Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
61.78% |
139 / 225 |
|
18.18% |
8 / 44 |
CRAP | |
0.00% |
0 / 1 |
| SearchEngine | |
61.78% |
139 / 225 |
|
18.18% |
8 / 44 |
598.96 | |
0.00% |
0 / 1 |
| searchText | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| searchArchiveTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchArchiveTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| searchTitle | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| maybePaginate | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
42 | |||
| supports | |
50.00% |
2 / 4 |
|
0.00% |
0 / 1 |
6.00 | |||
| setFeatureData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFeatureData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| normalizeText | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNearMatcher | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| defaultNearMatcher | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| legalSearchChars | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setLimitOffset | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setNamespaces | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
3.03 | |||
| setShowSuggestion | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getValidSorts | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setSort | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getSort | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| replacePrefixes | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| parseNamespacePrefixes | |
96.97% |
32 / 33 |
|
0.00% |
0 / 1 |
11 | |||
| userHighlightPrefs | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| update | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| updateTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| delete | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getTextFromContent | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| textAlreadyUpdatedForIndex | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| normalizeNamespaces | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| completionSearchBackendOverfetch | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| completionSearchBackend | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
3.02 | |||
| completionSearch | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
2.03 | |||
| completionSearchWithVariants | |
90.91% |
20 / 22 |
|
0.00% |
0 / 1 |
5.02 | |||
| extractTitles | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| processCompletionResults | |
80.00% |
28 / 35 |
|
0.00% |
0 / 1 |
7.39 | |||
| defaultPrefixSearch | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| simplePrefixSearch | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getProfiles | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| makeSearchFieldMapping | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getSearchIndexFields | |
81.82% |
18 / 22 |
|
0.00% |
0 / 1 |
7.29 | |||
| augmentSearchResults | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
56 | |||
| setHookContainer | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getHookContainer | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getHookRunner | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Basic search engine |
| 4 | * |
| 5 | * This program is free software; you can redistribute it and/or modify |
| 6 | * it under the terms of the GNU General Public License as published by |
| 7 | * the Free Software Foundation; either version 2 of the License, or |
| 8 | * (at your option) any later version. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 | * GNU General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License along |
| 16 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 17 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 18 | * http://www.gnu.org/copyleft/gpl.html |
| 19 | * |
| 20 | * @file |
| 21 | * @ingroup Search |
| 22 | */ |
| 23 | |
| 24 | /** |
| 25 | * @defgroup Search Search |
| 26 | */ |
| 27 | |
| 28 | use MediaWiki\Config\Config; |
| 29 | use MediaWiki\Content\Content; |
| 30 | use MediaWiki\Exception\MWUnknownContentModelException; |
| 31 | use MediaWiki\HookContainer\HookContainer; |
| 32 | use MediaWiki\HookContainer\HookRunner; |
| 33 | use MediaWiki\MediaWikiServices; |
| 34 | use MediaWiki\Search\TitleMatcher; |
| 35 | use MediaWiki\Status\Status; |
| 36 | use MediaWiki\Title\Title; |
| 37 | use MediaWiki\User\User; |
| 38 | |
| 39 | /** |
| 40 | * Contain a class for special pages |
| 41 | * @stable to extend |
| 42 | * @ingroup Search |
| 43 | */ |
| 44 | abstract class SearchEngine { |
| 45 | public const DEFAULT_SORT = 'relevance'; |
