Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
46.11% |
77 / 167 |
|
33.33% |
5 / 15 |
CRAP | |
0.00% |
0 / 1 |
| SearchRequestBuilder | |
46.11% |
77 / 167 |
|
33.33% |
5 / 15 |
584.54 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| build | |
31.86% |
36 / 113 |
|
0.00% |
0 / 1 |
399.76 | |||
| getOffset | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setOffset | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getLimit | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setLimit | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getTimeout | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setTimeout | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getIndex | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
4 | |||
| isCrossClusterSearch | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| inferIndexFromConcreteNamespaceMap | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
7 | |||
| setIndex | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getSort | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setSort | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getSearchContext | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Search; |
| 4 | |
| 5 | use CirrusSearch\CirrusConfigNames; |
| 6 | use CirrusSearch\Connection; |
| 7 | use CirrusSearch\LogChannel; |
| 8 | use CirrusSearch\SearchConfig; |
| 9 | use CirrusSearch\Util; |
| 10 | use Elastica\Index; |
| 11 | use Elastica\Query; |
| 12 | use MediaWiki\Logger\LoggerFactory; |
| 13 | use Wikimedia\Assert\Assert; |
| 14 | |
| 15 | /** |
| 16 | * Build the search request body |
| 17 | */ |
| 18 | class SearchRequestBuilder { |
| 19 | /** @var SearchContext */ |
| 20 | private $searchContext; |
| 21 | |
| 22 | /** @var Connection */ |
| 23 | private $connection; |
| 24 | |
| 25 | /** @var string */ |
| 26 | private $indexBaseName; |
| 27 | |
| 28 | /** @var int */ |
| 29 | private $offset = 0; |
| 30 | |
| 31 | /** @var int */ |
| 32 | private $limit = 20; |
| 33 | |
| 34 | /** @var string search timeout, string with time and unit, e.g. 20s for 20 seconds */ |
| 35 | private $timeout; |
| 36 | |
| 37 | /** |
| 38 | * @var Index|null force the index when set, use {@link Connection::pickIndexSuffixForNamespaces} |
| 39 | */ |
| 40 | private $index; |
| 41 | |
| 42 | /** @var string set the sort option, controls the use of rescore functions or elastic sort */ |
| 43 | private $sort = 'relevance'; |
| 44 | |
| 45 | /** |
| 46 | * @param SearchContext $searchContext |
| 47 | * @param Connection $connection |
| 48 | * @param string $indexBaseName |
| 49 | */ |
| 50 | public function __construct( SearchContext $searchContext, Connection $connection, $indexBaseName ) { |
| 51 | $this->searchContext = $searchContext; |
| 52 | $this->connection = $connection; |
| 53 | $this->indexBaseName = $indexBaseName; |
| 54 | } |
| 55 | |
| 56 | /** |
| 57 | * Build the search request |
| 58 | * @return \Elastica\Search |
| 59 | */ |
| 60 | public function build() { |
| 61 | $resultsType = $this->searchContext->getResultsType(); |
| 62 | |
| 63 | $query = new Query(); |
| 64 | // Track at least offset + limit + 1 hits if precise total_hits is not requested |
| 65 | // This useful to know if more results are available on the next page |
| 66 | $query->setTrackTotalHits( $this->searchContext->getTrackTotalHits() ? true : $this->offset + $this->limit + 1 ); |
| 67 | $query->setSource( $resultsType->getSourceFiltering() ); |
| 68 | $query->setParam( "fields", $resultsType->getFields() ); |
| 69 | |
| 70 | $extraIndexes = $this->searchContext->getExtraIndices(); |
| 71 | |
| 72 | if ( $extraIndexes && |
| 73 | $this->searchContext->getConfig()->getElement( CirrusConfigNames::DeduplicateInQuery ) !== false |
| 74 | ) { |
| 75 | $this->searchContext->addNotFilter( new \Elastica\Query\Term( |
| 76 | [ 'local_sites_with_dupe' => $this->indexBaseName ] |
| 77 | ) ); |
| 78 | } |
| 79 | |
| 80 | $mainQuery = $this->searchContext->getQuery(); |
| 81 | $query->setQuery( $mainQuery ); |
| 82 | |
| 83 | foreach ( $this->searchContext->getAggregations() as $agg ) { |
| 84 | $query->addAggregation( $agg ); |
| 85 | } |
| 86 | |
