Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
91.77% |
145 / 158 |
|
66.67% |
8 / 12 |
CRAP | |
0.00% |
0 / 1 |
| CognateStore | |
91.77% |
145 / 158 |
|
66.67% |
8 / 12 |
28.44 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| insertPage | |
96.97% |
32 / 33 |
|
0.00% |
0 / 1 |
5 | |||
| deletePage | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
2 | |||
| selectLinkDetailsForPage | |
100.00% |
25 / 25 |
|
100.00% |
1 / 1 |
2 | |||
| selectSitesForPage | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| insertPages | |
96.15% |
25 / 26 |
|
0.00% |
0 / 1 |
5 | |||
| buildRows | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| insertSites | |
94.44% |
17 / 18 |
|
0.00% |
0 / 1 |
5.00 | |||
| deletePagesForSite | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
12 | |||
| getStringHash | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNormalizedStringHash | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| throwReadOnlyException | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Cognate; |
| 4 | |
| 5 | use MediaWiki\Linker\LinkTarget; |
| 6 | use MediaWiki\Title\TitleValue; |
| 7 | use RuntimeException; |
| 8 | use Wikimedia\Rdbms\DBReadOnlyError; |
| 9 | use Wikimedia\Rdbms\IConnectionProvider; |
| 10 | |
| 11 | /** |
| 12 | * Database access for the Cognate tables. |
| 13 | * |
| 14 | * This class should generally not be accessed directly but instead via CognateRepo which contains |
| 15 | * extra business logic such as logging, stats and cache purges. |
| 16 | * |
| 17 | * @license GPL-2.0-or-later |
| 18 | * @author Gabriel Birke < gabriel.birke@wikimedia.de > |
| 19 | * @author Addshore |
| 20 | */ |
| 21 | class CognateStore { |
| 22 | |
| 23 | /** |
| 24 | * @var IConnectionProvider |
| 25 | */ |
| 26 | private $connectionProvider; |
| 27 | |
| 28 | /** |
| 29 | * @var StringNormalizer |
| 30 | */ |
| 31 | private $stringNormalizer; |
| 32 | |
| 33 | /** |
| 34 | * @var StringHasher |
| 35 | */ |
| 36 | private $stringHasher; |
| 37 | |
| 38 | /** |
| 39 | * @var bool |
| 40 | */ |
| 41 | private $readOnly; |
| 42 | |
| 43 | public const PAGES_TABLE_NAME = 'cognate_pages'; |
| 44 | public const SITES_TABLE_NAME = 'cognate_sites'; |
| 45 | public const TITLES_TABLE_NAME = 'cognate_titles'; |
| 46 | |
| 47 | /** |
| 48 | * @param IConnectionProvider $connectionProvider |
| 49 | * @param StringNormalizer $stringNormalizer |
| 50 | * @param StringHasher $stringHasher |
| 51 | * @param bool $readOnly Is Cognate in readonly mode? |
| 52 | */ |
| 53 | public function __construct( |
| 54 | IConnectionProvider $connectionProvider, |
| 55 | StringNormalizer $stringNormalizer, |
| 56 | StringHasher $stringHasher, |
| 57 | $readOnly |
| 58 | ) { |
| 59 | $this->connectionProvider = $connectionProvider; |
| 60 | $this->stringNormalizer = $stringNormalizer; |
| 61 | $this->stringHasher = $stringHasher; |
| 62 | $this->readOnly = $readOnly; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * Adds a page to the database. As well as adding the data to the pages table this also |
| 67 | * includes adding the data to the titles table where needed. |
| 68 | * |
| 69 | * @param string $dbName The dbName for the site |
| 70 | * @param LinkTarget $linkTarget |
| 71 | * |
| 72 | * @return bool|int number of inserts run on success, false if there was a key conflict |
| 73 | * @throws DBReadOnlyError |
