334 | $replicas = $this->getReplicaCount(); |
| 335 | $key = str_contains( $replicas, '-' ) ? 'auto_expand_replicas' : 'number_of_replicas'; |
| 336 | $indexSettings['settings']['index'][$key] = $replicas; |
| 337 | |
| 338 | $this->getIndex()->create( $indexSettings, $rebuild ); |
| 339 | } |
| 340 | |
| 341 | /** |
| 342 | * Begin the bootstrap process. |
| 343 | * @throws RuntimeException |
| 344 | */ |
| 345 | public function beginBootstrap(): void { |
| 346 | $this->checkElasticsearchVersion(); |
| 347 | $index = $this->getIndex(); |
| 348 | if ( $this->updateMapping ) { |
| 349 | $this->logOutput( 'Updating the index mappings...' ); |
| 350 | $this->createIndex( true ); |
| 351 | } elseif ( !$index->exists() ) { |
| 352 | $this->createIndex( false ); |
| 353 | } |
| 354 | |
| 355 | $settings = $index->getSettings(); |
| 356 | $settings->setRefreshInterval( '-1' ); |
| 357 | |
| 358 | $this->deleteByQuery( $this->getIndex(), Query::create( |
| 359 | ( new Term() )->setTerm( 'wiki', WikiMap::getCurrentWikiId() ) ) ); |
| 360 | |
| 361 | $properties = [ |
| 362 | 'wiki' => [ 'type' => 'keyword' ], |
| 363 | 'localid' => [ 'type' => 'keyword' ], |
| 364 | 'uri' => [ 'type' => 'keyword' ], |
| 365 | 'language' => [ 'type' => 'keyword' ], |
| 366 | 'group' => [ 'type' => 'keyword' ], |
| 367 | 'content' => [ |
| 368 | 'type' => 'text', |
| 369 | 'fields' => [ |
| 370 | 'content' => [ |
| 371 | 'type' => 'text', |
| 372 | 'term_vector' => 'yes' |
| 373 | ], |
| 374 | 'prefix_complete' => [ |
| 375 | 'type' => 'text', |
| 376 | 'analyzer' => 'prefix', |
| 377 | 'search_analyzer' => 'standard', |
| 378 | 'term_vector' => 'yes' |
| 379 | ], |
| 380 | 'case_sensitive' => [ |
| 381 | 'type' => 'text', |
| 382 | 'analyzer' => 'casesensitive', |
| 383 | 'term_vector' => 'yes' |
| 384 | ] |
| 385 | ] |
| 386 | ], |
| 387 | ]; |
| 388 | |
| 389 | $mapping = new Mapping( $properties ); |
| 390 | $mapping->send( $index, [ 'include_type_name' => 'false' ] ); |
| 391 | |
| 392 | $this->waitUntilReady(); |
| 393 | } |
| 394 | |
| 395 | public function beginBatch(): void { |
| 396 | } |
| 397 | |
| 398 | /** |
| 399 | * @param array[] $batch |
| 400 | * @phan-param array<int,array{0:MessageHandle,1:string,2:string}> $batch |
| 401 | */ |
| 402 | public function batchInsertDefinitions( array $batch ): void { |
| 403 | $lb = MediaWikiServices::getInstance()->getLinkBatchFactory()->newLinkBatch(); |
| 404 | foreach ( $batch as $data ) { |
| 405 | $lb->addObj( $data[0]->getTitle() ); |
| 406 | } |
| 407 | $lb->execute(); |
| 408 | |
| 409 | $this->batchInsertTranslations( $batch ); |
| 410 | } |
| 411 | |
| 412 | public function batchInsertTranslations( array $batch ): void { |
| 413 | $docs = []; |
| 414 | foreach ( $batch as $data ) { |
| 415 | [ $handle, $sourceLanguage, $text ] = $data; |
| 416 | $revId = $handle->getTitleForLanguage( $sourceLanguage )->getLatestRevID(); |
| 417 | $docs[] = $this->createDocument( $handle, $text, $revId ); |
| 418 | } |
| 419 | |
| 420 | MWElasticUtils::withRetry( self::BULK_INDEX_RETRY_ATTEMPTS, |
| 421 | function () use ( $docs ) { |
| 422 | $this->getIndex()->addDocuments( $docs ); |
| 423 | }, |
| 424 | function ( $e, $errors ) { |
| 425 | $c = get_class( $e ); |
| 426 | $msg = $e->getMessage(); |
| 427 | $this->logOutput( "Batch failed ($c: $msg), trying again in 10 seconds" ); |
| 428 | sleep( 10 ); |
| 429 | } |
| 430 | ); |
| 431 | } |
| 432 | |
| 433 | public function endBatch(): void { |
| 434 | } |
| 435 | |
| 436 | public function endBootstrap(): void { |
| 437 | $index = $this->getIndex(); |
| 438 | $index->refresh(); |
| 439 | $index->forcemerge(); |
| 440 | $index->getSettings()->setRefreshInterval( '5s' ); |
| 441 | } |
| 442 | |
| 443 | public function getClient(): Client { |
| 444 | if ( $this->client === null ) { |
