Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
45.59% |
93 / 204 |
|
31.25% |
20 / 64 |
CRAP | |
0.00% |
0 / 1 |
| SearchContext | |
45.59% |
93 / 204 |
|
31.25% |
20 / 64 |
1455.50 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| withConfig | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| loadConfig | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| __clone | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isDirty | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getConfig | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNamespaces | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setNamespaces | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getProfileContext | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getProfileContextParams | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setProfileContext | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| getRescoreProfile | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| setRescoreProfile | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| areResultsPossible | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setResultsPossible | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| isSyntaxUsed | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| isSpecialKeywordUsed | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| getSyntaxUsed | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getSyntaxDescription | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addSyntaxUsed | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getSearchType | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| addFilter | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| addNotFilter | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| setHighlightQuery | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addNonTextHighlightQuery | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getFetchPhaseBuilder | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getHighlight | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| getHighlightQuery | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| getRescore | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| getQuery | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
72 | |||
| setMainQuery | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addNonTextQuery | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getSearchQuery | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getLimitSearchToLocalWiki | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setLimitSearchToLocalWiki | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getCacheTtl | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setCacheTtl | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getOriginalSearchTerm | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setOriginalSearchTerm | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getCleanedSearchTerm | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setCleanedSearchTerm | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| escaper | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getExtraScoreBuilders | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addCustomRescoreComponent | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addWarning | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| getWarnings | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFulltextQueryBuilderProfile | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| setFulltextQueryBuilderProfile | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| setResultsType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getResultsType | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getExtraIndices | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| getPhraseRescoreQuery | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setPhraseRescoreQuery | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addAggregation | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getAggregations | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDebugOptions | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFilters | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| must | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| mustNot | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| fromSearchQuery | |
100.00% |
39 / 39 |
|
100.00% |
1 / 1 |
4 | |||