| 74 | */ |
| 75 | public function insertPage( $dbName, LinkTarget $linkTarget ) { |
| 76 | if ( $this->readOnly ) { |
| 77 | $this->throwReadOnlyException(); |
| 78 | } |
| 79 | |
| 80 | $dbr = $this->connectionProvider->getReplicaDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 81 | |
| 82 | [ $pagesToInsert, $titlesToInsert ] = $this->buildRows( |
| 83 | $linkTarget, |
| 84 | $dbName |
| 85 | ); |
| 86 | |
| 87 | $row = $dbr->newSelectQueryBuilder() |
| 88 | ->select( 'cgti_raw' ) |
| 89 | ->from( self::TITLES_TABLE_NAME ) |
| 90 | ->where( [ 'cgti_raw_key' => $this->getStringHash( $linkTarget->getDBkey() ) ] ) |
| 91 | ->caller( __METHOD__ ) |
| 92 | ->fetchRow(); |
| 93 | |
| 94 | if ( $row && $row->cgti_raw !== $linkTarget->getDBkey() ) { |
| 95 | return false; |
| 96 | } |
| 97 | |
| 98 | $insertQueryCounter = 0; |
| 99 | |
| 100 | $dbw = $this->connectionProvider->getPrimaryDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 101 | if ( !$row ) { |
| 102 | $dbw->newInsertQueryBuilder() |
| 103 | ->insertInto( self::TITLES_TABLE_NAME ) |
| 104 | ->ignore() |
| 105 | ->rows( $titlesToInsert ) |
| 106 | ->caller( __METHOD__ ) |
| 107 | ->execute(); |
| 108 | $insertQueryCounter++; |
| 109 | } |
| 110 | |
| 111 | $dbw->newInsertQueryBuilder() |
| 112 | ->insertInto( self::PAGES_TABLE_NAME ) |
| 113 | ->ignore() |
| 114 | ->rows( $pagesToInsert ) |
| 115 | ->caller( __METHOD__ ) |
| 116 | ->execute(); |
| 117 | $insertQueryCounter++; |
| 118 | |
| 119 | return $insertQueryCounter; |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * Note: this method will not remove any relevant entries from the titles table |
| 124 | * |
| 125 | * @param string $dbName The dbName for the site |
| 126 | * @param LinkTarget $linkTarget |
| 127 | * |
| 128 | * @return bool |
| 129 | * @throws DBReadOnlyError |
| 130 | */ |
| 131 | public function deletePage( $dbName, LinkTarget $linkTarget ) { |
| 132 | if ( $this->readOnly ) { |
| 133 | $this->throwReadOnlyException(); |
| 134 | } |
| 135 | |
| 136 | $pageData = [ |
| 137 | 'cgpa_site' => $this->getStringHash( $dbName ), |
| 138 | 'cgpa_title' => $this->getStringHash( $linkTarget->getDBkey() ), |
| 139 | 'cgpa_namespace' => $linkTarget->getNamespace(), |
| 140 | ]; |
| 141 | $dbw = $this->connectionProvider->getPrimaryDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 142 | $dbw->newDeleteQueryBuilder() |
| 143 | ->deleteFrom( self::PAGES_TABLE_NAME ) |
| 144 | ->where( $pageData ) |
| 145 | ->caller( __METHOD__ ) |
| 146 | ->execute(); |
| 147 | |
| 148 | return true; |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * @param string $dbName The dbName of the site being linked from |
| 153 | * @param LinkTarget $linkTarget of the page the links should be retrieved for |
| 154 | * |
| 155 | * @return array[] details used to create interwiki links. Each array will look like: |
| 156 | * [ 'interwiki' => 'en', 'namespaceID' => 0, 'title' => 'Berlin' ] |
| 157 | */ |
| 158 | public function selectLinkDetailsForPage( $dbName, LinkTarget $linkTarget ) { |
| 159 | $dbr = $this->connectionProvider->getReplicaDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 160 | $result = $dbr->newSelectQueryBuilder() |
| 161 | ->select( [ |
| 162 | 'cgsi_interwiki', |
| 163 | 'cgpa_namespace', |
| 164 | 'cgti_raw', |
| 165 | ] ) |
| 166 | ->from( self::TITLES_TABLE_NAME ) |