| 445 | if ( isset( $this->config['config'] ) ) { |
| 446 | $this->client = new Client( $this->config['config'] ); |
| 447 | } else { |
| 448 | $this->client = new Client(); |
| 449 | } |
| 450 | } |
| 451 | return $this->client; |
| 452 | } |
| 453 | |
| 454 | /** @return true if the backend is configured with the wikimedia extra plugin */ |
| 455 | public function useWikimediaExtraPlugin(): bool { |
| 456 | return isset( $this->config['use_wikimedia_extra'] ) && $this->config['use_wikimedia_extra']; |
| 457 | } |
| 458 | |
| 459 | private function getIndexName(): string { |
| 460 | return $this->config['index'] ?? 'ttmserver'; |
| 461 | } |
| 462 | |
| 463 | public function getIndex(): Index { |
| 464 | return $this->getClient() |
| 465 | ->getIndex( $this->getIndexName() ); |
| 466 | } |
| 467 | |
| 468 | private function getShardCount(): int { |
| 469 | return $this->config['shards'] ?? 1; |
| 470 | } |
| 471 | |
| 472 | private function getReplicaCount(): string { |
| 473 | return $this->config['replicas'] ?? '0-2'; |
| 474 | } |
| 475 | |
| 476 | private function waitUntilReady(): void { |
| 477 | $statuses = MWElasticUtils::waitForGreen( |
| 478 | $this->getClient(), |
| 479 | $this->getIndexName(), |
| 480 | self::WAIT_UNTIL_READY_TIMEOUT ); |
| 481 | $this->logOutput( "Waiting for the index to go green..." ); |
| 482 | foreach ( $statuses as $message ) { |
| 483 | $this->logOutput( $message ); |
| 484 | } |
| 485 | |
| 486 | if ( !$statuses->getReturn() ) { |
| 487 | die( "Timeout! Please check server logs for {$this->getIndexName()}." ); |
| 488 | } |
| 489 | } |
| 490 | |
| 491 | public function setLogger( TTMServerBootstrap $logger ): void { |
| 492 | $this->logger = $logger; |
| 493 | } |
| 494 | |
| 495 | // Can it get any uglier? |
| 496 | private function logOutput( string $text ): void { |
| 497 | if ( $this->logger !== null ) { |
| 498 | $this->logger->statusLine( "$text\n" ); |
| 499 | } |
| 500 | } |
| 501 | |
| 502 | public function setDoReIndex(): void { |
| 503 | $this->updateMapping = true; |
| 504 | } |
| 505 | |
| 506 | /** Parse query string and build the search query */ |
| 507 | private function parseQueryString( string $queryString, array $opts ): array { |
| 508 | $fields = $highlights = []; |
| 509 | $terms = preg_split( '/\s+/', $queryString ); |
| 510 | $match = $opts['match']; |
| 511 | $case = $opts['case']; |
| 512 | |
| 513 | // Map each word in the query string with its corresponding field |
| 514 | foreach ( $terms as $term ) { |
| 515 | $prefix = strstr( $term, '*', true ); |
| 516 | if ( $prefix ) { |
| 517 | // For wildcard search |
| 518 | $fields['content.prefix_complete'][] = $prefix; |
| 519 | } elseif ( $case === '1' ) { |
| 520 | // For case-sensitive search |
| 521 | $fields['content.case_sensitive'][] = $term; |
| 522 | } else { |
| 523 | $fields['content'][] = $term; |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | // Allow searching either by message content or message id (page name |
| 528 | // without language subpage) with exact match only. |
| 529 | $searchQuery = new BoolQuery(); |
| 530 | foreach ( $fields as $analyzer => $words ) { |
| 531 | foreach ( $words as $word ) { |
| 532 | $boolQuery = new BoolQuery(); |
| 533 | $contentQuery = new MatchQuery(); |
| 534 | $contentQuery->setFieldQuery( $analyzer, $word ); |
| 535 | $boolQuery->addShould( $contentQuery ); |
| 536 | $messageQuery = new Term(); |
| 537 | $messageQuery->setTerm( 'localid', $word ); |
| 538 | $boolQuery->addShould( $messageQuery ); |
| 539 | |
| 540 | if ( $match === 'all' ) { |
| 541 | $searchQuery->addMust( $boolQuery ); |
| 542 | } else { |
| 543 | $searchQuery->addShould( $boolQuery ); |
| 544 | } |
| 545 | |
| 546 | // Fields for highlighting |
| 547 | $highlights[$analyzer] = [ |
| 548 | 'number_of_fragments' => 0 |
| 549 | ]; |
| 550 | |
