Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
72.11% |
181 / 251 |
|
62.50% |
10 / 16 |
CRAP | |
0.00% |
0 / 1 |
| FullTextQueryStringQueryBuilder | |
72.11% |
181 / 251 |
|
62.50% |
10 / 16 |
110.65 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| build | |
76.67% |
92 / 120 |
|
0.00% |
0 / 1 |
22.12 | |||
| isPathologicalWildcard | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| buildDegraded | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
6 | |||
| buildSearchTextQuery | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
2 | |||
| buildQueryString | |
90.91% |
10 / 11 |
|
0.00% |
0 / 1 |
2.00 | |||
| getMultiTermRewriteMethod | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| switchSearchToExactForWildcards | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| switchSearchToExact | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| buildFullTextSearchFields | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
4 | |||
| replaceAllPartsOfQuery | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| replacePartsOfQuery | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
5 | |||
| buildHighlightQuery | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| buildPhraseRescoreQuery | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| isPhraseRescoreNeeded | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
5 | |||
| maybeWrapWithTokenCountRouter | |
10.00% |
2 / 20 |
|
0.00% |
0 / 1 |
9.56 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Query; |
| 4 | |
| 5 | use CirrusSearch\Extra\Query\TokenCountRouter; |
| 6 | use CirrusSearch\Query\Builder\NearMatchFieldQueryBuilder; |
| 7 | use CirrusSearch\Search\SearchContext; |
| 8 | use CirrusSearch\SearchConfig; |
| 9 | use Elastica\Query\AbstractQuery; |
| 10 | use Elastica\Query\MatchAll; |
| 11 | use Elastica\Query\MatchNone; |
| 12 | use MediaWiki\Logger\LoggerFactory; |
| 13 | |
| 14 | /** |
| 15 | * Builds an Elastica query backed by an elasticsearch QueryString query |
| 16 | * Has many warts and edge cases that are hardly desirable. |
| 17 | */ |
| 18 | class FullTextQueryStringQueryBuilder implements FullTextQueryBuilder { |
| 19 | /** |
| 20 | * @var SearchConfig |
| 21 | */ |
| 22 | protected $config; |
| 23 | |
| 24 | /** |
| 25 | * @var KeywordFeature[] |
| 26 | */ |
| 27 | private $features; |
| 28 | |
| 29 | /** |
| 30 | * @var string |
| 31 | */ |
| 32 | private $queryStringQueryString = ''; |
| 33 | |
| 34 | /** |
| 35 | * @var bool |
| 36 | */ |
| 37 | private $useTokenCountRouter; |
| 38 | |
| 39 | private NearMatchFieldQueryBuilder $nearMatchFieldQueryBuilder; |
| 40 | |
| 41 | /** |
| 42 | * @param SearchConfig $config |
| 43 | * @param KeywordFeature[] $features |
| 44 | * @param array[] $settings currently ignored |
| 45 | */ |
| 46 | public function __construct( SearchConfig $config, array $features, array $settings = [] ) { |
| 47 | $this->config = $config; |
| 48 | $this->features = $features; |
| 49 | $this->useTokenCountRouter = $this->config->getElement( |
| 50 | 'CirrusSearchWikimediaExtraPlugin', 'token_count_router' ) === true; |
| 51 | $this->nearMatchFieldQueryBuilder = NearMatchFieldQueryBuilder::defaultFromSearchConfig( $config ); |
| 52 | } |
| 53 | |
| 54 | /** |
| 55 | * Search articles with provided term. |
| 56 | * |