| 46 | |
| 47 | /** @var string */ |
| 48 | public $prefix = ''; |
| 49 | |
| 50 | /** @var int[]|null */ |
| 51 | public $namespaces = [ NS_MAIN ]; |
| 52 | |
| 53 | /** @var int */ |
| 54 | protected $limit = 10; |
| 55 | |
| 56 | /** @var int */ |
| 57 | protected $offset = 0; |
| 58 | |
| 59 | /** |
| 60 | * @var string[] |
| 61 | * @deprecated since 1.34 |
| 62 | */ |
| 63 | protected $searchTerms = []; |
| 64 | |
| 65 | /** @var bool */ |
| 66 | protected $showSuggestion = true; |
| 67 | /** @var string */ |
| 68 | private $sort = self::DEFAULT_SORT; |
| 69 | |
| 70 | /** @var array Feature values */ |
| 71 | protected $features = []; |
| 72 | |
| 73 | /** @var HookContainer */ |
| 74 | private $hookContainer; |
| 75 | |
| 76 | /** @var HookRunner */ |
| 77 | private $hookRunner; |
| 78 | |
| 79 | /** Profile type for completionSearch */ |
| 80 | public const COMPLETION_PROFILE_TYPE = 'completionSearchProfile'; |
| 81 | |
| 82 | /** Profile type for query independent ranking features */ |
| 83 | public const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile'; |
| 84 | |
| 85 | /** Integer flag for legalSearchChars: includes all chars allowed in a search query */ |
| 86 | protected const CHARS_ALL = 1; |
| 87 | |
| 88 | /** Integer flag for legalSearchChars: includes all chars allowed in a search term */ |
| 89 | protected const CHARS_NO_SYNTAX = 2; |
| 90 | |
| 91 | /** |
| 92 | * Perform a full text search query and return a result set. |
| 93 | * If full text searches are not supported or disabled, return null. |
| 94 | * |
| 95 | * @note As of 1.32 overriding this function is deprecated. It will |
| 96 | * be converted to final in 1.34. Override self::doSearchText(). |
| 97 | * |
| 98 | * @param string $term Raw search term |
| 99 | * @return ISearchResultSet|Status|null |
| 100 | */ |
| 101 | public function searchText( $term ) { |
| 102 | return $this->maybePaginate( function () use ( $term ) { |
| 103 | return $this->doSearchText( $term ); |
| 104 | } ); |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * Perform a full text search query and return a result set. |
| 109 | * |
| 110 | * @stable to override |
| 111 | * |
| 112 | * @param string $term Raw search term |
| 113 | * @return ISearchResultSet|Status|null |
| 114 | * @since 1.32 |
| 115 | */ |
| 116 | protected function doSearchText( $term ) { |
| 117 | return null; |
| 118 | } |
| 119 | |
| 120 | /** |
| 121 | * Perform a title search in the article archive. |
| 122 | * NOTE: these results still should be filtered by |
| 123 | * matching against PageArchive, permissions checks etc |
| 124 | * The results returned by this methods are only suggestions and |
| 125 | * may not end up being shown to the user. |
| 126 | * |
| 127 | * @note As of 1.32 overriding this function is deprecated. It will |
| 128 | * be converted to final in 1.34. Override self::doSearchArchiveTitle(). |
| 129 | * |
| 130 | * @param string $term Raw search term |
| 131 | * @return Status |
| 132 | * @since 1.29 |
| 133 | */ |
| 134 | public function searchArchiveTitle( $term ) { |
| 135 | return $this->doSearchArchiveTitle( $term ); |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * Perform a title search in the article archive. |
| 140 | * |
| 141 | * @stable to override |
| 142 | * |
| 143 | * @param string $term Raw search term |
| 144 | * @return Status |
| 145 | * @since 1.32 |
| 146 | */ |
| 147 | protected function doSearchArchiveTitle( $term ) { |
| 148 | return Status::newGood( [] ); |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Perform a title-only search query and return a result set. |
| 153 | * If title searches are not supported or disabled, return null. |
| 154 | * STUB |
| 155 | * |
| 156 | * @note As of 1.32 overriding this function is deprecated. It will |
| 157 | * be converted to final in 1.34. Override self::doSearchTitle(). |
| 158 | * |
| 159 | * @param string $term Raw search term |
| 160 | * @return ISearchResultSet|null |
| 161 | */ |
| 162 | public function searchTitle( $term ) { |
| 163 | return $this->maybePaginate( function () use ( $term ) { |
| 164 | return $this->doSearchTitle( $term ); |
| 165 | } ); |
| 166 | } |
| 167 | |
| 168 | /** |
| 169 | * Perform a title-only search query and return a result set. |
| 170 | * |
| 171 | * @stable to override |
| 172 | * |
| 173 | * @param string $term Raw search term |
| 174 | * @return ISearchResultSet|null |
| 175 | * @since 1.32 |
| 176 | */ |
| 177 | protected function doSearchTitle( $term ) { |
| 178 | return null; |
| 179 | } |
| 180 | |
| 181 | /** |
| 182 | * Performs an overfetch and shrink operation to determine if |