| getFallbackRunner | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| disableFallbackRunner | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setTrackTotalHits | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getTrackTotalHits | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Search; |
| 4 | |
| 5 | use CirrusSearch\CirrusDebugOptions; |
| 6 | use CirrusSearch\CirrusSearchHookRunner; |
| 7 | use CirrusSearch\ExternalIndex; |
| 8 | use CirrusSearch\Fallbacks\FallbackRunner; |
| 9 | use CirrusSearch\OtherIndexesUpdater; |
| 10 | use CirrusSearch\Parser\BasicQueryClassifier; |
| 11 | use CirrusSearch\Profile\SearchProfileService; |
| 12 | use CirrusSearch\Query\Builder\FilterBuilder; |
| 13 | use CirrusSearch\Search\Fetch\FetchPhaseConfigBuilder; |
| 14 | use CirrusSearch\Search\Rescore\BoostFunctionBuilder; |
| 15 | use CirrusSearch\Search\Rescore\RescoreBuilder; |
| 16 | use CirrusSearch\SearchConfig; |
| 17 | use CirrusSearch\WarningCollector; |
| 18 | use Elastica\Aggregation\AbstractAggregation; |
| 19 | use Elastica\Query\AbstractQuery; |
| 20 | use MediaWiki\MainConfigNames; |
| 21 | use MediaWiki\MediaWikiServices; |
| 22 | use Wikimedia\Assert\Assert; |
| 23 | |
| 24 | /** |
| 25 | * The search context, maintains the state of the current search query. |
| 26 | * |
| 27 | * @license GPL-2.0-or-later |
| 28 | */ |
| 29 | |
| 30 | /** |
| 31 | * The SearchContext stores the various states maintained |
| 32 | * during the query building process. |
| 33 | */ |
| 34 | class SearchContext implements WarningCollector, FilterBuilder { |
| 35 | /** |
| 36 | * @var SearchConfig |
| 37 | */ |
| 38 | private $config; |
| 39 | |
| 40 | /** |
| 41 | * @var int[]|null list of namespaces |
| 42 | */ |
| 43 | private $namespaces; |
| 44 | |
| 45 | /** |
| 46 | * @var string |
| 47 | */ |
| 48 | private $profileContext = SearchProfileService::CONTEXT_DEFAULT; |
| 49 | |
| 50 | /** |
| 51 | * @var string[] |
| 52 | */ |
| 53 | private $profileContextParams = []; |
| 54 | |
| 55 | /** |
| 56 | * @var string rescore profile to use |
| 57 | */ |
| 58 | private $rescoreProfile; |
| 59 | |
| 60 | /** |
| 61 | * @var BoostFunctionBuilder[] Extra scoring builders to use. |
| 62 | */ |
| 63 | private $extraScoreBuilders = []; |
| 64 | |
| 65 | /** |
| 66 | * @var bool Could this query possibly return results? |
| 67 | */ |
| 68 | private $resultsPossible = true; |
| 69 | |
| 70 | /** |
| 71 | * @var int[] List of features in the user suplied query string. Features are |
| 72 | * held in the array key, value is how "complex" the feature is. |
| 73 | */ |
| 74 | private $syntaxUsed = []; |
| 75 | |
| 76 | /** |
| 77 | * @var AbstractQuery[] List of filters that query results must match |
| 78 | */ |
| 79 | private $filters = []; |
| 80 | |
| 81 | /** |
| 82 | * @var AbstractQuery[] List of filters that query results must not match |
| 83 | */ |
| 84 | private $notFilters = []; |
| 85 | |
| 86 | /** |
| 87 | * @var AbstractQuery|null Query that should be used for highlighting if different |
| 88 | * from the query used for selecting. |
| 89 | */ |
| 90 | private $highlightQuery; |
| 91 | |
| 92 | /** |
| 93 | * @var AbstractQuery[] queries that don't use Elastic's "query string" query, |
| 94 | * for more advanced highlighting (e.g. match_phrase_prefix for regular |
| 95 | * quoted strings). |
| 96 | */ |
| 97 | private $nonTextHighlightQueries = []; |
| 98 | |
| 99 | /** |
| 100 | * @var AbstractQuery|null phrase rescore query |
| 101 | */ |
| 102 | private $phraseRescoreQuery; |
| 103 | |
| 104 | /** |
| 105 | * @var AbstractQuery|null main query. null defaults to MatchAll |
| 106 | */ |
| 107 | private $mainQuery; |
| 108 | |
| 109 | /** |
| 110 | * @var AbstractQuery[] Queries that don't use Elastic's "query string" query, for |
| 111 | * more advanced searching (e.g. match_phrase_prefix for regular quoted strings). |
| 112 | */ |
| 113 | private $nonTextQueries = []; |
| 114 | |
| 115 | /** |
| 116 | * @var SearchQuery |
| 117 | */ |
| 118 | private $searchQuery; |
| 119 | |
| 120 | /** |
| 121 | * @var bool Should this search limit results to the local wiki? |
| 122 | */ |
| 123 | private $limitSearchToLocalWiki = false; |
| 124 | |
| 125 | /** |
| 126 | * @var int The number of seconds to cache results for |
| 127 | */ |
| 128 | private $cacheTtl = 0; |
| 129 | |
| 130 | /** |
| 131 | * @var string The original search |
| 132 | */ |
| 133 | private $originalSearchTerm; |
| 134 | |
| 135 | /** |
| 136 | * @var string The users search term with keywords removed |