| 87 | $highlight = $this->searchContext->getHighlight( $resultsType, $mainQuery ); |
| 88 | if ( $highlight ) { |
| 89 | $query->setHighlight( $highlight ); |
| 90 | } |
| 91 | |
| 92 | $suggestQueries = $this->searchContext->getFallbackRunner()->getElasticSuggesters(); |
| 93 | if ( $suggestQueries ) { |
| 94 | $query->setParam( 'suggest', [ |
| 95 | // TODO: remove special case on 1-elt array, added to not change the test fixtures |
| 96 | // We should switch to explicit naming |
| 97 | 'suggest' => count( $suggestQueries ) === 1 ? reset( $suggestQueries ) : $suggestQueries |
| 98 | ] ); |
| 99 | $query->addParam( 'stats', 'suggest' ); |
| 100 | } |
| 101 | |
| 102 | foreach ( $this->searchContext->getSyntaxUsed() as $syntax ) { |
| 103 | $query->addParam( 'stats', $syntax ); |
| 104 | } |
| 105 | |
| 106 | // See also CirrusSearch::getValidSorts() |
| 107 | switch ( $this->sort ) { |
| 108 | case 'just_match': |
| 109 | // Use just matching scores, without any rescoring, and default sort. |
| 110 | break; |
| 111 | case 'relevance': |
| 112 | // Add some rescores to improve relevance |
| 113 | $rescores = $this->searchContext->getRescore(); |
| 114 | if ( $rescores !== [] ) { |
| 115 | $query->setParam( 'rescore', $rescores ); |
| 116 | } |
| 117 | break; // The default |
| 118 | case 'create_timestamp_asc': |
| 119 | $query->setSort( [ 'create_timestamp' => 'asc' ] ); |
| 120 | break; |
| 121 | case 'create_timestamp_desc': |
| 122 | $query->setSort( [ 'create_timestamp' => 'desc' ] ); |
| 123 | break; |
| 124 | case 'last_edit_asc': |
| 125 | $query->setSort( [ 'timestamp' => 'asc' ] ); |
| 126 | break; |
| 127 | case 'last_edit_desc': |
| 128 | $query->setSort( [ 'timestamp' => 'desc' ] ); |
| 129 | break; |
| 130 | case 'incoming_links_asc': |
| 131 | $query->setSort( [ 'incoming_links' => [ |
| 132 | 'order' => 'asc', |
| 133 | 'missing' => '_first', |
| 134 | ] ] ); |
| 135 | break; |
| 136 | case 'incoming_links_desc': |
| 137 | $query->setSort( [ 'incoming_links' => [ |
| 138 | 'order' => 'desc', |
| 139 | 'missing' => '_last', |
| 140 | ] ] ); |
| 141 | break; |
| 142 | case 'none': |
| 143 | // Return documents in index order |
| 144 | $query->setSort( [ '_doc' ] ); |
| 145 | break; |
| 146 | case 'random': |
| 147 | $randomSeed = $this->searchContext->getSearchQuery()->getRandomSeed(); |
| 148 | if ( $randomSeed === null && $this->offset !== 0 ) { |
| 149 | $this->searchContext->addWarning( 'cirrussearch-offset-not-allowed-with-random-sort' ); |
| 150 | $this->offset = 0; |
| 151 | } |
| 152 | // Can't use an empty array, it would JSONify to [] instead of {}. |
| 153 | $scoreParams = ( $randomSeed === null ) ? (object)[] : [ 'seed' => $randomSeed, 'field' => '_seq_no' ]; |
| 154 | // Instead of setting a sort field wrap the whole query in a |
| 155 | // bool filter and add a must clause for the random score. This |
| 156 | // could alternatively be a rescore over a limited document |
| 157 | // set, but in basic testing the filter was more performant |
| 158 | // than an 8k rescore window even with 50M total hits. |
| 159 | $query->setQuery( ( new Query\BoolQuery() ) |
| 160 | ->addFilter( $mainQuery ) |
| 161 | ->addMust( ( new Query\FunctionScore() ) |
| 162 | ->setQuery( new Query\MatchAll() ) |
| 163 | ->addFunction( 'random_score', $scoreParams ) ) ); |
| 164 | |
| 165 | break; |
| 166 | case 'user_random': |
| 167 | // Randomly ordered, but consistent for a single user |
| 168 | $query->setQuery( ( new Query\BoolQuery() ) |
| 169 | ->addFilter( $mainQuery ) |
| 170 | ->addMust( ( new Query\FunctionScore() ) |
| 171 | ->setQuery( new Query\MatchAll() ) |
| 172 | ->addFunction( 'random_score', [ |
| 173 | 'seed' => Util::generateIdentToken(), |
| 174 | 'field' => '_seq_no', |
| 175 | ] ) ) ); |
| 176 | break; |
| 177 | |
| 178 | case 'title_natural_asc': |
| 179 | case 'title_natural_desc': |
| 180 | if ( $this->searchContext->getConfig()->getElement( CirrusConfigNames::NaturalTitleSort, 'use' ) ) { |
| 181 | $dir = explode( '_', $this->sort, 3 )[2]; |
| 182 | $query->setSort( [ |
| 183 | 'namespace_text' => $dir, |
| 184 | 'title.natural_sort' => $dir, |
| 185 | ] ); |
| 186 | break; |
| 187 | } |