| 57 | * @param SearchContext $searchContext |
| 58 | * @param string $term term to search |
| 59 | * searches that might be better? |
| 60 | */ |
| 61 | public function build( SearchContext $searchContext, $term ) { |
| 62 | $searchContext->addSyntaxUsed( 'full_text' ); |
| 63 | // Transform MediaWiki specific syntax to filters and extra |
| 64 | // (pre-escaped) query string |
| 65 | foreach ( $this->features as $feature ) { |
| 66 | $term = $feature->apply( $searchContext, $term ); |
| 67 | } |
| 68 | |
| 69 | if ( !$searchContext->areResultsPossible() ) { |
| 70 | return; |
| 71 | } |
| 72 | |
| 73 | $term = $searchContext->escaper()->escapeQuotes( $term ); |
| 74 | $term = trim( $term ); |
| 75 | |
| 76 | // Match quoted phrases including those containing escaped quotes. |
| 77 | // Those phrases can optionally be followed by ~ then a number (this is |
| 78 | // the phrase slop). That can optionally be followed by a ~ (this |
| 79 | // matches stemmed words in phrases). The following all match: |
| 80 | // "a", "a boat", "a\"boat", "a boat"~, "a boat"~9, |
| 81 | // "a boat"~9~, -"a boat", -"a boat"~9~ |
| 82 | $slop = $this->config->get( 'CirrusSearchPhraseSlop' ); |
| 83 | $matchQuotesRegex = '(?<![\]])(?<negate>-|!)?(?<main>"((?:[^"]|(?<=\\\)")+)"(?<slop>~\d+)?)(?<fuzzy>~)?'; |
| 84 | $query = self::replacePartsOfQuery( |
| 85 | $term, |
| 86 | "/$matchQuotesRegex/", |
| 87 | function ( $matches ) use ( $searchContext, $slop ) { |
| 88 | $negate = $matches[ 'negate' ][ 0 ] ? 'NOT ' : ''; |
| 89 | $main = $searchContext->escaper()->fixupQueryStringPart( $matches[ 'main' ][ 0 ] ); |
| 90 | |
| 91 | if ( !$negate && !isset( $matches[ 'fuzzy' ] ) && !isset( $matches[ 'slop' ] ) && |
| 92 | preg_match( '/^"([^"*]+)[*]"/', $main, $matches ) |
| 93 | ) { |
| 94 | $phraseMatch = new \Elastica\Query\MatchPhrasePrefix(); |
| 95 | $phraseMatch->setFieldQuery( "all.plain", $matches[1] ); |
| 96 | $searchContext->addNonTextQuery( $phraseMatch ); |
| 97 | $searchContext->addSyntaxUsed( 'phrase_match_prefix' ); |
| 98 | |
| 99 | $phraseHighlightMatch = new \Elastica\Query\QueryString(); |
| 100 | $phraseHighlightMatch->setQuery( $matches[1] . '*' ); |
| 101 | $phraseHighlightMatch->setFields( [ 'all.plain' ] ); |
| 102 | $searchContext->addNonTextHighlightQuery( $phraseHighlightMatch ); |
| 103 | |
| 104 | return []; |
| 105 | } |
| 106 | |
| 107 | if ( !isset( $matches[ 'fuzzy' ] ) ) { |
| 108 | if ( !isset( $matches[ 'slop' ] ) ) { |
| 109 | $main .= '~' . $slop[ 'precise' ]; |
| 110 | } |
| 111 | // Got to collect phrases that don't use the all field so we can highlight them. |
| 112 | // The highlighter locks phrases to the fields that specify them. It doesn't do |
| 113 | // that with terms. |
| 114 | return [ |
| 115 | 'escaped' => $negate . self::switchSearchToExact( $searchContext, $main, true ), |
| 116 | 'nonAll' => $negate . self::switchSearchToExact( $searchContext, $main, false ), |
| 117 | ]; |
| 118 | } |
| 119 | return [ 'escaped' => $negate . $main ]; |
| 120 | } ); |
| 121 | // Find prefix matches and force them to only match against the plain analyzed fields. This |
| 122 | // prevents prefix matches from getting confused by stemming. Users really don't expect stemming |
| 123 | // in prefix queries. |