| 137 | */ |
| 138 | private $cleanedSearchTerm; |
| 139 | |
| 140 | /** |
| 141 | * @var Escaper |
| 142 | */ |
| 143 | private $escaper; |
| 144 | |
| 145 | /** |
| 146 | * @var FetchPhaseConfigBuilder |
| 147 | */ |
| 148 | private $fetchPhaseBuilder; |
| 149 | |
| 150 | /** |
| 151 | * @var int[] weights of different syntaxes |
| 152 | */ |
| 153 | private static $syntaxWeights = [ |
| 154 | // Deep category searches can have thousands of terms |
| 155 | 'deepcategory' => 20, |
| 156 | 'full_text' => 10, |
| 157 | 'more_like' => 100, |
| 158 | 'near_match' => 10, |
| 159 | 'prefix' => 2, |
| 160 | // regex is really tough |
| 161 | 'regex' => PHP_INT_MAX, |
| 162 | // depends, filtered semantic is pretty expensive |
| 163 | 'semantic' => 20, |
| 164 | ]; |
| 165 | |
| 166 | /** |
| 167 | * @var array[] Warnings to be passed into StatusValue::warning() |
| 168 | */ |
| 169 | private $warnings = []; |
| 170 | |
| 171 | /** |
| 172 | * @var string name of the fulltext query builder profile |
| 173 | */ |
| 174 | private $fulltextQueryBuilderProfile; |
| 175 | |
| 176 | /** |
| 177 | * @var bool Have custom options that effect the search results been set |
| 178 | * outside the defaults from config? |
| 179 | */ |
| 180 | private $isDirty = false; |
| 181 | |
| 182 | /** |
| 183 | * @var ResultsType|FullTextResultsType Type of the result for the context. |
| 184 | */ |
| 185 | private $resultsType; |
| 186 | |
| 187 | /** |
| 188 | * @var AbstractAggregation[] Aggregations to perform |
| 189 | */ |
| 190 | private $aggs = []; |
| 191 | |
| 192 | /** |
| 193 | * @var CirrusDebugOptions |
| 194 | */ |
| 195 | private $debugOptions; |
| 196 | |
| 197 | /** |
| 198 | * @var FallbackRunner|null |
| 199 | */ |
| 200 | private $fallbackRunner; |
| 201 | /** |
| 202 | * @var CirrusSearchHookRunner|null |
| 203 | */ |
| 204 | private $cirrusSearchHookRunner; |
| 205 | |
| 206 | /** |
| 207 | * @var bool |
| 208 | */ |
| 209 | private $trackTotalHits = true; |
| 210 | |
| 211 | /** |
| 212 | * @param SearchConfig $config |
| 213 | * @param int[]|null $namespaces |
| 214 | * @param CirrusDebugOptions|null $options |
| 215 | * @param FallbackRunner|null $fallbackRunner |
| 216 | * @param FetchPhaseConfigBuilder|null $fetchPhaseConfigBuilder |
| 217 | * @param CirrusSearchHookRunner|null $cirrusSearchHookRunner |
| 218 | */ |
| 219 | public function __construct( |
| 220 | SearchConfig $config, |
| 221 | ?array $namespaces = null, |
| 222 | ?CirrusDebugOptions $options = null, |
| 223 | ?FallbackRunner $fallbackRunner = null, |
| 224 | ?FetchPhaseConfigBuilder $fetchPhaseConfigBuilder = null, |
| 225 | ?CirrusSearchHookRunner $cirrusSearchHookRunner = null |
| 226 | ) { |
| 227 | $this->config = $config; |
| 228 | $this->namespaces = $namespaces; |
| 229 | $this->debugOptions = $options ?? CirrusDebugOptions::defaultOptions(); |
| 230 | $this->fallbackRunner = $fallbackRunner ?? FallbackRunner::noopRunner(); |
| 231 | $this->fetchPhaseBuilder = $fetchPhaseConfigBuilder ?? new FetchPhaseConfigBuilder( $config ); |
| 232 | $this->loadConfig(); |
| 233 | $this->cirrusSearchHookRunner = $cirrusSearchHookRunner ?? new CirrusSearchHookRunner( |
| 234 | MediaWikiServices::getInstance()->getHookContainer() ); |
| 235 | } |
| 236 | |
| 237 | /** |
| 238 | * Return a copy of this context with a new configuration. |
| 239 | * |
| 240 | * @param SearchConfig $config The new configuration |
| 241 | * @return self |
| 242 | */ |
| 243 | public function withConfig( SearchConfig $config ) { |
| 244 | $other = clone $this; |
| 245 | $other->config = $config; |
| 246 | $other->fetchPhaseBuilder = $this->fetchPhaseBuilder->withConfig( $config ); |
| 247 | if ( $other->resultsType instanceof FullTextResultsType ) { |
| 248 | $other->resultsType = $this->resultsType->withFetchPhaseBuilder( $other->fetchPhaseBuilder ); |
| 249 | } |
| 250 | $other->loadConfig(); |
| 251 | |
| 252 | return $other; |
| 253 | } |
| 254 | |
| 255 | private function loadConfig(): void { |
| 256 | $this->escaper = new Escaper( |
| 257 | $this->config->get( MainConfigNames::LanguageCode ), |
| 258 | $this->config->get( 'CirrusSearchAllowLeadingWildcard' ) |
| 259 | ); |
| 260 | } |
| 261 | |
| 262 | public function __clone() { |
| 263 | if ( $this->mainQuery ) { |
| 264 | $this->mainQuery = clone $this->mainQuery; |
| 265 | } |
| 266 | } |
| 267 | |
| 268 | /** |
| 269 | * Have custom options that effect the search results been set outside the |