| 188 | // Intentional fall-through to default error case. |
| 189 | |
| 190 | default: |
| 191 | // Same as just_match. No user warning since an invalid sort |
| 192 | // getting this far is a bug in the calling code which should |
| 193 | // be validating it's input. |
| 194 | LoggerFactory::getInstance( LogChannel::DEFAULT )->warning( |
| 195 | "Invalid sort type: {sort}", |
| 196 | [ 'sort' => $this->sort ] |
| 197 | ); |
| 198 | } |
| 199 | |
| 200 | if ( $this->offset ) { |
| 201 | $query->setFrom( $this->offset ); |
| 202 | } |
| 203 | if ( $this->limit ) { |
| 204 | $query->setSize( $this->limit ); |
| 205 | } |
| 206 | |
| 207 | // Setup the search |
| 208 | $queryOptions = []; |
| 209 | if ( $this->isCrossClusterSearch() ) { |
| 210 | // Ignore unavailable indices during cross-cluster searches. This feature is currently used only by interwiki searches |
| 211 | // for which we purposedly ignore errors. Instead of logging a search failure we let opensearch ignore this problem for us. |
| 212 | // Missing indices might be intermittent because ccs_minimize_roundtrips is on by default, this means that the coordinator will |
| 213 | // do minimal efforts to check that the target index exists in its view of the remote cluster cluster state. |
| 214 | // Disabling ccs_minimize_roundtrips could be an option but at the cost of increased latencies which is probably |
| 215 | // not something worth the cost for interwiki searches. |
| 216 | $queryOptions[\Elastica\Search::OPTION_SEARCH_IGNORE_UNAVAILABLE] = true; |
| 217 | } |
| 218 | if ( $this->timeout ) { |
| 219 | $queryOptions[\Elastica\Search::OPTION_TIMEOUT] = $this->timeout; |
| 220 | } |
| 221 | // @todo when switching to multi-search this has to be provided at the top level |
| 222 | if ( $this->searchContext->getConfig()->get( CirrusConfigNames::MoreAccurateScoringMode ) ) { |
| 223 | $queryOptions[\Elastica\Search::OPTION_SEARCH_TYPE] = \Elastica\Search::OPTION_SEARCH_TYPE_DFS_QUERY_THEN_FETCH; |
| 224 | } |
| 225 | |
| 226 | $search = $this->getIndex()->createSearch( $query, $queryOptions ); |
| 227 | $crossClusterName = $this->connection->getConfig()->getClusterAssignment()->getCrossClusterName(); |
| 228 | foreach ( $extraIndexes as $i ) { |
| 229 | $search->addIndex( $this->connection->getIndex( $i->getSearchIndex( $crossClusterName ) ) ); |
| 230 | } |
| 231 | |
| 232 | $this->searchContext->getDebugOptions()->applyDebugOptions( $query ); |
| 233 | return $search; |
| 234 | } |
| 235 | |
| 236 | /** |
| 237 | * @return int |
| 238 | */ |
| 239 | public function getOffset() { |
| 240 | return $this->offset; |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * @param int $offset |
| 245 | * @return self |
| 246 | */ |
| 247 | public function setOffset( $offset ) { |
| 248 | $this->offset = $offset; |
| 249 | |
| 250 | return $this; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * @return int |
| 255 | */ |
| 256 | public function getLimit() { |
| 257 | return $this->limit; |
| 258 | } |
| 259 | |
| 260 | /** |
| 261 | * @param int $limit |
| 262 | * @return self |
| 263 | */ |
| 264 | public function setLimit( $limit ) { |
| 265 | $this->limit = $limit; |
| 266 | |
| 267 | return $this; |
| 268 | } |
| 269 | |
| 270 | /** |
| 271 | * @return string |
| 272 | */ |
| 273 | public function getTimeout() { |
| 274 | return $this->timeout; |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * @param string $timeout |
| 279 | * @return self |
| 280 | */ |
| 281 | public function setTimeout( $timeout ) { |
| 282 | $this->timeout = $timeout; |
| 283 | |
| 284 | return $this; |
| 285 | } |
| 286 | |
| 287 | /** |
| 288 | * @return \Elastica\Index An elastica type suitable for searching against |
| 289 | * the configured wiki over the host wiki's default connection. |
| 290 | */ |
| 291 | public function getIndex(): \Elastica\Index { |
| 292 | if ( $this->index ) { |
| 293 | return $this->index; |
| 294 | } else { |
| 295 | $indexBaseName = $this->indexBaseName; |
| 296 | $config = $this->searchContext->getConfig(); |
| 297 | $hostConfig = $config->getHostWikiConfig(); |
| 298 | |
| 299 | $indexName = $this->inferIndexFromConcreteNamespaceMap( $config ); |
| 300 | if ( $indexName === null ) { |
