Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 127 |
|
0.00% |
0 / 15 |
CRAP | |
0.00% |
0 / 1 |
| SearchSqlite | |
0.00% |
0 / 127 |
|
0.00% |
0 / 15 |
1332 | |
0.00% |
0 / 1 |
| fulltextSearchSupported | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| parseQuery | |
0.00% |
0 / 48 |
|
0.00% |
0 / 1 |
156 | |||
| regexTerm | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| legalSearchChars | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| doSearchTextInDB | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchTitleInDB | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| searchInternal | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
12 | |||
| queryNamespaces | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| limitResult | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getQuery | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| getIndexField | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| queryMain | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| getCountQuery | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| update | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
6 | |||
| updateTitle | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * SQLite search backend, based upon SearchMysql |
| 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 | use MediaWiki\MediaWikiServices; |
| 25 | use Wikimedia\AtEase\AtEase; |
| 26 | use Wikimedia\Rdbms\IDatabase; |
| 27 | |
| 28 | /** |
| 29 | * Search engine hook for SQLite |
| 30 | * @ingroup Search |
| 31 | */ |
| 32 | class SearchSqlite extends SearchDatabase { |
| 33 | /** |
| 34 | * Whether fulltext search is supported by current schema |
| 35 | * @return bool |
| 36 | */ |
| 37 | private function fulltextSearchSupported() { |
| 38 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 39 | $sql = (string)$dbr->newSelectQueryBuilder() |
| 40 | ->select( 'sql' ) |
| 41 | ->from( 'sqlite_master' ) |
| 42 | ->where( [ 'tbl_name' => $dbr->tableName( 'searchindex', 'raw' ) ] ) |
| 43 | ->caller( __METHOD__ )->fetchField(); |
| 44 | |
| 45 | return ( stristr( $sql, 'fts' ) !== false ); |
| 46 | } |
| 47 | |
| 48 | /** |
| 49 | * Parse the user's query and transform it into an SQL fragment which will |
| 50 | * become part of a WHERE clause |
| 51 | * |
| 52 | * @param string $filteredText |
| 53 | * @param bool $fulltext |
| 54 | * @return string |
| 55 | */ |
| 56 | private function parseQuery( $filteredText, $fulltext ) { |
| 57 | $lc = $this->legalSearchChars( self::CHARS_NO_SYNTAX ); // Minus syntax chars (" and *) |
| 58 | $searchon = ''; |
| 59 | $this->searchTerms = []; |
| 60 | |
| 61 | $m = []; |
| 62 | if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| 63 | $filteredText, $m, PREG_SET_ORDER ) ) { |
| 64 | foreach ( $m as $bits ) { |
| 65 | AtEase::suppressWarnings(); |
| 66 | [ /* all */, $modifier, $term, $nonQuoted, $wildcard ] = $bits; |
| 67 | AtEase::restoreWarnings(); |
| 68 | |
| 69 | if ( $nonQuoted != '' ) { |
| 70 | $term = $nonQuoted; |
| 71 | $quote = ''; |
| 72 | } else { |
| 73 | $term = str_replace( '"', '', $term ); |
| 74 | $quote = '"'; |
| 75 | } |
| 76 | |
| 77 | if ( $searchon !== '' ) { |
| 78 | $searchon .= ' '; |
| 79 | } |
| 80 | |
| 81 | // Some languages such as Serbian store the input form in the search index, |
| 82 | // so we may need to search for matches in multiple writing system variants. |
| 83 | |
| 84 | $converter = MediaWikiServices::getInstance()->getLanguageConverterFactory() |
| 85 | ->getLanguageConverter(); |
| 86 | $convertedVariants = $converter->autoConvertToAllVariants( $term ); |
| 87 | if ( is_array( $convertedVariants ) ) { |
| 88 | $variants = array_unique( array_values( $convertedVariants ) ); |
| 89 | } else { |
| 90 | $variants = [ $term ]; |
| 91 | } |
| 92 | |
| 93 | // The low-level search index does some processing on input to work |
| 94 | // around problems with minimum lengths and encoding in MySQL's |
| 95 | // fulltext engine. |
| 96 | // For Chinese this also inserts spaces between adjacent Han characters. |
| 97 | $strippedVariants = array_map( |
| 98 | [ MediaWikiServices::getInstance()->getContentLanguage(), |
| 99 | 'normalizeForSearch' ], |
| 100 | $variants ); |
| 101 | |
| 102 | // Some languages such as Chinese force all variants to a canonical |
| 103 | // form when stripping to the low-level search index, so to be sure |