| 270 | * defaults from config? |
| 271 | * |
| 272 | * @return bool |
| 273 | */ |
| 274 | public function isDirty() { |
| 275 | return $this->isDirty; |
| 276 | } |
| 277 | |
| 278 | /** |
| 279 | * @return SearchConfig the Cirrus config object |
| 280 | */ |
| 281 | public function getConfig() { |
| 282 | return $this->config; |
| 283 | } |
| 284 | |
| 285 | /** |
| 286 | * mediawiki namespace id's being requested. |
| 287 | * NOTE: this value may change during the Searcher process. |
| 288 | * |
| 289 | * @return int[]|null |
| 290 | */ |
| 291 | public function getNamespaces() { |
| 292 | return $this->namespaces; |
| 293 | } |
| 294 | |
| 295 | /** |
| 296 | * set the mediawiki namespace id's |
| 297 | * |
| 298 | * @param int[]|null $namespaces array of integer |
| 299 | */ |
| 300 | public function setNamespaces( $namespaces ) { |
| 301 | $this->isDirty = true; |
| 302 | $this->namespaces = $namespaces; |
| 303 | } |
| 304 | |
| 305 | /** |
| 306 | * @return string |
| 307 | */ |
| 308 | public function getProfileContext() { |
| 309 | return $this->profileContext; |
| 310 | } |
| 311 | |
| 312 | /** |
| 313 | * @return string[] |
| 314 | */ |
| 315 | public function getProfileContextParams(): array { |
| 316 | return $this->profileContextParams; |
| 317 | } |
| 318 | |
| 319 | /** |
| 320 | * @param string $profileContext |
| 321 | * @param string[] $contextParams |
| 322 | */ |
| 323 | public function setProfileContext( $profileContext, array $contextParams = [] ) { |
| 324 | $this->isDirty = $this->isDirty || |
| 325 | $this->profileContext !== $profileContext || |
| 326 | $this->profileContextParams !== $contextParams; |
| 327 | $this->profileContext = $profileContext; |
| 328 | $this->profileContextParams = $contextParams; |
| 329 | } |
| 330 | |
| 331 | /** |
| 332 | * @return string the rescore profile to use |
| 333 | */ |
| 334 | public function getRescoreProfile() { |
| 335 | if ( $this->rescoreProfile === null ) { |
| 336 | $this->rescoreProfile = $this->config->getProfileService() |
| 337 | ->getProfileName( SearchProfileService::RESCORE, $this->profileContext, $this->profileContextParams ); |
| 338 | } |
| 339 | return $this->rescoreProfile; |
| 340 | } |
| 341 | |
| 342 | /** |
| 343 | * @param string $rescoreProfile the rescore profile to use |
| 344 | */ |
| 345 | public function setRescoreProfile( $rescoreProfile ) { |
| 346 | $this->isDirty = true; |
| 347 | $this->rescoreProfile = $rescoreProfile; |
| 348 | } |
| 349 | |
| 350 | /** |
| 351 | * @return bool Could this query possibly return results? |
| 352 | */ |
| 353 | public function areResultsPossible() { |
| 354 | return $this->resultsPossible; |
| 355 | } |
| 356 | |
| 357 | /** |
| 358 | * @param bool $possible Could this query possible return results? Defaults to true |
| 359 | * if not called. |
| 360 | */ |
| 361 | public function setResultsPossible( $possible ) { |
| 362 | $this->isDirty = true; |
| 363 | $this->resultsPossible = $possible; |
| 364 | } |
| 365 | |
| 366 | /** |
| 367 | * @param string|null $type type of syntax to check, null for any type |
| 368 | * @return bool True when the query uses $type kind of syntax |
| 369 | */ |
| 370 | public function isSyntaxUsed( $type = null ) { |
| 371 | if ( $type === null ) { |
| 372 | return $this->syntaxUsed !== []; |
| 373 | } |
| 374 | return isset( $this->syntaxUsed[$type] ); |
| 375 | } |
| 376 | |
| 377 | /** |
| 378 | * @return bool true if a special keyword or syntax was used in the query |
| 379 | */ |
| 380 | public function isSpecialKeywordUsed() { |
| 381 | // full_text is not considered a special keyword |
| 382 | // TODO: investigate using BasicQueryClassifier::SIMPLE_BAG_OF_WORDS instead |
| 383 | return array_diff_key( $this->syntaxUsed, [ |
| 384 | 'full_text' => true, |
| 385 | 'full_text_simple_match' => true, |
| 386 | 'full_text_querystring' => true, |
| 387 | BasicQueryClassifier::SIMPLE_BAG_OF_WORDS => true, |
| 388 | BasicQueryClassifier::SIMPLE_PHRASE => true, |
| 389 | BasicQueryClassifier::BAG_OF_WORDS_WITH_PHRASE => true, |
| 390 | ] ) !== []; |
| 391 | } |
| 392 | |
| 393 | /** |
| 394 | * @return string[] List of syntax used in the query |
| 395 | */ |
| 396 | public function getSyntaxUsed() { |
| 397 | return array_keys( $this->syntaxUsed ); |
| 398 | } |
| 399 | |
| 400 | /** |
| 401 | * @return string Text description of syntax used by query. |
| 402 | */ |
| 403 | public function getSyntaxDescription() { |
| 404 | return implode( ',', $this->getSyntaxUsed() ); |