| 183 | * the next page is available for search engines that do not |
| 184 | * explicitly implement their own pagination. |
| 185 | * |
| 186 | * @param Closure $fn Takes no arguments |
| 187 | * @return ISearchResultSet|Status<ISearchResultSet>|null Result of calling $fn |
| 188 | */ |
| 189 | private function maybePaginate( Closure $fn ) { |
| 190 | if ( $this instanceof PaginatingSearchEngine ) { |
| 191 | return $fn(); |
| 192 | } |
| 193 | $this->limit++; |
| 194 | try { |
| 195 | $resultSetOrStatus = $fn(); |
| 196 | } finally { |
| 197 | $this->limit--; |
| 198 | } |
| 199 | |
| 200 | $resultSet = null; |
| 201 | if ( $resultSetOrStatus instanceof ISearchResultSet ) { |
| 202 | $resultSet = $resultSetOrStatus; |
| 203 | } elseif ( $resultSetOrStatus instanceof Status && |
| 204 | $resultSetOrStatus->getValue() instanceof ISearchResultSet |
| 205 | ) { |
| 206 | $resultSet = $resultSetOrStatus->getValue(); |
| 207 | } |
| 208 | if ( $resultSet ) { |
| 209 | $resultSet->shrink( $this->limit ); |
| 210 | } |
| 211 | |
| 212 | return $resultSetOrStatus; |
| 213 | } |
| 214 | |
| 215 | /** |
| 216 | * @since 1.18 |
| 217 | * @stable to override |
| 218 | * |
| 219 | * @param string $feature |
| 220 | * @return bool |
| 221 | */ |
| 222 | public function supports( $feature ) { |
| 223 | switch ( $feature ) { |
| 224 | case 'search-update': |
| 225 | return true; |
| 226 | case 'title-suffix-filter': |
| 227 | default: |
| 228 | return false; |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | /** |
| 233 | * Way to pass custom data for engines |
| 234 | * @since 1.18 |
| 235 | * @param string $feature |
| 236 | * @param mixed $data |
| 237 | */ |
| 238 | public function setFeatureData( $feature, $data ) { |
| 239 | $this->features[$feature] = $data; |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Way to retrieve custom data set by setFeatureData |
| 244 | * or by the engine itself. |
| 245 | * @since 1.29 |
| 246 | * @param string $feature feature name |
| 247 | * @return mixed the feature value or null if unset |
| 248 | */ |
| 249 | public function getFeatureData( $feature ) { |
| 250 | return $this->features[$feature] ?? null; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * When overridden in derived class, performs database-specific conversions |
| 255 | * on text to be used for searching or updating search index. |
| 256 | * Default implementation does nothing (simply returns $string). |
| 257 | * |
| 258 | * @param string $string String to process |
| 259 | * @return string |
| 260 | */ |
| 261 | public function normalizeText( $string ) { |
| 262 | // Some languages such as Chinese require word segmentation |
| 263 | return MediaWikiServices::getInstance()->getContentLanguage()->segmentByWord( $string ); |
| 264 | } |
| 265 | |
| 266 | /** |
| 267 | * Get service class to finding near matches. |
| 268 | * |
| 269 | * @return TitleMatcher |
| 270 | * @deprecated since 1.40, use MediaWikiServices::getInstance()->getTitleMatcher() |
| 271 | */ |
| 272 | public function getNearMatcher( Config $config ) { |
| 273 | return MediaWikiServices::getInstance()->getTitleMatcher(); |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Get near matcher for default SearchEngine. |
| 278 | * |
| 279 | * @return TitleMatcher |
| 280 | * @deprecated since 1.40, MediaWikiServices::getInstance()->getTitleMatcher() |
| 281 | */ |
| 282 | protected static function defaultNearMatcher() { |
| 283 | wfDeprecated( __METHOD__, '1.40' ); |
| 284 | return MediaWikiServices::getInstance()->getTitleMatcher(); |
| 285 | } |
| 286 | |
| 287 | /** |
| 288 | * Get chars legal for search |
| 289 | * @param int $type type of search chars (see self::CHARS_ALL |
| 290 | * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL |
| 291 | * @return string |
| 292 | */ |
| 293 | public function legalSearchChars( $type = self::CHARS_ALL ) { |
| 294 | return "A-Za-z_'.0-9\\x80-\\xFF\\-"; |
| 295 | } |
| 296 | |
| 297 | /** |
| 298 | * Set the maximum number of results to return |
| 299 | * and how many to skip before returning the first. |
| 300 | * |
| 301 | * @param int $limit |
| 302 | * @param int $offset |
| 303 | */ |
| 304 | public function setLimitOffset( $limit, $offset = 0 ) { |
| 305 | $this->limit = intval( $limit ); |
| 306 | $this->offset = intval( $offset ); |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Set which namespaces the search should include. |
| 311 | * Give an array of namespace index numbers. |
| 312 | * |
| 313 | * @param int[]|null $namespaces |
| 314 | */ |
| 315 | public function setNamespaces( $namespaces ) { |
| 316 | if ( $namespaces ) { |