| 551 | // Allow searching by exact message title (page name with |
| 552 | // language subpage). |
| 553 | $title = Title::newFromText( $word ); |
| 554 | if ( !$title ) { |
| 555 | continue; |
| 556 | } |
| 557 | $handle = new MessageHandle( $title ); |
| 558 | if ( $handle->isValid() && $handle->getCode() !== '' ) { |
| 559 | $localid = $handle->getTitleForBase()->getPrefixedText(); |
| 560 | $boolQuery = new BoolQuery(); |
| 561 | $messageId = new Term(); |
| 562 | $messageId->setTerm( 'localid', $localid ); |
| 563 | $boolQuery->addMust( $messageId ); |
| 564 | $searchQuery->addShould( $boolQuery ); |
| 565 | } |
| 566 | } |
| 567 | } |
| 568 | |
| 569 | return [ $searchQuery, $highlights ]; |
| 570 | } |
| 571 | |
| 572 | /** Search interface */ |
| 573 | public function createSearch( string $queryString, array $opts, array $highlight ): Search { |
| 574 | $query = new Query(); |
| 575 | |
| 576 | [ $searchQuery, $highlights ] = $this->parseQueryString( $queryString, $opts ); |
| 577 | $query->setQuery( $searchQuery ); |
| 578 | |
| 579 | $language = new Terms( 'language' ); |
| 580 | $language->setField( 'language' ); |
| 581 | $language->setSize( 500 ); |
| 582 | $query->addAggregation( $language ); |
| 583 | |
| 584 | $group = new Terms( 'group' ); |
| 585 | $group->setField( 'group' ); |
| 586 | // Would like to prioritize the top level groups and not show subgroups |
| 587 | // if the top group has only few hits, but that doesn't seem to be possile. |
| 588 | $group->setSize( 500 ); |
| 589 | $query->addAggregation( $group ); |
| 590 | |
| 591 | $query->setSize( $opts['limit'] ); |
| 592 | $query->setFrom( $opts['offset'] ); |
| 593 | |
| 594 | // BoolAnd filters are executed in sequence per document. Bool filters with |
| 595 | // multiple must clauses are executed by converting each filter into a bit |
| 596 | // field then anding them together. The latter is normally faster if either |
| 597 | // of the subfilters are reused. May not make a difference in this context. |
| 598 | $filters = new BoolQuery(); |
| 599 | |
| 600 | $language = $opts['language']; |
| 601 | if ( $language !== '' ) { |
| 602 | $languageFilter = new Term(); |
| 603 | $languageFilter->setTerm( 'language', $language ); |
| 604 | $filters->addFilter( $languageFilter ); |
| 605 | } |
| 606 | |
| 607 | $group = $opts['group']; |
| 608 | if ( $group !== '' ) { |
| 609 | $groupFilter = new Term(); |
| 610 | $groupFilter->setTerm( 'group', $group ); |
| 611 | $filters->addFilter( $groupFilter ); |
| 612 | } |
| 613 | |
| 614 | // Check that we have at least one filter to avoid invalid query errors. |
| 615 | if ( $language !== '' || $group !== '' ) { |
| 616 | // TODO: This seems wrong, but perhaps for aggregation purposes? |
| 617 | // should make $search a must clause and use the bool query |
| 618 | // as main. |
| 619 | $query->setPostFilter( $filters ); |
| 620 | } |
| 621 | |
| 622 | [ $pre, $post ] = $highlight; |
| 623 | $query->setHighlight( [ |
| 624 | // The value must be an object |
| 625 | 'pre_tags' => [ $pre ], |
| 626 | 'post_tags' => [ $post ], |
| 627 | 'fields' => $highlights, |
| 628 | ] ); |
| 629 | |
| 630 | return $this->getIndex()->createSearch( $query ); |
| 631 | } |
| 632 | |
| 633 | /** |
| 634 | * Search interface |
| 635 | * @throws TtmServerException |
| 636 | */ |
| 637 | public function search( string $queryString, array $opts, array $highlight ): ResultSet { |
| 638 | $search = $this->createSearch( $queryString, $opts, $highlight ); |
| 639 | |
| 640 | try { |
| 641 | return $search->search(); |
| 642 | } catch ( ExceptionInterface $e ) { |
| 643 | throw new TtmServerException( $e->getMessage() ); |
| 644 | } |
| 645 | } |
| 646 | |
| 647 | /** @inheritDoc */ |
| 648 | public function getFacets( $resultset ): array { |
| 649 | $this->assertResultSetInstance( $resultset ); |