| 405 | } |
| 406 | |
| 407 | /** |
| 408 | * @param string $feature Name of a syntax feature used in the query string |
| 409 | * @param int|null $weight How "complex" is this feature. |
| 410 | */ |
| 411 | public function addSyntaxUsed( $feature, $weight = null ) { |
| 412 | $this->isDirty = true; |
| 413 | $this->syntaxUsed[$feature] = $weight ?? self::$syntaxWeights[$feature] ?? 1; |
| 414 | } |
| 415 | |
| 416 | /** |
| 417 | * @return string The type of search being performed, ex: full_text, near_match, prefix, etc. |
| 418 | * Using getSyntaxUsed() is better in most cases. |
| 419 | */ |
| 420 | public function getSearchType() { |
| 421 | if ( !$this->syntaxUsed ) { |
| 422 | return 'full_text'; |
| 423 | } |
| 424 | arsort( $this->syntaxUsed ); |
| 425 | // Return the first heaviest syntax |
| 426 | return key( $this->syntaxUsed ); |
| 427 | } |
| 428 | |
| 429 | /** |
| 430 | * @param AbstractQuery $filter Query results must match this filter |
| 431 | */ |
| 432 | public function addFilter( AbstractQuery $filter ) { |
| 433 | $this->isDirty = true; |
| 434 | $this->filters[] = $filter; |
| 435 | } |
| 436 | |
| 437 | /** |
| 438 | * @param AbstractQuery $filter Query results must not match this filter |
| 439 | */ |
| 440 | public function addNotFilter( AbstractQuery $filter ) { |
| 441 | $this->isDirty = true; |
| 442 | $this->notFilters[] = $filter; |
| 443 | } |
| 444 | |
| 445 | /** |
| 446 | * @param AbstractQuery|null $query Query that should be used for highlighting if different |
| 447 | * from the query used for selecting. |
| 448 | */ |
| 449 | public function setHighlightQuery( ?AbstractQuery $query = null ) { |
| 450 | $this->isDirty = true; |
| 451 | $this->highlightQuery = $query; |
| 452 | } |
| 453 | |
| 454 | /** |
| 455 | * @param AbstractQuery $query queries that don't use Elastic's "query |
| 456 | * string" query, for more advanced highlighting (e.g. match_phrase_prefix |
| 457 | * for regular quoted strings). |
| 458 | */ |
| 459 | public function addNonTextHighlightQuery( AbstractQuery $query ) { |
| 460 | $this->isDirty = true; |
| 461 | $this->nonTextHighlightQueries[] = $query; |
| 462 | } |
| 463 | |
| 464 | public function getFetchPhaseBuilder(): FetchPhaseConfigBuilder { |
| 465 | return $this->fetchPhaseBuilder; |
| 466 | } |
| 467 | |
| 468 | /** |
| 469 | * @param ResultsType $resultsType |
| 470 | * @param AbstractQuery $mainQuery Will be combined with highlighting query |
| 471 | * to provide highlightable terms. |
| 472 | * @return array|null Fetch portion of query to be sent to elasticsearch |
| 473 | */ |
| 474 | public function getHighlight( ResultsType $resultsType, AbstractQuery $mainQuery ) { |
| 475 | $highlight = $resultsType->getHighlightingConfiguration( [] ); |
| 476 | if ( !$highlight ) { |
| 477 | return null; |
| 478 | } |
| 479 | |
| 480 | $query = $this->getHighlightQuery( $mainQuery ); |
| 481 | if ( $query ) { |
| 482 | $highlight['highlight_query'] = $query->toArray(); |
| 483 | } |
| 484 | |
| 485 | return $highlight; |
| 486 | } |
| 487 | |
| 488 | /** |
| 489 | * @param AbstractQuery $mainQuery |
| 490 | * @return AbstractQuery|null Query that should be used for highlighting if different |
| 491 | * from the query used for selecting. |
| 492 | */ |
| 493 | private function getHighlightQuery( AbstractQuery $mainQuery ) { |
| 494 | if ( !$this->nonTextHighlightQueries ) { |
| 495 | // If no explicit highlight query is provided elastic |
| 496 | // will fallback to $mainQuery without specifying it. |
| 497 | return $this->highlightQuery; |
| 498 | } |
| 499 | |
| 500 | $bool = new \Elastica\Query\BoolQuery(); |
| 501 | // If no explicit highlight query is provided we |
| 502 | // need to include the main query along with |
| 503 | // the non-text queries to highlight those fields. |
| 504 | $bool->addShould( $this->highlightQuery ?: $mainQuery ); |
| 505 | foreach ( $this->nonTextHighlightQueries as $nonTextHighlightQuery ) { |
| 506 | $bool->addShould( $nonTextHighlightQuery ); |
| 507 | } |
| 508 | |
| 509 | return $bool; |
| 510 | } |
| 511 | |
| 512 | /** |
| 513 | * rescore_query has to be in array form before we send it to Elasticsearch but it is way |
| 514 | * easier to work with if we leave it in query form until now |
| 515 | * |
| 516 | * @return array[] Rescore configurations as used by elasticsearch. |
| 517 | */ |
| 518 | public function getRescore() { |
| 519 | $rescores = ( new RescoreBuilder( $this, $this->cirrusSearchHookRunner ) )->build(); |