| 317 | // Filter namespaces to only keep valid ones |
| 318 | $validNs = MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces(); |
| 319 | $namespaces = array_filter( $namespaces, static function ( $ns ) use( $validNs ) { |
| 320 | return $ns < 0 || isset( $validNs[$ns] ); |
| 321 | } ); |
| 322 | } else { |
| 323 | $namespaces = []; |
| 324 | } |
| 325 | $this->namespaces = $namespaces; |
| 326 | } |
| 327 | |
| 328 | /** |
| 329 | * Set whether the searcher should try to build a suggestion. Note: some searchers |
| 330 | * don't support building a suggestion in the first place and others don't respect |
| 331 | * this flag. |
| 332 | * |
| 333 | * @param bool $showSuggestion Should the searcher try to build suggestions |
| 334 | */ |
| 335 | public function setShowSuggestion( $showSuggestion ) { |
| 336 | $this->showSuggestion = $showSuggestion; |
| 337 | } |
| 338 | |
| 339 | /** |
| 340 | * Get the valid sort directions. All search engines support 'relevance' but others |
| 341 | * might support more. The default in all implementations must be 'relevance.' |
| 342 | * |
| 343 | * @since 1.25 |
| 344 | * @stable to override |
| 345 | * |
| 346 | * @return string[] the valid sort directions for setSort |
| 347 | */ |
| 348 | public function getValidSorts() { |
| 349 | return [ self::DEFAULT_SORT ]; |
| 350 | } |
| 351 | |
| 352 | /** |
| 353 | * Set the sort direction of the search results. Must be one returned by |
| 354 | * SearchEngine::getValidSorts() |
| 355 | * |
| 356 | * @since 1.25 |
| 357 | * @param string $sort sort direction for query result |
| 358 | */ |
| 359 | public function setSort( $sort ) { |
| 360 | if ( !in_array( $sort, $this->getValidSorts() ) ) { |
| 361 | throw new InvalidArgumentException( "Invalid sort: $sort. " . |
| 362 | "Must be one of: " . implode( ', ', $this->getValidSorts() ) ); |
| 363 | } |
| 364 | $this->sort = $sort; |
| 365 | } |
| 366 | |
| 367 | /** |
| 368 | * Get the sort direction of the search results |
| 369 | * |
| 370 | * @since 1.25 |
| 371 | * @return string |
| 372 | */ |
| 373 | public function getSort() { |
| 374 | return $this->sort; |
| 375 | } |
| 376 | |
| 377 | /** |
| 378 | * Parse some common prefixes: all (search everything) |
| 379 | * or namespace names and set the list of namespaces |
| 380 | * of this class accordingly. |
| 381 | * |
| 382 | * @deprecated since 1.32; should be handled internally by the search engine |
| 383 | * @param string $query |
| 384 | * @return string |
| 385 | */ |
| 386 | public function replacePrefixes( $query ) { |
| 387 | return $query; |
| 388 | } |
| 389 | |
| 390 | /** |
| 391 | * Parse some common prefixes: all (search everything) |
| 392 | * or namespace names |
| 393 | * |
| 394 | * @param string $query |
| 395 | * @param bool $withAllKeyword activate support of the "all:" keyword and its |
| 396 | * translations to activate searching on all namespaces. |
| 397 | * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook |
| 398 | * if classic namespace identification did not match. |
| 399 | * @return false|array false if no namespace was extracted, an array |
| 400 | * with the parsed query at index 0 and an array of namespaces at index |
| 401 | * 1 (or null for all namespaces). |
| 402 | */ |
| 403 | public static function parseNamespacePrefixes( |
| 404 | $query, |
| 405 | $withAllKeyword = true, |
| 406 | $withPrefixSearchExtractNamespaceHook = false |
| 407 | ) { |
| 408 | $parsed = $query; |
| 409 | if ( strpos( $query, ':' ) === false ) { // nothing to do |
| 410 | return false; |
| 411 | } |
| 412 | $extractedNamespace = null; |
| 413 | |
| 414 | $allQuery = false; |
| 415 | if ( $withAllKeyword ) { |
| 416 | $allkeywords = []; |
| 417 | |
| 418 | $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":"; |
| 419 | // force all: so that we have a common syntax for all the wikis |
| 420 | if ( !in_array( 'all:', $allkeywords ) ) { |
| 421 | $allkeywords[] = 'all:'; |
| 422 | } |
| 423 | |
| 424 | foreach ( $allkeywords as $kw ) { |
| 425 | if ( str_starts_with( $query, $kw ) ) { |
| 426 | $parsed = substr( $query, strlen( $kw ) ); |
| 427 | $allQuery = true; |
| 428 | break; |
| 429 | } |
| 430 | } |
| 431 | } |
| 432 | |
| 433 | if ( !$allQuery && strpos( $query, ':' ) !== false ) { |
| 434 | $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) ); |
| 435 | $services = MediaWikiServices::getInstance(); |
| 436 | $index = $services->getContentLanguage()->getNsIndex( $prefix ); |
| 437 | if ( $index !== false ) { |
| 438 | $extractedNamespace = [ $index ]; |
| 439 | $parsed = substr( $query, strlen( $prefix ) + 1 ); |