| 104 | // let's check our variants list for unique items after stripping. |
| 105 | $strippedVariants = array_unique( $strippedVariants ); |
| 106 | |
| 107 | $searchon .= $modifier; |
| 108 | if ( count( $strippedVariants ) > 1 ) { |
| 109 | $searchon .= '('; |
| 110 | } |
| 111 | $count = 0; |
| 112 | foreach ( $strippedVariants as $stripped ) { |
| 113 | if ( $nonQuoted && strpos( $stripped, ' ' ) !== false ) { |
| 114 | // Hack for Chinese: we need to toss in quotes for |
| 115 | // multiple-character phrases since normalizeForSearch() |
| 116 | // added spaces between them to make word breaks. |
| 117 | $stripped = '"' . trim( $stripped ) . '"'; |
| 118 | } |
| 119 | if ( $count > 0 ) { |
| 120 | $searchon .= " OR "; |
| 121 | } |
| 122 | $searchon .= "$quote$stripped$quote$wildcard "; |
| 123 | ++$count; |
| 124 | } |
| 125 | if ( count( $strippedVariants ) > 1 ) { |
| 126 | $searchon .= ')'; |
| 127 | } |
| 128 | |
| 129 | // Match individual terms or quoted phrase in result highlighting... |
| 130 | // Note that variants will be introduced in a later stage for highlighting! |
| 131 | $regexp = $this->regexTerm( $term, $wildcard ); |
| 132 | $this->searchTerms[] = $regexp; |
| 133 | } |
| 134 | |
| 135 | } else { |
| 136 | wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'" ); |
| 137 | } |
| 138 | |
| 139 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 140 | $searchon = $dbr->addQuotes( $searchon ); |
| 141 | $field = $this->getIndexField( $fulltext ); |
| 142 | |
| 143 | return " $field MATCH $searchon "; |
| 144 | } |
| 145 | |
| 146 | private function regexTerm( $string, $wildcard ) { |
| 147 | $regex = preg_quote( $string, '/' ); |
| 148 | if ( MediaWikiServices::getInstance()->getContentLanguage()->hasWordBreaks() ) { |
| 149 | if ( $wildcard ) { |
| 150 | // Don't cut off the final bit! |
| 151 | $regex = "\b$regex"; |
| 152 | } else { |
| 153 | $regex = "\b$regex\b"; |
| 154 | } |
| 155 | } else { |
| 156 | // For Chinese, words may legitimately abut other words in the text literal. |
| 157 | // Don't add \b boundary checks... note this could cause false positives |
| 158 | // for Latin chars. |
| 159 | } |
| 160 | return $regex; |
| 161 | } |
| 162 | |
| 163 | public function legalSearchChars( $type = self::CHARS_ALL ) { |
| 164 | $searchChars = parent::legalSearchChars( $type ); |
| 165 | if ( $type === self::CHARS_ALL ) { |
| 166 | // " for phrase, * for wildcard |
| 167 | $searchChars = "\"*" . $searchChars; |
| 168 | } |
| 169 | return $searchChars; |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * Perform a full text search query and return a result set. |
| 174 | * |
| 175 | * @param string $term Raw search term |
| 176 | * @return SqlSearchResultSet|null |
| 177 | */ |
| 178 | protected function doSearchTextInDB( $term ) { |
| 179 | return $this->searchInternal( $term, true ); |
| 180 | } |
| 181 | |
| 182 | /** |
| 183 | * Perform a title-only search query and return a result set. |
| 184 | * |
| 185 | * @param string $term Raw search term |
| 186 | * @return SqlSearchResultSet|null |
| 187 | */ |
| 188 | protected function doSearchTitleInDB( $term ) { |
| 189 | return $this->searchInternal( $term, false ); |
| 190 | } |
| 191 | |
| 192 | protected function searchInternal( $term, $fulltext ) { |
| 193 | if ( !$this->fulltextSearchSupported() ) { |
| 194 | return null; |
| 195 | } |
| 196 | |
| 197 | $filteredTerm = |
| 198 | $this->filter( MediaWikiServices::getInstance()->getContentLanguage()->lc( $term ) ); |
| 199 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 200 | // The real type is still IDatabase, but IReplicaDatabase is used for safety. |
| 201 | '@phan-var IDatabase $dbr'; |
| 202 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 203 | $resultSet = $dbr->query( $this->getQuery( $filteredTerm, $fulltext ), __METHOD__ ); |
| 204 | |
| 205 | $total = null; |
| 206 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 207 | $totalResult = $dbr->query( $this->getCountQuery( $filteredTerm, $fulltext ), __METHOD__ ); |
| 208 | $row = $totalResult->fetchObject(); |
| 209 | if ( $row ) { |
| 210 | $total = intval( $row->c ); |
| 211 | } |
| 212 | $totalResult->free(); |
| 213 | |
| 214 | return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total ); |
| 215 | } |
| 216 | |
| 217 | /** |
| 218 | * Return a partial WHERE clause to limit the search to the given namespaces |