| 124 | $maxWildcards = $this->config->get( 'CirrusSearchQueryStringMaxWildcards' ); |
| 125 | $query = self::replaceAllPartsOfQuery( $query, '/\w+\*(?:\w*\*?)*/u', |
| 126 | function ( $matches ) use ( $searchContext, $maxWildcards ) { |
| 127 | // hack to detect pathological wildcard |
| 128 | // relates to T102589 but elastic7 seems to have broken our fix by stopping |
| 129 | // to propagate the max_determinized_states param to the wildcard queries |
| 130 | // We might consider fixing this upstream again when switch to opensearch. |
| 131 | // In the meantine simply count the number of wildcard chars and mimic the previous |
| 132 | // if we detect such problematic queries |
| 133 | if ( self::isPathologicalWildcard( $matches[ 0 ][ 0 ], $maxWildcards ) ) { |
| 134 | $searchContext->addWarning( 'cirrussearch-regex-too-complex-error' ); |
| 135 | $searchContext->setResultsPossible( false ); |
| 136 | } |
| 137 | $term = $searchContext->escaper()->fixupQueryStringPart( $matches[ 0 ][ 0 ] ); |
| 138 | return [ |
| 139 | 'escaped' => self::switchSearchToExactForWildcards( $searchContext, $term ), |
| 140 | 'nonAll' => self::switchSearchToExactForWildcards( $searchContext, $term ) |
| 141 | ]; |
| 142 | } ); |
| 143 | |
| 144 | $escapedQuery = []; |
| 145 | $nonAllQuery = []; |
| 146 | $nearMatchQuery = []; |
| 147 | foreach ( $query as $queryPart ) { |
| 148 | if ( isset( $queryPart[ 'escaped' ] ) ) { |
| 149 | $escapedQuery[] = $queryPart[ 'escaped' ]; |
| 150 | $nonAllQuery[] = $queryPart['nonAll'] ?? $queryPart['escaped']; |
| 151 | continue; |
| 152 | } |
| 153 | if ( isset( $queryPart[ 'raw' ] ) ) { |
| 154 | $fixed = $searchContext->escaper()->fixupQueryStringPart( $queryPart[ 'raw' ] ); |
| 155 | $escapedQuery[] = $fixed; |
| 156 | $nonAllQuery[] = $fixed; |
| 157 | $nearMatchQuery[] = $queryPart[ 'raw' ]; |
| 158 | continue; |
| 159 | } |
| 160 | LoggerFactory::getInstance( 'CirrusSearch' )->warning( |
| 161 | 'Unknown query part: {queryPart}', |
| 162 | [ 'queryPart' => serialize( $queryPart ) ] |
| 163 | ); |
| 164 | } |
| 165 | |
| 166 | // Actual text query |
| 167 | $this->queryStringQueryString = |
| 168 | $searchContext->escaper()->fixupWholeQueryString( implode( ' ', $escapedQuery ) ); |
| 169 | $searchContext->setCleanedSearchTerm( $this->queryStringQueryString ); |
| 170 | |
| 171 | if ( $this->queryStringQueryString === '' ) { |
| 172 | $searchContext->addSyntaxUsed( 'filter_only' ); |
| 173 | $searchContext->setHighlightQuery( new MatchAll() ); |
| 174 | return; |
| 175 | } |
| 176 | |
| 177 | // Note that no escaping is required for near_match's match query. |
| 178 | $nearMatchQuery = implode( ' ', $nearMatchQuery ); |
| 179 | // If the near match is made only of spaces disable it. |
| 180 | if ( preg_match( '/^\s+$/', $nearMatchQuery ) === 1 ) { |
| 181 | $nearMatchQuery = ''; |
| 182 | } |
| 183 | |
| 184 | $queryStringRegex = |
| 185 | '(' . |
| 186 | // quoted strings |
| 187 | $matchQuotesRegex . |
| 188 | ')|(' . |
| 189 | // patterns that are seen before tokens. |
| 190 | '(^|\s)[+!-]\S' . |
| 191 | ')|(' . |
| 192 | // patterns seen after tokens. |
| 193 | '\S(?<!\\\\)~[0-9]?(\s|$)' . |
| 194 | ')|(' . |
| 195 | // patterns that are separated from tokens by whitespace |
| 196 | // on both sides. |
| 197 | '\s(AND|OR|NOT|&&|\\|\\|)\s' . |
| 198 | ')|(' . |