| 440 | } elseif ( $withPrefixSearchExtractNamespaceHook ) { |
| 441 | $hookNamespaces = [ NS_MAIN ]; |
| 442 | $hookQuery = $query; |
| 443 | ( new HookRunner( $services->getHookContainer() ) ) |
| 444 | ->onPrefixSearchExtractNamespace( $hookNamespaces, $hookQuery ); |
| 445 | if ( $hookQuery !== $query ) { |
| 446 | $parsed = $hookQuery; |
| 447 | $extractedNamespace = $hookNamespaces; |
| 448 | } else { |
| 449 | return false; |
| 450 | } |
| 451 | } else { |
| 452 | return false; |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | return [ $parsed, $extractedNamespace ]; |
| 457 | } |
| 458 | |
| 459 | /** |
| 460 | * Find snippet highlight settings for all users |
| 461 | * @return array Contextlines, contextchars |
| 462 | * @deprecated since 1.34; use the SearchHighlighter constants directly |
| 463 | * @see SearchHighlighter::DEFAULT_CONTEXT_CHARS |
| 464 | * @see SearchHighlighter::DEFAULT_CONTEXT_LINES |
| 465 | */ |
| 466 | public static function userHighlightPrefs() { |
| 467 | $contextlines = SearchHighlighter::DEFAULT_CONTEXT_LINES; |
| 468 | $contextchars = SearchHighlighter::DEFAULT_CONTEXT_CHARS; |
| 469 | return [ $contextlines, $contextchars ]; |
| 470 | } |
| 471 | |
| 472 | /** |
| 473 | * Create or update the search index record for the given page. |
| 474 | * Title and text should be pre-processed. |
| 475 | * STUB |
| 476 | * |
| 477 | * @param int $id |
| 478 | * @param string $title |
| 479 | * @param string $text |
| 480 | */ |
| 481 | public function update( $id, $title, $text ) { |
| 482 | // no-op |
| 483 | } |
| 484 | |
| 485 | /** |
| 486 | * Update a search index record's title only. |
| 487 | * Title should be pre-processed. |
| 488 | * STUB |
| 489 | * |
| 490 | * @param int $id |
| 491 | * @param string $title |
| 492 | */ |
| 493 | public function updateTitle( $id, $title ) { |
| 494 | // no-op |
| 495 | } |
| 496 | |
| 497 | /** |
| 498 | * Delete an indexed page |
| 499 | * Title should be pre-processed. |
| 500 | * STUB |
| 501 | * |
| 502 | * @param int $id Page id that was deleted |
| 503 | * @param string $title Title of page that was deleted |
| 504 | */ |
| 505 | public function delete( $id, $title ) { |
| 506 | // no-op |
| 507 | } |
| 508 | |
| 509 | /** |
| 510 | * Get the raw text for updating the index from a content object |
| 511 | * Nicer search backends could possibly do something cooler than |
| 512 | * just returning raw text |
| 513 | * |
| 514 | * @todo This isn't ideal, we'd really like to have content-specific handling here |
| 515 | * @param Title $t Title we're indexing |
| 516 | * @param Content|null $c Content of the page to index |
| 517 | * @return string |
| 518 | * @deprecated since 1.34 use Content::getTextForSearchIndex directly |
| 519 | */ |
| 520 | public function getTextFromContent( Title $t, ?Content $c = null ) { |
| 521 | return $c ? $c->getTextForSearchIndex() : ''; |
| 522 | } |
| 523 | |
| 524 | /** |
| 525 | * If an implementation of SearchEngine handles all of its own text processing |
| 526 | * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s |
| 527 | * rather silly handling, it should return true here instead. |
| 528 | * |
| 529 | * @return bool |
| 530 | * @deprecated since 1.34 no longer needed since getTextFromContent is being deprecated |
| 531 | */ |
| 532 | public function textAlreadyUpdatedForIndex() { |
| 533 | return false; |
| 534 | } |
| 535 | |
| 536 | /** |
| 537 | * Makes search simple string if it was namespaced. |
| 538 | * Sets namespaces of the search to namespaces extracted from string. |
| 539 | * @param string $search |
| 540 | * @return string Simplified search string |
| 541 | */ |
| 542 | protected function normalizeNamespaces( $search ) { |
| 543 | $queryAndNs = self::parseNamespacePrefixes( $search, false, true ); |
| 544 | if ( $queryAndNs !== false ) { |
| 545 | $this->setNamespaces( $queryAndNs[1] ); |
| 546 | return $queryAndNs[0]; |
| 547 | } |
| 548 | return $search; |
| 549 | } |
| 550 | |
| 551 | /** |
| 552 | * Perform an overfetch of completion search results. This allows |
| 553 | * determining if another page of results is available. |
| 554 | * |
| 555 | * @param string $search |
| 556 | * @return SearchSuggestionSet |
| 557 | */ |
| 558 | protected function completionSearchBackendOverfetch( $search ) { |
| 559 | $this->limit++; |
| 560 | try { |
| 561 | return $this->completionSearchBackend( $search ); |
| 562 | } finally { |
| 563 | $this->limit--; |
| 564 | } |
| 565 | } |
| 566 | |
| 567 | /** |
| 568 | * Perform a completion search. |
| 569 | * Does not resolve namespaces and does not check variants. |
| 570 | * Search engine implementations may want to override this function. |