| 650 | $aggs = $resultset->getAggregations(); |
| 651 | '@phan-var array[][][] $aggs'; |
| 652 | |
| 653 | $ret = [ |
| 654 | 'language' => [], |
| 655 | 'group' => [] |
| 656 | ]; |
| 657 | |
| 658 | foreach ( $aggs as $type => $info ) { |
| 659 | foreach ( $info['buckets'] as $row ) { |
| 660 | $ret[$type][$row['key']] = $row['doc_count']; |
| 661 | } |
| 662 | } |
| 663 | |
| 664 | return $ret; |
| 665 | } |
| 666 | |
| 667 | /** @inheritDoc */ |
| 668 | public function getTotalHits( $resultset ): int { |
| 669 | $this->assertResultSetInstance( $resultset ); |
| 670 | return $resultset->getTotalHits(); |
| 671 | } |
| 672 | |
| 673 | /** @inheritDoc */ |
| 674 | public function getDocuments( $resultset ): array { |
| 675 | $this->assertResultSetInstance( $resultset ); |
| 676 | $ret = []; |
| 677 | foreach ( $resultset->getResults() as $document ) { |
| 678 | $data = $document->getData(); |
| 679 | $hl = $document->getHighlights(); |
| 680 | if ( isset( $hl['content.prefix_complete'][0] ) ) { |
| 681 | $data['content'] = $hl['content.prefix_complete'][0]; |
| 682 | } elseif ( isset( $hl['content.case_sensitive'][0] ) ) { |
| 683 | $data['content'] = $hl['content.case_sensitive'][0]; |
| 684 | } elseif ( isset( $hl['content'][0] ) ) { |
| 685 | $data['content'] = $hl['content'][0]; |
| 686 | } |
| 687 | $ret[] = $data; |
| 688 | } |
| 689 | |
| 690 | return $ret; |
| 691 | } |
| 692 | |
| 693 | /** |
| 694 | * Delete docs by query by using the scroll API. |
| 695 | * TODO: Elastica\Index::deleteByQuery() ? was removed |
| 696 | * in 2.x and returned in 5.x. |
| 697 | * @throws RuntimeException |
| 698 | */ |
| 699 | private function deleteByQuery( Index $sourceIndex, Query $query ): void { |
| 700 | try { |
| 701 | MWElasticUtils::deleteByQuery( $sourceIndex, $query, /* $allowConflicts = */ true ); |
| 702 | } catch ( Exception $e ) { |
| 703 | LoggerFactory::getInstance( LogNames::ELASTIC_SEARCH_TTMSERVER )->error( |
| 704 | 'Problem encountered during deletion.', |
| 705 | [ 'exception' => $e ] |
| 706 | ); |
| 707 | |
| 708 | throw new RuntimeException( "Problem encountered during deletion.\n" . $e ); |
| 709 | } |
| 710 | } |
| 711 | |
| 712 | /* @throws RuntimeException */ |
| 713 | private function getElasticsearchVersion(): string { |
| 714 | $response = $this->getClient()->request( '' ); |
| 715 | if ( !$response->isOK() ) { |
| 716 | throw new RuntimeException( "Cannot fetch elasticsearch version: " . $response->getError() ); |
| 717 | } |
| 718 | |
| 719 | $result = $response->getData(); |
| 720 | if ( !isset( $result['version']['number'] ) ) { |
| 721 | throw new RuntimeException( 'Unable to determine elasticsearch version, aborting.' ); |
| 722 | } |
| 723 | |
| 724 | return $result[ 'version' ][ 'number' ]; |
| 725 | } |
| 726 | |
| 727 | private function checkElasticsearchVersion(): void { |
| 728 | $version = $this->getElasticsearchVersion(); |
| 729 | if ( !str_starts_with( $version, '6.8' ) && !str_starts_with( $version, '7.' ) ) { |
| 730 | throw new RuntimeException( "Only Elasticsearch 6.8.x and 7.x are supported. Your version: $version." ); |
| 731 | } |
| 732 | } |
| 733 | |
| 734 | /** @param mixed $resultSet */ |
| 735 | private function assertResultSetInstance( $resultSet ): void { |
| 736 | if ( $resultSet instanceof ResultSet ) { |
| 737 | return; |
| 738 | } |
| 739 | |
| 740 | throw new RuntimeException( |
| 741 | "Expected resultset to be an instance of " . ResultSet::class |
| 742 | ); |
| 743 | } |
| 744 | } |
| 745 | |
| 746 | // Translation memory configuration ($wgTranslateTranslationServices) uses class as ElasticSearchTTMServer |
| 747 | // See: https://www.mediawiki.org/wiki/Help:Extension:Translate/Translation_memories#Configuration |
| 748 | class_alias( ElasticSearchTtmServer::class, "ElasticSearchTTMServer" ); |