| 199 | // patterns that can be at the start of the string |
| 200 | '^NOT\s' . |
| 201 | ')|(' . |
| 202 | // patterns that can be inside tokens |
| 203 | // Note that question mark stripping has already been applied |
| 204 | '(?<!\\\\)[?*]' . |
| 205 | ')'; |
| 206 | if ( preg_match( "/$queryStringRegex/", $this->queryStringQueryString ) ) { |
| 207 | $searchContext->addSyntaxUsed( 'query_string' ); |
| 208 | } |
| 209 | $fields = array_merge( |
| 210 | self::buildFullTextSearchFields( $searchContext, 1, '.plain', true ), |
| 211 | self::buildFullTextSearchFields( $searchContext, |
| 212 | $this->config->get( 'CirrusSearchStemmedWeight' ), '', true ) ); |
| 213 | |
| 214 | $searchContext->setMainQuery( |
| 215 | $this->buildSearchTextQuery( |
| 216 | $searchContext, |
| 217 | $fields, |
| 218 | $this->nearMatchFieldQueryBuilder->buildFromQueryString( $nearMatchQuery ), |
| 219 | $this->queryStringQueryString |
| 220 | ) |
| 221 | ); |
| 222 | |
| 223 | // The highlighter doesn't know about the weighting from the all fields so we have to send |
| 224 | // it a query without the all fields. This swaps one in. |
| 225 | $nonAllFields = array_merge( |
| 226 | self::buildFullTextSearchFields( $searchContext, 1, '.plain', false ), |
| 227 | self::buildFullTextSearchFields( $searchContext, |
| 228 | $this->config->get( 'CirrusSearchStemmedWeight' ), '', false ) ); |
| 229 | $nonAllQueryString = $searchContext->escaper() |
| 230 | ->fixupWholeQueryString( implode( ' ', $nonAllQuery ) ); |
| 231 | $searchContext->setHighlightQuery( |
| 232 | $this->buildHighlightQuery( $searchContext, $nonAllFields, $nonAllQueryString, 1 ) |
| 233 | ); |
| 234 | |
| 235 | if ( $this->isPhraseRescoreNeeded( $searchContext ) ) { |
| 236 | $rescoreFields = $fields; |
| 237 | |
| 238 | $searchContext->setPhraseRescoreQuery( $this->buildPhraseRescoreQuery( |
| 239 | $searchContext, |
| 240 | $rescoreFields, |
| 241 | $this->queryStringQueryString, |
| 242 | $this->config->getElement( 'CirrusSearchPhraseSlop', 'boost' ) |
| 243 | ) ); |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | private function isPathologicalWildcard( string $term, int $maxWildcard ): bool { |
| 248 | $ret = preg_match_all( "/[*?]+/", $term ); |
| 249 | if ( $ret === false ) { |
| 250 | // we failed the regex, out of caution fail the query |
| 251 | return true; |
| 252 | } |
| 253 | return $ret > $maxWildcard; |
| 254 | } |
| 255 | |
| 256 | /** |
| 257 | * Attempt to build a degraded query from the query already built into $context. Must be |
| 258 | * called *after* self::build(). |
| 259 | * |
| 260 | * @param SearchContext $searchContext |
| 261 | * @return bool True if a degraded query was built |
| 262 | */ |
| 263 | public function buildDegraded( SearchContext $searchContext ) { |
| 264 | if ( $this->queryStringQueryString === '' ) { |
| 265 | return false; |
| 266 | } |
| 267 | |
| 268 | $fields = array_merge( |
| 269 | self::buildFullTextSearchFields( $searchContext, 1, '.plain', true ), |
| 270 | self::buildFullTextSearchFields( $searchContext, |
| 271 | $this->config->get( 'CirrusSearchStemmedWeight' ), '', true ) |
| 272 | ); |
| 273 | |
| 274 | $searchContext->addSyntaxUsed( 'degraded_full_text' ); |
| 275 | $simpleQuery = new \Elastica\Query\Simple( [ 'simple_query_string' => [ |
| 276 | 'fields' => $fields, |