| 571 | * |
| 572 | * @stable to override |
| 573 | * |
| 574 | * @param string $search |
| 575 | * @return SearchSuggestionSet |
| 576 | */ |
| 577 | protected function completionSearchBackend( $search ) { |
| 578 | $results = []; |
| 579 | |
| 580 | $search = trim( $search ); |
| 581 | |
| 582 | if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search |
| 583 | !$this->getHookRunner()->onPrefixSearchBackend( |
| 584 | $this->namespaces, $search, $this->limit, $results, $this->offset ) |
| 585 | ) { |
| 586 | // False means hook worked. |
| 587 | // FIXME: Yes, the API is weird. That's why it is going to be deprecated. |
| 588 | |
| 589 | return SearchSuggestionSet::fromStrings( $results ); |
| 590 | } else { |
| 591 | // Hook did not do the job, use default simple search |
| 592 | $results = $this->simplePrefixSearch( $search ); |
| 593 | return SearchSuggestionSet::fromTitles( $results ); |
| 594 | } |
| 595 | } |
| 596 | |
| 597 | /** |
| 598 | * Perform a completion search. |
| 599 | * @param string $search |
| 600 | * @return SearchSuggestionSet |
| 601 | */ |
| 602 | public function completionSearch( $search ) { |
| 603 | if ( trim( $search ) === '' ) { |
| 604 | return SearchSuggestionSet::emptySuggestionSet(); // Return empty result |
| 605 | } |
| 606 | $search = $this->normalizeNamespaces( $search ); |
| 607 | $suggestions = $this->completionSearchBackendOverfetch( $search ); |
| 608 | return $this->processCompletionResults( $search, $suggestions ); |
| 609 | } |
| 610 | |
| 611 | /** |
| 612 | * Perform a completion search with variants. |
| 613 | * @stable to override |
| 614 | * |
| 615 | * @param string $search |
| 616 | * @return SearchSuggestionSet |
| 617 | */ |
| 618 | public function completionSearchWithVariants( $search ) { |
| 619 | if ( trim( $search ) === '' ) { |
| 620 | return SearchSuggestionSet::emptySuggestionSet(); // Return empty result |
| 621 | } |
| 622 | $search = $this->normalizeNamespaces( $search ); |
| 623 | |
| 624 | $results = $this->completionSearchBackendOverfetch( $search ); |
| 625 | $fallbackLimit = 1 + $this->limit - $results->getSize(); |
| 626 | if ( $fallbackLimit > 0 ) { |
| 627 | $services = MediaWikiServices::getInstance(); |
| 628 | $fallbackSearches = $services->getLanguageConverterFactory() |
| 629 | ->getLanguageConverter( $services->getContentLanguage() ) |
| 630 | ->autoConvertToAllVariants( $search ); |
| 631 | $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] ); |
| 632 | |
| 633 | $origLimit = $this->limit; |
| 634 | $origOffset = $this->offset; |
| 635 | foreach ( $fallbackSearches as $fbs ) { |
| 636 | try { |
| 637 | $this->setLimitOffset( $fallbackLimit ); |
| 638 | $fallbackSearchResult = $this->completionSearch( $fbs ); |
| 639 | $results->appendAll( $fallbackSearchResult ); |
| 640 | $fallbackLimit -= $fallbackSearchResult->getSize(); |
| 641 | } finally { |
| 642 | $this->setLimitOffset( $origLimit, $origOffset ); |
| 643 | } |
| 644 | if ( $fallbackLimit <= 0 ) { |
| 645 | break; |
| 646 | } |
| 647 | } |
| 648 | } |
| 649 | return $this->processCompletionResults( $search, $results ); |
| 650 | } |
| 651 | |
| 652 | /** |
| 653 | * Extract titles from completion results |
| 654 | * @param SearchSuggestionSet $completionResults |
| 655 | * @return Title[] |
| 656 | */ |
| 657 | public function extractTitles( SearchSuggestionSet $completionResults ) { |
| 658 | return $completionResults->map( static function ( SearchSuggestion $sugg ) { |
| 659 | return $sugg->getSuggestedTitle(); |
| 660 | } ); |
| 661 | } |
| 662 | |
| 663 | /** |
| 664 | * Process completion search results. |
| 665 | * Resolves the titles and rescores. |
| 666 | * @param string $search |
| 667 | * @param SearchSuggestionSet $suggestions |
| 668 | * @return SearchSuggestionSet |
| 669 | */ |
| 670 | protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) { |
| 671 | // We over-fetched to determine pagination. Shrink back down if we have extra results |
| 672 | // and mark if pagination is possible |
| 673 | $suggestions->shrink( $this->limit ); |
| 674 | |
| 675 | $search = trim( $search ); |
| 676 | // preload the titles with LinkBatch |
| 677 | $suggestedTitles = $suggestions->map( static function ( SearchSuggestion $sugg ) { |
| 678 | return $sugg->getSuggestedTitle(); |
| 679 | } ); |
| 680 | $linkBatchFactory = MediaWikiServices::getInstance()->getLinkBatchFactory(); |
| 681 | $linkBatchFactory->newLinkBatch( $suggestedTitles ) |
| 682 | ->setCaller( __METHOD__ ) |
| 683 | ->execute(); |
| 684 | |
| 685 | $diff = $suggestions->filter( static function ( SearchSuggestion $sugg ) { |