| 301 | $indexSuffix = $this->connection->pickIndexSuffixForNamespaces( |
| 302 | $this->searchContext->getNamespaces() ); |
| 303 | $indexName = $this->connection->getIndexName( $indexBaseName, $indexSuffix ); |
| 304 | } |
| 305 | |
| 306 | if ( $this->isCrossClusterSearch() ) { |
| 307 | $local = $hostConfig->getClusterAssignment()->getCrossClusterName(); |
| 308 | $current = $config->getClusterAssignment()->getCrossClusterName(); |
| 309 | Assert::invariant( $local !== $current, 'clusters are different' ); |
| 310 | $indexName = $current . ':' . $indexName; |
| 311 | } |
| 312 | return $this->connection->getIndex( $indexName ); |
| 313 | } |
| 314 | } |
| 315 | |
| 316 | private function isCrossClusterSearch(): bool { |
| 317 | $config = $this->searchContext->getConfig(); |
| 318 | $hostConfig = $config->getHostWikiConfig(); |
| 319 | if ( $hostConfig->get( CirrusConfigNames::CrossClusterSearch ) ) { |
| 320 | $local = $hostConfig->getClusterAssignment()->getCrossClusterName(); |
| 321 | $current = $config->getClusterAssignment()->getCrossClusterName(); |
| 322 | return $local !== $current; |
| 323 | } |
| 324 | return false; |
| 325 | } |
| 326 | |
| 327 | /** |
| 328 | * Attempt to infer the index from the concrete namespace map. |
| 329 | * This is used mainly during crossproject searches where the concrete namespace map |
| 330 | * is provided by the config dump API. |
| 331 | * Since we might want to query namespaces that are unknown to the host wiki, we |
| 332 | * can't use the connection to pick the index suffix. |
| 333 | * Instead, we use the concrete namespace map to infer the index suffix. |
| 334 | * Returns null if the concrete namespace map is not available or if multiple index types |
| 335 | * might be required, in which case we rely on the host wiki connection to pick up the |
| 336 | * right index. |
| 337 | * |
| 338 | * @param SearchConfig $config |
| 339 | * @return string|null |
| 340 | */ |
| 341 | private function inferIndexFromConcreteNamespaceMap( SearchConfig $config ): ?string { |
| 342 | if ( $this->searchContext->getNamespaces() && $config->has( CirrusConfigNames::ConcreteNamespaceMap ) ) { |
| 343 | // Attempt to skip Connection::pickIndexSuffixForNamespaces() and use the |
| 344 | // concrete namespace map. |
| 345 | // Reason is that the connection is built against the host wiki config but |
| 346 | // the concrete namespace map is likely obtained for the target wiki. |
| 347 | $concreteNamespaceMap = $config->get( CirrusConfigNames::ConcreteNamespaceMap ); |
| 348 | $indices = []; |
| 349 | $inconsistentNamespaces = false; |
| 350 | foreach ( $this->searchContext->getNamespaces() as $ns ) { |
| 351 | if ( !isset( $concreteNamespaceMap[$ns] ) ) { |
| 352 | // Something's odd here we can trust the target wiki config |
| 353 | $inconsistentNamespaces = true; |
| 354 | continue; |
| 355 | } |
| 356 | $indices[$concreteNamespaceMap[$ns]] = $concreteNamespaceMap[$ns]; |
| 357 | } |
| 358 | // the concrete namespace map contains the full index name $basename_$suffix |
| 359 | // if there's only one index requested we target this one |
| 360 | // otherwise we target the main alias. |
| 361 | if ( !$inconsistentNamespaces && count( $indices ) === 1 ) { |
| 362 | $indexBaseName = reset( $indices ); |
| 363 | return $this->connection->getIndexName( $indexBaseName ); |
| 364 | } |
| 365 | } |
| 366 | return null; |
| 367 | } |
| 368 | |
| 369 | /** |
| 370 | * @param ?Index $index |
| 371 | * @return $this |
| 372 | */ |
| 373 | public function setIndex( ?Index $index ): self { |
| 374 | $this->index = $index; |
| 375 | return $this; |
| 376 | } |
| 377 | |
| 378 | /** |
| 379 | * @return string |
| 380 | */ |
| 381 | public function getSort() { |
| 382 | return $this->sort; |
| 383 | } |
| 384 | |
| 385 | /** |
| 386 | * @param string $sort |
| 387 | * @return self |
| 388 | */ |
| 389 | public function setSort( $sort ) { |
| 390 | $this->sort = $sort; |
| 391 | |
| 392 | return $this; |
| 393 | } |
| 394 | |
| 395 | /** |
| 396 | * @return SearchContext |
| 397 | */ |
| 398 | public function getSearchContext() { |
| 399 | return $this->searchContext; |
| 400 | } |
| 401 | } |