| 277 | 'query' => $this->queryStringQueryString, |
| 278 | 'default_operator' => 'AND', |
| 279 | // Disable all costly operators |
| 280 | 'flags' => 'OR|AND' |
| 281 | ] ] ); |
| 282 | $searchContext->setMainQuery( $simpleQuery ); |
| 283 | $searchContext->setHighlightQuery( $simpleQuery ); |
| 284 | |
| 285 | return true; |
| 286 | } |
| 287 | |
| 288 | /** |
| 289 | * Build the primary query used for full text search. This will be a |
| 290 | * QueryString query, and optionally a MultiMatch if a $nearMatchQuery |
| 291 | * is provided. |
| 292 | * |
| 293 | * @param SearchContext $searchContext |
| 294 | * @param string[] $fields |
| 295 | * @param AbstractQuery $nearMatchQuery |
| 296 | * @param string $queryString |
| 297 | * @return \Elastica\Query\AbstractQuery |
| 298 | */ |
| 299 | protected function buildSearchTextQuery( |
| 300 | SearchContext $searchContext, |
| 301 | array $fields, |
| 302 | AbstractQuery $nearMatchQuery, |
| 303 | $queryString |
| 304 | ) { |
| 305 | $slop = $this->config->getElement( 'CirrusSearchPhraseSlop', 'default' ); |
| 306 | $queryForMostFields = $this->buildQueryString( $fields, $queryString, $slop ); |
| 307 | $searchContext->addSyntaxUsed( 'full_text_querystring', 5 ); |
| 308 | if ( $nearMatchQuery instanceof MatchNone ) { |
| 309 | return $queryForMostFields; |
| 310 | } |
| 311 | |
| 312 | // Build one query for the full text fields and one for the near match fields so that |
| 313 | // the near match can run unescaped. |
| 314 | $bool = new \Elastica\Query\BoolQuery(); |
| 315 | $bool->setMinimumShouldMatch( 1 ); |
| 316 | $bool->addShould( $queryForMostFields ); |
| 317 | $bool->addShould( $nearMatchQuery ); |
| 318 | |
| 319 | return $bool; |
| 320 | } |
| 321 | |
| 322 | /** |
| 323 | * Builds the query using the QueryString, this is the default builder |
| 324 | * used by cirrus and uses a default AND between clause. |
| 325 | * The query 'the query' and the fields all and all.plain will be like |
| 326 | * (all:the OR all.plain:the) AND (all:query OR all.plain:query) |
| 327 | * |
| 328 | * @param string[] $fields |
| 329 | * @param string $queryString |
| 330 | * @param int $phraseSlop |
| 331 | * @return \Elastica\Query\QueryString |
| 332 | */ |
| 333 | private function buildQueryString( array $fields, $queryString, $phraseSlop ) { |
| 334 | $query = new \Elastica\Query\QueryString( $queryString ); |
| 335 | $query->setFields( $fields ); |
| 336 | $query->setPhraseSlop( $phraseSlop ); |
| 337 | $query->setDefaultOperator( 'AND' ); |
| 338 | $query->setAllowLeadingWildcard( (bool)$this->config->get( 'CirrusSearchAllowLeadingWildcard' ) ); |
| 339 | $query->setFuzzyPrefixLength( 2 ); |
| 340 | $query->setRewrite( $this->getMultiTermRewriteMethod() ); |
| 341 | $states = $this->config->get( 'CirrusSearchQueryStringMaxDeterminizedStates' ); |
| 342 | if ( $states !== null ) { |
| 343 | $query->setParam( 'max_determinized_states', $states ); |
| 344 | } |
| 345 | return $query; |
| 346 | } |
| 347 | |
| 348 | /** |
| 349 | * the rewrite method to use for multi term queries |
| 350 | * @return string |
| 351 | */ |
| 352 | protected function getMultiTermRewriteMethod() { |
| 353 | return 'top_terms_boost_1024'; |
| 354 | } |
| 355 | |
| 356 | /** |