| 520 | $result = []; |
| 521 | foreach ( $rescores as $rescore ) { |
| 522 | $rescore['query']['rescore_query'] = $rescore['query']['rescore_query']->toArray(); |
| 523 | $result[] = $rescore; |
| 524 | } |
| 525 | |
| 526 | return $result; |
| 527 | } |
| 528 | |
| 529 | /** |
| 530 | * @return AbstractQuery The primary query to be sent to elasticsearch. Includes |
| 531 | * the main quedry, non text queries, and any additional filters. |
| 532 | */ |
| 533 | public function getQuery() { |
| 534 | if ( !$this->nonTextQueries ) { |
| 535 | $mainQuery = $this->mainQuery ?: new \Elastica\Query\MatchAll(); |
| 536 | } else { |
| 537 | $mainQuery = new \Elastica\Query\BoolQuery(); |
| 538 | if ( $this->mainQuery ) { |
| 539 | $mainQuery->addMust( $this->mainQuery ); |
| 540 | } |
| 541 | foreach ( $this->nonTextQueries as $nonTextQuery ) { |
| 542 | $mainQuery->addMust( $nonTextQuery ); |
| 543 | } |
| 544 | } |
| 545 | $filters = $this->filters; |
| 546 | if ( $this->getNamespaces() ) { |
| 547 | // We must take an array_values here, or it can be json-encoded into an object instead |
| 548 | // of a list which elasticsearch will interpret as terms lookup. |
| 549 | $filters[] = new \Elastica\Query\Terms( 'namespace', array_values( $this->getNamespaces() ) ); |
| 550 | } |
| 551 | |
| 552 | // Wrap $mainQuery in a filtered query if there are any filters |
| 553 | $unifiedFilter = Filters::unify( $filters, $this->notFilters ); |
| 554 | if ( $unifiedFilter !== null ) { |
| 555 | if ( !( $mainQuery instanceof \Elastica\Query\BoolQuery ) ) { |
| 556 | $bool = new \Elastica\Query\BoolQuery(); |
| 557 | $bool->addMust( $mainQuery ); |
| 558 | $mainQuery = $bool; |
| 559 | } |
| 560 | $mainQuery->addFilter( $unifiedFilter ); |
| 561 | } |
| 562 | |
| 563 | return $mainQuery; |
| 564 | } |
| 565 | |
| 566 | /** |
| 567 | * @param AbstractQuery $query The primary query to be passed to |
| 568 | * elasticsearch. |
| 569 | */ |
| 570 | public function setMainQuery( AbstractQuery $query ) { |
| 571 | $this->isDirty = true; |
| 572 | $this->mainQuery = $query; |
| 573 | } |
| 574 | |
| 575 | /** |
| 576 | * @param \Elastica\Query\AbstractQuery $match Queries that don't use Elastic's |
| 577 | * "query string" query, for more advanced searching (e.g. |
| 578 | * match_phrase_prefix for regular quoted strings). |
| 579 | */ |
| 580 | public function addNonTextQuery( \Elastica\Query\AbstractQuery $match ) { |
| 581 | $this->isDirty = true; |
| 582 | $this->nonTextQueries[] = $match; |
| 583 | } |
| 584 | |
| 585 | /** |
| 586 | * @return SearchQuery |
| 587 | */ |
| 588 | public function getSearchQuery() { |
| 589 | return $this->searchQuery; |
| 590 | } |
| 591 | |
| 592 | /** |
| 593 | * @return bool Should this search limit results to the local wiki? If |
| 594 | * not called the default is false. |
| 595 | */ |
| 596 | public function getLimitSearchToLocalWiki() { |
| 597 | return $this->limitSearchToLocalWiki; |
| 598 | } |
| 599 | |
| 600 | /** |
| 601 | * @param bool $localWikiOnly Should this search limit results to the local wiki? If |
| 602 | * not called the default is false. |
| 603 | */ |
| 604 | public function setLimitSearchToLocalWiki( $localWikiOnly ) { |
| 605 | if ( $localWikiOnly !== $this->limitSearchToLocalWiki ) { |
| 606 | $this->isDirty = true; |
| 607 | $this->limitSearchToLocalWiki = $localWikiOnly; |
| 608 | } |
| 609 | } |
| 610 | |
| 611 | /** |
| 612 | * @return int The number of seconds to cache results for |
| 613 | */ |
| 614 | public function getCacheTtl() { |
| 615 | return $this->cacheTtl; |
| 616 | } |
| 617 | |
| 618 | /** |
| 619 | * @param int $ttl The number of seconds to cache results for |
| 620 | */ |
| 621 | public function setCacheTtl( $ttl ) { |
| 622 | $this->isDirty = true; |
| 623 | $this->cacheTtl = $ttl; |
| 624 | } |
| 625 | |
| 626 | /** |
| 627 | * @return string the original search term |
| 628 | */ |
| 629 | public function getOriginalSearchTerm() { |
| 630 | return $this->originalSearchTerm; |
| 631 | } |
| 632 | |
| 633 | /** |
| 634 | * @param string $term |
| 635 | */ |
| 636 | public function setOriginalSearchTerm( $term ) { |
| 637 | // Intentionally does not set dirty to true. This is used only |
| 638 | // for logging, as of july 2017. |
| 639 | $this->originalSearchTerm = $term; |
| 640 | } |
| 641 | |
| 642 | /** |
| 643 | * @return string The search term with keywords removed |
| 644 | */ |
| 645 | public function getCleanedSearchTerm() { |