| 686 | return $sugg->getSuggestedTitle()->isKnown(); |
| 687 | } ); |
| 688 | if ( $diff > 0 ) { |
| 689 | $statsFactory = MediaWikiServices::getInstance()->getStatsFactory(); |
| 690 | $statsFactory->getCounter( 'search_completion_missing_total' ) |
| 691 | ->incrementBy( $diff ); |
| 692 | } |
| 693 | |
| 694 | // SearchExactMatchRescorer should probably be refactored to work directly on top of a SearchSuggestionSet |
| 695 | // instead of converting it to array and trying to infer if it has re-scored anything by inspected the head |
| 696 | // of the returned array. |
| 697 | $results = $suggestions->map( static function ( SearchSuggestion $sugg ) { |
| 698 | return $sugg->getSuggestedTitle()->getPrefixedText(); |
| 699 | } ); |
| 700 | |
| 701 | $rescorer = new SearchExactMatchRescorer(); |
| 702 | if ( $this->offset === 0 ) { |
| 703 | // Rescore results with an exact title match |
| 704 | // NOTE: in some cases like cross-namespace redirects |
| 705 | // (frequently used as shortcuts e.g. WP:WP on huwiki) some |
| 706 | // backends like Cirrus will return no results. We should still |
| 707 | // try an exact title match to workaround this limitation |
| 708 | $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit ); |
| 709 | } else { |
| 710 | // No need to rescore if offset is not 0 |
| 711 | // The exact match must have been returned at position 0 |
| 712 | // if it existed. |
| 713 | $rescoredResults = $results; |
| 714 | } |
| 715 | |
| 716 | if ( count( $rescoredResults ) > 0 ) { |
| 717 | $found = array_search( $rescoredResults[0], $results ); |
| 718 | if ( $found === false ) { |
| 719 | // If the first result is not in the previous array it |
| 720 | // means that we found a new exact match |
| 721 | $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) ); |
| 722 | $suggestions->prepend( $exactMatch ); |
| 723 | if ( $rescorer->getReplacedRedirect() !== null ) { |
| 724 | // the exact match rescorer replaced one of the suggestion found by the search engine |
| 725 | // let's remove it from our suggestions set to avoid showing duplicates |
| 726 | $suggestions->remove( SearchSuggestion::fromTitle( 0, |
| 727 | Title::newFromText( $rescorer->getReplacedRedirect() ) ) ); |
| 728 | } |
| 729 | $suggestions->shrink( $this->limit ); |
| 730 | } else { |
| 731 | // if the first result is not the same we need to rescore |
| 732 | if ( $found > 0 ) { |
| 733 | $suggestions->rescore( $found ); |
| 734 | } |
| 735 | } |
| 736 | } |
| 737 | |
| 738 | return $suggestions; |
| 739 | } |
| 740 | |
| 741 | /** |
| 742 | * Simple prefix search for subpages. |
| 743 | * @param string $search |
| 744 | * @return Title[] |
| 745 | */ |
| 746 | public function defaultPrefixSearch( $search ) { |
| 747 | if ( trim( $search ) === '' ) { |
| 748 | return []; |
| 749 | } |
| 750 | |
| 751 | $search = $this->normalizeNamespaces( $search ); |
| 752 | return $this->simplePrefixSearch( $search ); |
| 753 | } |
| 754 | |
| 755 | /** |
| 756 | * Call out to simple search backend. |
| 757 | * Defaults to TitlePrefixSearch. |
| 758 | * @param string $search |
| 759 | * @return Title[] |
| 760 | */ |
| 761 | protected function simplePrefixSearch( $search ) { |
| 762 | // Use default database prefix search |
| 763 | $backend = new TitlePrefixSearch; |
| 764 | return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset ); |
| 765 | } |
| 766 | |
| 767 | /** |
| 768 | * Get a list of supported profiles. |
| 769 | * Some search engine implementations may expose specific profiles to fine-tune |
| 770 | * its behaviors. |
| 771 | * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName ) |
| 772 | * The array returned by this function contains the following keys: |
| 773 | * - name: the profile name to use with setFeatureData |
| 774 | * - desc-message: the i18n description |
| 775 | * - default: set to true if this profile is the default |
| 776 | * |
| 777 | * @since 1.28 |
| 778 | * @stable to override |
| 779 | * |
| 780 | * @param string $profileType the type of profiles |
| 781 | * @param User|null $user the user requesting the list of profiles |
| 782 | * @return array|null the list of profiles or null if none available |
| 783 | * @phan-return null|array{name:string,desc-message:string,default?:bool} |
| 784 | */ |
| 785 | public function getProfiles( $profileType, ?User $user = null ) { |
| 786 | return null; |
| 787 | } |
| 788 | |
| 789 | /** |
| 790 | * Create a search field definition. |
| 791 | * Specific search engines should override this method to create search fields. |
| 792 | * @stable to override |
| 793 | * |
| 794 | * @param string $name |