| 357 | * Expand wildcard queries to the all.plain and title.plain fields this is reasonable tradeoff |
| 358 | * between perf and precision. |
| 359 | * |
| 360 | * @param SearchContext $context |
| 361 | * @param string $term |
| 362 | * @return string |
| 363 | */ |
| 364 | private static function switchSearchToExactForWildcards( SearchContext $context, $term ) { |
| 365 | // Try to limit the expansion of wildcards to all the subfields |
| 366 | // We still need to add title.plain with a high boost otherwise |
| 367 | // match in titles be poorly scored (actually it breaks some tests). |
| 368 | $titleWeight = $context->getConfig()->getElement( 'CirrusSearchWeights', 'title' ); |
| 369 | return "(title.plain:$term^$titleWeight OR all.plain:$term)"; |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * Build a QueryString query where all fields being searched are |
| 374 | * queried for $term, joined with an OR. This is primarily for the |
| 375 | * benefit of the highlighter, the primary search is typically against |
| 376 | * the special all field. |
| 377 | * |
| 378 | * @param SearchContext $context |
| 379 | * @param string $term |
| 380 | * @param bool $allFieldAllowed |
| 381 | * @return string |
| 382 | */ |
| 383 | private static function switchSearchToExact( SearchContext $context, $term, $allFieldAllowed ) { |
| 384 | $exact = implode( ' OR ', |
| 385 | self::buildFullTextSearchFields( $context, 1, ".plain:$term", $allFieldAllowed ) ); |
| 386 | return "($exact)"; |
| 387 | } |
| 388 | |
| 389 | /** |
| 390 | * Build fields searched by full text search. |
| 391 | * |
| 392 | * @param SearchContext $context |
| 393 | * @param float $weight weight to multiply by all fields |
| 394 | * @param string $fieldSuffix suffix to add to field names |
| 395 | * @param bool $allFieldAllowed can we use the all field? False for |
| 396 | * collecting phrases for the highlighter. |
| 397 | * @return string[] array of fields to query |
| 398 | */ |
| 399 | private static function buildFullTextSearchFields( |
| 400 | SearchContext $context, |
| 401 | $weight, |
| 402 | $fieldSuffix, |
| 403 | $allFieldAllowed |
| 404 | ) { |
| 405 | $searchWeights = $context->getConfig()->get( 'CirrusSearchWeights' ); |
| 406 | |
| 407 | if ( $allFieldAllowed ) { |
| 408 | return [ "all{$fieldSuffix}^{$weight}" ]; |
| 409 | } |
| 410 | |
| 411 | $fields = []; |
| 412 | $titleWeight = $weight * $searchWeights[ 'title' ]; |
| 413 | $redirectWeight = $weight * $searchWeights[ 'redirect' ]; |
| 414 | $fields[] = "title{$fieldSuffix}^{$titleWeight}"; |
| 415 | $fields[] = "redirect.title{$fieldSuffix}^{$redirectWeight}"; |
| 416 | $categoryWeight = $weight * $searchWeights[ 'category' ]; |
| 417 | $headingWeight = $weight * $searchWeights[ 'heading' ]; |
| 418 | $openingTextWeight = $weight * $searchWeights[ 'opening_text' ]; |
| 419 | $textWeight = $weight * $searchWeights[ 'text' ]; |
| 420 | $auxiliaryTextWeight = $weight * $searchWeights[ 'auxiliary_text' ]; |
| 421 | $fields[] = "category{$fieldSuffix}^{$categoryWeight}"; |
| 422 | $fields[] = "heading{$fieldSuffix}^{$headingWeight}"; |
| 423 | $fields[] = "opening_text{$fieldSuffix}^{$openingTextWeight}"; |
| 424 | $fields[] = "text{$fieldSuffix}^{$textWeight}"; |
| 425 | $fields[] = "auxiliary_text{$fieldSuffix}^{$auxiliaryTextWeight}"; |