| 219 | * @return string |
| 220 | */ |
| 221 | private function queryNamespaces() { |
| 222 | if ( $this->namespaces === null ) { |
| 223 | return ''; # search all |
| 224 | } |
| 225 | if ( $this->namespaces === [] ) { |
| 226 | $namespaces = NS_MAIN; |
| 227 | } else { |
| 228 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 229 | $namespaces = $dbr->makeList( $this->namespaces ); |
| 230 | } |
| 231 | return 'AND page_namespace IN (' . $namespaces . ')'; |
| 232 | } |
| 233 | |
| 234 | /** |
| 235 | * Returns a query with limit for number of results set. |
| 236 | * @param string $sql |
| 237 | * @return string |
| 238 | */ |
| 239 | private function limitResult( $sql ) { |
| 240 | return $this->dbProvider->getReplicaDatabase()->limitResult( $sql, $this->limit, $this->offset ); |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * Construct the full SQL query to do the search. |
| 245 | * The guts shoulds be constructed in queryMain() |
| 246 | * @param string $filteredTerm |
| 247 | * @param bool $fulltext |
| 248 | * @return string |
| 249 | */ |
| 250 | private function getQuery( $filteredTerm, $fulltext ) { |
| 251 | return $this->limitResult( |
| 252 | $this->queryMain( $filteredTerm, $fulltext ) . ' ' . |
| 253 | $this->queryNamespaces() |
| 254 | ); |
| 255 | } |
| 256 | |
| 257 | /** |
| 258 | * Picks which field to index on, depending on what type of query. |
| 259 | * @param bool $fulltext |
| 260 | * @return string |
| 261 | */ |
| 262 | private function getIndexField( $fulltext ) { |
| 263 | return $fulltext ? 'si_text' : 'si_title'; |
| 264 | } |
| 265 | |
| 266 | /** |
| 267 | * Get the base part of the search query. |
| 268 | * |
| 269 | * @param string $filteredTerm |
| 270 | * @param bool $fulltext |
| 271 | * @return string |
| 272 | */ |
| 273 | private function queryMain( $filteredTerm, $fulltext ) { |
| 274 | $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 275 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 276 | $page = $dbr->tableName( 'page' ); |
| 277 | $searchindex = $dbr->tableName( 'searchindex' ); |
| 278 | return "SELECT $searchindex.rowid, page_namespace, page_title " . |
| 279 | "FROM $searchindex CROSS JOIN $page " . |
| 280 | "WHERE page_id=$searchindex.rowid AND $match"; |
| 281 | } |
| 282 | |
| 283 | private function getCountQuery( $filteredTerm, $fulltext ) { |
| 284 | $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 285 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 286 | $page = $dbr->tableName( 'page' ); |
| 287 | $searchindex = $dbr->tableName( 'searchindex' ); |
| 288 | return "SELECT COUNT(*) AS c " . |
| 289 | "FROM $searchindex CROSS JOIN $page " . |
| 290 | "WHERE page_id=$searchindex.rowid AND $match " . |
| 291 | $this->queryNamespaces(); |
| 292 | } |
| 293 | |
| 294 | /** |
| 295 | * Create or update the search index record for the given page. |
| 296 | * Title and text should be pre-processed. |
| 297 | * |
| 298 | * @param int $id |
| 299 | * @param string $title |
| 300 | * @param string $text |
| 301 | */ |
| 302 | public function update( $id, $title, $text ) { |
| 303 | if ( !$this->fulltextSearchSupported() ) { |
| 304 | return; |
| 305 | } |
| 306 | // @todo find a method to do it in a single request, |
| 307 | // couldn't do it so far due to typelessness of FTS3 tables. |
| 308 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 309 | $dbw->newDeleteQueryBuilder() |
| 310 | ->deleteFrom( 'searchindex' ) |
| 311 | ->where( [ 'rowid' => $id ] ) |
| 312 | ->caller( __METHOD__ )->execute(); |
| 313 | $dbw->newInsertQueryBuilder() |
| 314 | ->insertInto( 'searchindex' ) |
| 315 | ->row( [ 'rowid' => $id, 'si_title' => $title, 'si_text' => $text ] ) |
| 316 | ->caller( __METHOD__ )->execute(); |
| 317 | } |
| 318 | |
| 319 | /** |
| 320 | * Update a search index record's title only. |
| 321 | * Title should be pre-processed. |
| 322 | * |
| 323 | * @param int $id |
| 324 | * @param string $title |
| 325 | */ |
| 326 | public function updateTitle( $id, $title ) { |
| 327 | if ( !$this->fulltextSearchSupported() ) { |
| 328 | return; |
| 329 | } |
| 330 | |
| 331 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 332 | $dbw->newUpdateQueryBuilder() |
| 333 | ->update( 'searchindex' ) |
| 334 | ->set( [ 'si_title' => $title ] ) |
| 335 | ->where( [ 'rowid' => $id ] ) |
| 336 | ->caller( __METHOD__ )->execute(); |
| 337 | } |
| 338 | } |