| 795 | * @param string $type One of the types in SearchIndexField::INDEX_TYPE_* |
| 796 | * @return SearchIndexField |
| 797 | * @since 1.28 |
| 798 | */ |
| 799 | public function makeSearchFieldMapping( $name, $type ) { |
| 800 | return new NullIndexField(); |
| 801 | } |
| 802 | |
| 803 | /** |
| 804 | * Get fields for search index |
| 805 | * @since 1.28 |
| 806 | * @return SearchIndexField[] Index field definitions for all content handlers |
| 807 | */ |
| 808 | public function getSearchIndexFields() { |
| 809 | $models = MediaWikiServices::getInstance()->getContentHandlerFactory()->getContentModels(); |
| 810 | $fields = []; |
| 811 | $seenHandlers = new SplObjectStorage(); |
| 812 | foreach ( $models as $model ) { |
| 813 | try { |
| 814 | $handler = MediaWikiServices::getInstance() |
| 815 | ->getContentHandlerFactory() |
| 816 | ->getContentHandler( $model ); |
| 817 | } catch ( MWUnknownContentModelException ) { |
| 818 | // If we can find no handler, ignore it |
| 819 | continue; |
| 820 | } |
| 821 | // Several models can have the same handler, so avoid processing it repeatedly |
| 822 | if ( $seenHandlers->contains( $handler ) ) { |
| 823 | // We already did this one |
| 824 | continue; |
| 825 | } |
| 826 | $seenHandlers->attach( $handler ); |
| 827 | $handlerFields = $handler->getFieldsForSearchIndex( $this ); |
| 828 | foreach ( $handlerFields as $fieldName => $fieldData ) { |
| 829 | if ( empty( $fields[$fieldName] ) ) { |
| 830 | $fields[$fieldName] = $fieldData; |
| 831 | } else { |
| 832 | // TODO: do we allow some clashes with the same type or reject all of them? |
| 833 | $mergeDef = $fields[$fieldName]->merge( $fieldData ); |
| 834 | if ( !$mergeDef ) { |
| 835 | throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" ); |
| 836 | } |
| 837 | $fields[$fieldName] = $mergeDef; |
| 838 | } |
| 839 | } |
| 840 | } |
| 841 | // Hook to allow extensions to produce search mapping fields |
| 842 | $this->getHookRunner()->onSearchIndexFields( $fields, $this ); |
| 843 | return $fields; |
| 844 | } |
| 845 | |
| 846 | /** |
| 847 | * Augment search results with extra data. |
| 848 | */ |
| 849 | public function augmentSearchResults( ISearchResultSet $resultSet ) { |
| 850 | $setAugmentors = []; |
| 851 | $rowAugmentors = []; |
| 852 | $this->getHookRunner()->onSearchResultsAugment( $setAugmentors, $rowAugmentors ); |
| 853 | if ( !$setAugmentors && !$rowAugmentors ) { |
| 854 | // We're done here |
| 855 | return; |
| 856 | } |
| 857 | |
| 858 | // Convert row augmentors to set augmentor |
| 859 | foreach ( $rowAugmentors as $name => $row ) { |
| 860 | if ( isset( $setAugmentors[$name] ) ) { |
| 861 | throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" ); |
| 862 | } |
| 863 | $setAugmentors[$name] = new PerRowAugmentor( $row ); |
| 864 | } |
| 865 | |
| 866 | /** |
| 867 | * @var string $name |
| 868 | * @var ResultSetAugmentor $augmentor |
| 869 | */ |
| 870 | foreach ( $setAugmentors as $name => $augmentor ) { |
| 871 | $data = $augmentor->augmentAll( $resultSet ); |
| 872 | if ( $data ) { |
| 873 | $resultSet->setAugmentedData( $name, $data ); |
| 874 | } |
| 875 | } |
| 876 | } |
| 877 | |
| 878 | /** |
| 879 | * @since 1.35 |
| 880 | * @internal |
| 881 | * @param HookContainer $hookContainer |
| 882 | */ |
| 883 | public function setHookContainer( HookContainer $hookContainer ) { |
| 884 | $this->hookContainer = $hookContainer; |
| 885 | $this->hookRunner = new HookRunner( $hookContainer ); |
| 886 | } |
| 887 | |
| 888 | /** |
| 889 | * Get a HookContainer, for running extension hooks or for hook metadata. |
| 890 | * |
| 891 | * @since 1.35 |
| 892 | * @return HookContainer |
| 893 | */ |
| 894 | protected function getHookContainer(): HookContainer { |
| 895 | if ( !$this->hookContainer ) { |
| 896 | // This shouldn't be hit in core, but it is needed for CirrusSearch |
| 897 | // which commonly creates a CirrusSearch object without cirrus being |
| 898 | // configured in $wgSearchType/$wgSearchTypeAlternatives. |
| 899 | $this->hookContainer = MediaWikiServices::getInstance()->getHookContainer(); |
| 900 | } |
| 901 | return $this->hookContainer; |
| 902 | } |
| 903 | |
| 904 | /** |
| 905 | * Get a HookRunner for running core hooks. |
| 906 | * |
| 907 | * @internal This is for use by core only. Hook interfaces may be removed |
| 908 | * without notice. |
| 909 | * @since 1.35 |
| 910 | * @return HookRunner |
| 911 | */ |
| 912 | protected function getHookRunner(): HookRunner { |
| 913 | if ( !$this->hookRunner ) { |
| 914 | $this->hookRunner = new HookRunner( $this->getHookContainer() ); |
| 915 | } |
| 916 | return $this->hookRunner; |
| 917 | } |
| 918 | |
| 919 | } |