| 426 | $namespaces = $context->getNamespaces(); |
| 427 | if ( !$namespaces || in_array( NS_FILE, $namespaces ) ) { |
| 428 | $fileTextWeight = $weight * $searchWeights[ 'file_text' ]; |
| 429 | $fields[] = "file_text{$fieldSuffix}^{$fileTextWeight}"; |
| 430 | } |
| 431 | return $fields; |
| 432 | } |
| 433 | |
| 434 | /** |
| 435 | * Walks through an array of query pieces, as built by |
| 436 | * self::replacePartsOfQuery, and replaecs all raw pieces by the result of |
| 437 | * self::replacePartsOfQuery when called with the provided regex and |
| 438 | * callable. One query piece may turn into one or more query pieces in the |
| 439 | * result. |
| 440 | * |
| 441 | * @param array[] $query The set of query pieces to apply against |
| 442 | * @param string $regex Pieces of $queryPart that match this regex will |
| 443 | * be provided to $callable |
| 444 | * @param callable $callable A function accepting the $matches from preg_match |
| 445 | * and returning either a raw or escaped query piece. |
| 446 | * @return array[] The set of query pieces after applying regex and callable |
| 447 | */ |
| 448 | private static function replaceAllPartsOfQuery( array $query, $regex, $callable ) { |
| 449 | $result = []; |
| 450 | foreach ( $query as $queryPart ) { |
| 451 | if ( isset( $queryPart[ 'raw' ] ) ) { |
| 452 | $result = array_merge( $result, |
| 453 | self::replacePartsOfQuery( $queryPart[ 'raw' ], $regex, $callable ) ); |
| 454 | } else { |
| 455 | $result[] = $queryPart; |
| 456 | } |
| 457 | } |
| 458 | return $result; |
| 459 | } |
| 460 | |
| 461 | /** |
| 462 | * Splits a query string into one or more sequential pieces. Each piece |
| 463 | * of the query can either be raw (['raw'=>'stuff']), or escaped |
| 464 | * (['escaped'=>'stuff']). escaped can also optionally include a nonAll |
| 465 | * query (['escaped'=>'stuff','nonAll'=>'stuff']). If nonAll is not set |
| 466 | * the escaped query will be used. |
| 467 | * |
| 468 | * Pieces of $queryPart that do not match the provided $regex are tagged |
| 469 | * as 'raw' and may see further parsing. $callable receives pieces of |
| 470 | * the string that match the regex and must return either a raw or escaped |
| 471 | * query piece. |
| 472 | * |
| 473 | * @param string $queryPart Raw piece of a user supplied query string |
| 474 | * @param string $regex Pieces of $queryPart that match this regex will |
| 475 | * be provided to $callable |
| 476 | * @param callable $callable A function accepting the $matches from preg_match |
| 477 | * and returning either a raw or escaped query piece. |
| 478 | * @return array[] The sequential set of quer ypieces $queryPart was |
| 479 | * converted into. |
| 480 | */ |
| 481 | private static function replacePartsOfQuery( $queryPart, $regex, $callable ) { |
| 482 | $destination = []; |
| 483 | $matches = []; |
| 484 | $offset = 0; |
| 485 | while ( preg_match( $regex, $queryPart, $matches, PREG_OFFSET_CAPTURE, $offset ) ) { |
| 486 | $startOffset = $matches[0][1]; |
| 487 | if ( $startOffset > $offset ) { |
| 488 | $destination[] = [ |
| 489 | 'raw' => substr( $queryPart, $offset, $startOffset - $offset ) |
| 490 | ]; |
| 491 | } |
| 492 | |
| 493 | $callableResult = $callable( $matches ); |
| 494 | if ( $callableResult ) { |
| 495 | $destination[] = $callableResult; |
| 496 | } |
| 497 | |