| 646 | return $this->cleanedSearchTerm; |
| 647 | } |
| 648 | |
| 649 | /** |
| 650 | * @param string $term The search term with keywords removed |
| 651 | */ |
| 652 | public function setCleanedSearchTerm( $term ) { |
| 653 | $this->isDirty = true; |
| 654 | $this->cleanedSearchTerm = $term; |
| 655 | } |
| 656 | |
| 657 | /** |
| 658 | * @return Escaper |
| 659 | */ |
| 660 | public function escaper() { |
| 661 | return $this->escaper; |
| 662 | } |
| 663 | |
| 664 | /** |
| 665 | * @return BoostFunctionBuilder[] |
| 666 | */ |
| 667 | public function getExtraScoreBuilders() { |
| 668 | return $this->extraScoreBuilders; |
| 669 | } |
| 670 | |
| 671 | /** |
| 672 | * Add custom scoring function to the context. |
| 673 | * The rescore builder will pick it up. |
| 674 | */ |
| 675 | public function addCustomRescoreComponent( BoostFunctionBuilder $rescore ) { |
| 676 | $this->isDirty = true; |
| 677 | $this->extraScoreBuilders[] = $rescore; |
| 678 | } |
| 679 | |
| 680 | /** |
| 681 | * @param string $message i18n message key |
| 682 | * @param mixed ...$params |
| 683 | */ |
| 684 | public function addWarning( $message, ...$params ) { |
| 685 | $this->isDirty = true; |
| 686 | $this->warnings[] = [ $message, ...array_filter( $params, static function ( $v ) { |
| 687 | return $v !== null; |
| 688 | } ) ]; |
| 689 | } |
| 690 | |
| 691 | /** |
| 692 | * @return array[] Array of arrays. Each sub array is a set of values |
| 693 | * suitable for creating an i18n message. |
| 694 | * @phan-return non-empty-array[] |
| 695 | */ |
| 696 | public function getWarnings() { |
| 697 | return $this->warnings; |
| 698 | } |
| 699 | |
| 700 | /** |
| 701 | * @return string the name of the fulltext query builder profile |
| 702 | */ |
| 703 | public function getFulltextQueryBuilderProfile() { |
| 704 | if ( $this->fulltextQueryBuilderProfile === null ) { |
| 705 | $this->fulltextQueryBuilderProfile = $this->config->getProfileService() |
| 706 | ->getProfileName( SearchProfileService::FT_QUERY_BUILDER, $this->profileContext ); |
| 707 | } |
| 708 | return $this->fulltextQueryBuilderProfile; |
| 709 | } |
| 710 | |
| 711 | /** |
| 712 | * @param string $profile set the name of the fulltext query builder profile |
| 713 | */ |
| 714 | public function setFulltextQueryBuilderProfile( $profile ) { |
| 715 | $this->isDirty = true; |
| 716 | $this->fulltextQueryBuilderProfile = $profile; |
| 717 | } |
| 718 | |
| 719 | /** |
| 720 | * @param ResultsType $resultsType results type to return |
| 721 | */ |
| 722 | public function setResultsType( $resultsType ) { |
| 723 | $this->resultsType = $resultsType; |
| 724 | } |
| 725 | |
| 726 | /** |
| 727 | * @return ResultsType |
| 728 | */ |
| 729 | public function getResultsType() { |
| 730 | Assert::precondition( $this->resultsType !== null, "resultsType unset" ); |
| 731 | return $this->resultsType; |
| 732 | } |
| 733 | |
| 734 | /** |
| 735 | * Get the list of extra indices to query. |
| 736 | * Generally needed to query externilized file index. |
| 737 | * Must be called only once the list of namespaces has been set. |
| 738 | * |
| 739 | * @return ExternalIndex[] |
| 740 | * @see OtherIndexesUpdater::getExtraIndexesForNamespaces() |
| 741 | */ |
| 742 | public function getExtraIndices() { |
| 743 | if ( $this->getLimitSearchToLocalWiki() || !$this->getNamespaces() ) { |
| 744 | return []; |
| 745 | } |
| 746 | return OtherIndexesUpdater::getExtraIndexesForNamespaces( |
| 747 | $this->config, |
| 748 | $this->getNamespaces() |
| 749 | ); |
| 750 | } |
| 751 | |
| 752 | /** |
| 753 | * Get the phrase rescore query if available |
| 754 | * @return AbstractQuery|null |
| 755 | */ |
| 756 | public function getPhraseRescoreQuery() { |
| 757 | return $this->phraseRescoreQuery; |
| 758 | } |
| 759 | |
| 760 | /** |
| 761 | * @param AbstractQuery|null $phraseRescoreQuery |
| 762 | */ |
| 763 | public function setPhraseRescoreQuery( $phraseRescoreQuery ) { |
| 764 | $this->phraseRescoreQuery = $phraseRescoreQuery; |
| 765 | $this->isDirty = true; |
| 766 | } |
| 767 | |
| 768 | /** |
| 769 | * Add aggregation to perform on search. |
| 770 | */ |
| 771 | public function addAggregation( AbstractAggregation $agg ) { |
| 772 | $this->aggs[] = $agg; |
| 773 | $this->isDirty = true; |
| 774 | } |
| 775 | |
| 776 | /** |
| 777 | * Get the list of aggregations. |
| 778 | * @return AbstractAggregation[] |
| 779 | */ |
| 780 | public function getAggregations() { |
| 781 | return $this->aggs; |