| 498 | $offset = $startOffset + strlen( $matches[0][0] ); |
| 499 | } |
| 500 | |
| 501 | if ( $offset < strlen( $queryPart ) ) { |
| 502 | $destination[] = [ |
| 503 | 'raw' => substr( $queryPart, $offset ), |
| 504 | ]; |
| 505 | } |
| 506 | |
| 507 | return $destination; |
| 508 | } |
| 509 | |
| 510 | /** |
| 511 | * Builds the highlight query |
| 512 | * @param SearchContext $context |
| 513 | * @param string[] $fields |
| 514 | * @param string $queryText |
| 515 | * @param int $slop |
| 516 | * @return \Elastica\Query\AbstractQuery |
| 517 | */ |
| 518 | protected function buildHighlightQuery( SearchContext $context, array $fields, $queryText, $slop ) { |
| 519 | return $this->buildQueryString( $fields, $queryText, $slop ); |
| 520 | } |
| 521 | |
| 522 | /** |
| 523 | * Builds the phrase rescore query |
| 524 | * @param SearchContext $context |
| 525 | * @param string[] $fields |
| 526 | * @param string $queryText |
| 527 | * @param int $slop |
| 528 | * @return \Elastica\Query\AbstractQuery |
| 529 | */ |
| 530 | protected function buildPhraseRescoreQuery( SearchContext $context, array $fields, $queryText, $slop ) { |
| 531 | return $this->maybeWrapWithTokenCountRouter( |
| 532 | $queryText, |
| 533 | $this->buildQueryString( $fields, '"' . $queryText . '"', $slop ) |
| 534 | ); |
| 535 | } |
| 536 | |
| 537 | /** |
| 538 | * Determines if a phrase rescore is needed |
| 539 | * @param SearchContext $searchContext |
| 540 | * @return bool true if we can a phrase rescore |
| 541 | */ |
| 542 | protected function isPhraseRescoreNeeded( SearchContext $searchContext ) { |
| 543 | // Only do a phrase match rescore if the query doesn't include |
| 544 | // any quotes and has a space or the token count router is |
| 545 | // active. |
| 546 | // Queries without spaces are either single term or have a |
| 547 | // phrase query generated. |
| 548 | // Queries with the quote already contain a phrase query and we |
| 549 | // can't build phrase queries out of phrase queries at this |
| 550 | // point. |
| 551 | if ( !$searchContext->isSpecialKeywordUsed() && |
| 552 | strpos( $this->queryStringQueryString, '"' ) === false && |
| 553 | ( $this->useTokenCountRouter || strpos( $this->queryStringQueryString, ' ' ) !== false ) |
| 554 | ) { |
| 555 | return true; |
| 556 | } |
| 557 | return false; |
| 558 | } |
| 559 | |
| 560 | /** |
| 561 | * @param string $queryText |
| 562 | * @param AbstractQuery $query |
| 563 | * @return AbstractQuery |
| 564 | */ |
| 565 | protected function maybeWrapWithTokenCountRouter( $queryText, \Elastica\Query\AbstractQuery $query ) { |
| 566 | if ( $this->useTokenCountRouter ) { |
| 567 | $tokCount = new TokenCountRouter( |
| 568 | // text |
| 569 | $queryText, |
| 570 | // fallack |
| 571 | new \Elastica\Query\MatchNone(), |
| 572 | // field |
| 573 | 'text' |
| 574 | ); |
| 575 | $maxTokens = $this->config->get( 'CirrusSearchMaxPhraseTokens' ); |
| 576 | if ( $maxTokens ) { |
| 577 | $tokCount->addCondition( |
| 578 | TokenCountRouter::GT, |
| 579 | $maxTokens, |
| 580 | new \Elastica\Query\MatchNone() |
| 581 | ); |
| 582 | } |
| 583 | $tokCount->addCondition( |
| 584 | TokenCountRouter::GT, |
| 585 | 1, |
| 586 | $query |
| 587 | ); |
| 588 | return $tokCount; |
| 589 | } |
| 590 | return $query; |
| 591 | } |
| 592 | } |