| 782 | } |
| 783 | |
| 784 | /** |
| 785 | * @return CirrusDebugOptions |
| 786 | */ |
| 787 | public function getDebugOptions() { |
| 788 | return $this->debugOptions; |
| 789 | } |
| 790 | |
| 791 | /** |
| 792 | * NOTE: public for testing purposes. |
| 793 | * @return AbstractQuery[] |
| 794 | */ |
| 795 | public function getFilters(): array { |
| 796 | return $this->filters; |
| 797 | } |
| 798 | |
| 799 | public function must( AbstractQuery $query ) { |
| 800 | $this->addFilter( $query ); |
| 801 | } |
| 802 | |
| 803 | public function mustNot( AbstractQuery $query ) { |
| 804 | $this->addNotFilter( $query ); |
| 805 | } |
| 806 | |
| 807 | /** |
| 808 | * Builds a SearchContext based on a SearchQuery. |
| 809 | * |
| 810 | * Helper function used for building blocks that still work on top |
| 811 | * of the SearchContext+queryString instead of SearchQuery. |
| 812 | * |
| 813 | * States initialized: |
| 814 | * - limitSearchToLocalWiki |
| 815 | * - suggestion |
| 816 | * - custom rescoreProfile/fulltextQueryBuilderProfile |
| 817 | * - contextual filters: (eg. SearchEngine::$prefix) |
| 818 | * - SuggestPrefix (DYM prefix: ~ and/or namespace header) |
| 819 | * |
| 820 | * @param SearchQuery $query |
| 821 | * @param FallbackRunner|null $fallbackRunner |
| 822 | * @param CirrusSearchHookRunner|null $cirrusSearchHookRunner |
| 823 | * @return self |
| 824 | * @throws \CirrusSearch\Parser\ParsedQueryClassifierException |
| 825 | */ |
| 826 | public static function fromSearchQuery( |
| 827 | SearchQuery $query, |
| 828 | ?FallbackRunner $fallbackRunner = null, |
| 829 | ?CirrusSearchHookRunner $cirrusSearchHookRunner = null |
| 830 | ): self { |
| 831 | $searchContext = new self( |
| 832 | $query->getSearchConfig(), |
| 833 | $query->getNamespaces(), |
| 834 | $query->getDebugOptions(), |
| 835 | $fallbackRunner, |
| 836 | new FetchPhaseConfigBuilder( |
| 837 | $query->getSearchConfig(), |
| 838 | $query->getSearchEngineEntryPoint(), |
| 839 | $query->shouldProvideAllSnippets() |
| 840 | ), |
| 841 | $cirrusSearchHookRunner |
| 842 | ); |
| 843 | |
| 844 | $searchContext->searchQuery = $query; |
| 845 | |
| 846 | $searchContext->limitSearchToLocalWiki = !$query->getCrossSearchStrategy()->isExtraIndicesSearchSupported(); |
| 847 | |
| 848 | $searchContext->rescoreProfile = $query->getForcedProfile( SearchProfileService::RESCORE ); |
| 849 | |
| 850 | $profileContext = $query->getSearchConfig() |
| 851 | ->getProfileService() |
| 852 | ->getDispatchService() |
| 853 | ->bestRoute( $query ) |
| 854 | ->getProfileContext(); |
| 855 | $searchContext->setProfileContext( $profileContext ); |
| 856 | $parsedQuery = $query->getParsedQuery(); |
| 857 | $basicQueryClasses = [ |
| 858 | BasicQueryClassifier::SIMPLE_BAG_OF_WORDS, |
| 859 | BasicQueryClassifier::SIMPLE_PHRASE, |
| 860 | BasicQueryClassifier::BAG_OF_WORDS_WITH_PHRASE, |
| 861 | BasicQueryClassifier::COMPLEX_QUERY, |
| 862 | ]; |
| 863 | |
| 864 | foreach ( $basicQueryClasses as $klass ) { |
| 865 | if ( $parsedQuery->isQueryOfClass( $klass ) ) { |
| 866 | $searchContext->syntaxUsed[$klass] = 1; |
| 867 | } |
| 868 | } |
| 869 | // TODO: Clarify what happens when user forces a profile, should we disable the dispatch service? |
| 870 | $searchContext->fulltextQueryBuilderProfile = $query->getForcedProfile( SearchProfileService::FT_QUERY_BUILDER ); |
| 871 | $searchContext->profileContextParams = $query->getProfileContextParameters(); |
| 872 | |
| 873 | foreach ( $query->getContextualFilters() as $filter ) { |
| 874 | $filter->populate( $searchContext ); |
| 875 | } |
| 876 | $pQuery = $query->getParsedQuery(); |
| 877 | $searchContext->originalSearchTerm = $pQuery->getRawQuery(); |
| 878 | $searchContext->trackTotalHits = $query->mustTrackTotalHits(); |
| 879 | return $searchContext; |
| 880 | } |
| 881 | |
| 882 | public function getFallbackRunner(): FallbackRunner { |
| 883 | return $this->fallbackRunner; |
| 884 | } |
| 885 | |
| 886 | public function disableFallbackRunner(): void { |
| 887 | $this->fallbackRunner = FallbackRunner::noopRunner(); |
| 888 | } |
| 889 | |
| 890 | public function setTrackTotalHits( bool $trackTotalHits ): void { |
| 891 | if ( $trackTotalHits !== $this->trackTotalHits ) { |
| 892 | $this->isDirty = true; |
| 893 | $this->trackTotalHits = $trackTotalHits; |
| 894 | } |
| 895 | } |
| 896 | |
| 897 | public function getTrackTotalHits(): bool { |
| 898 | return $this->trackTotalHits; |
| 899 | } |
| 900 | } |