| 167 | ->join( self::PAGES_TABLE_NAME, null, 'cgti_raw_key = cgpa_title' ) |
| 168 | ->join( self::SITES_TABLE_NAME, null, 'cgpa_site = cgsi_key' ) |
| 169 | ->where( [ |
| 170 | $dbr->expr( 'cgsi_dbname', '!=', $dbName ), |
| 171 | 'cgti_normalized_key' => $this->getNormalizedStringHash( $linkTarget->getDBkey() ), |
| 172 | 'cgpa_namespace' => $linkTarget->getNamespace(), |
| 173 | ] ) |
| 174 | ->caller( __METHOD__ ) |
| 175 | ->fetchResultSet(); |
| 176 | |
| 177 | $linkDetails = []; |
| 178 | foreach ( $result as $row ) { |
| 179 | $linkDetails[] = [ |
| 180 | 'interwiki' => $row->cgsi_interwiki, |
| 181 | 'namespaceID' => intval( $row->cgpa_namespace ), |
| 182 | 'title' => $row->cgti_raw, |
| 183 | ]; |
| 184 | } |
| 185 | |
| 186 | return $linkDetails; |
| 187 | } |
| 188 | |
| 189 | /** |
| 190 | * @param LinkTarget $linkTarget |
| 191 | * |
| 192 | * @return string[] array of dbnames |
| 193 | */ |
| 194 | public function selectSitesForPage( LinkTarget $linkTarget ) { |
| 195 | $dbr = $this->connectionProvider->getReplicaDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 196 | return $dbr->newSelectQueryBuilder() |
| 197 | ->select( 'cgsi_dbname' ) |
| 198 | ->from( self::TITLES_TABLE_NAME ) |
| 199 | ->join( self::PAGES_TABLE_NAME, null, 'cgti_raw_key = cgpa_title' ) |
| 200 | ->join( self::SITES_TABLE_NAME, null, 'cgpa_site = cgsi_key' ) |
| 201 | ->where( [ |
| 202 | 'cgti_normalized_key' => $this->getNormalizedStringHash( $linkTarget->getDBkey() ), |
| 203 | 'cgpa_namespace' => $linkTarget->getNamespace(), |
| 204 | ] ) |
| 205 | ->caller( __METHOD__ ) |
| 206 | ->fetchFieldValues(); |
| 207 | } |
| 208 | |
| 209 | /** |
| 210 | * Adds pages to the database. As well as adding the data to the pages table this also |
| 211 | * includes adding the data to the titles table where needed. |
| 212 | * |
| 213 | * @note Errors during insertion are totally ignored by this method. If there were duplicate |
| 214 | * keys in the DB then you will not find out about them here. |
| 215 | * |
| 216 | * @param array[] $pageDetailsArray where each element contains the keys 'site', 'namespace', |
| 217 | * and 'title', e.g. [ [ 'site' => 'enwiktionary', 'namespace' => 0, 'title' => 'Berlin' ] ]. |
| 218 | * |
| 219 | * @throws RuntimeException |
| 220 | */ |
| 221 | public function insertPages( array $pageDetailsArray ) { |
| 222 | if ( !defined( 'RUN_MAINTENANCE_IF_MAIN' ) && !defined( 'MW_PHPUNIT_TEST' ) ) { |
| 223 | throw new RuntimeException( __METHOD__ . ' can only be used for maintenance or tests.' ); |
| 224 | } |
| 225 | |
| 226 | if ( !$pageDetailsArray ) { |
| 227 | return; |
| 228 | } |
| 229 | |
| 230 | $pagesToInsert = []; |
| 231 | $titlesToInsert = []; |
| 232 | foreach ( $pageDetailsArray as $pageDetails ) { |
| 233 | $this->buildRows( |
| 234 | new TitleValue( $pageDetails['namespace'], $pageDetails['title'] ), |
| 235 | $pageDetails['site'], |
| 236 | $pagesToInsert, |
| 237 | $titlesToInsert |
| 238 | ); |
| 239 | } |
| 240 | |
| 241 | $dbw = $this->connectionProvider->getPrimaryDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 242 | $dbw->newInsertQueryBuilder() |
| 243 | ->insertInto( self::TITLES_TABLE_NAME ) |
| 244 | ->ignore() |
| 245 | ->rows( $titlesToInsert ) |
| 246 | ->caller( __METHOD__ ) |
| 247 | ->execute(); |
| 248 | |
| 249 | $dbw->newInsertQueryBuilder() |
| 250 | ->insertInto( self::PAGES_TABLE_NAME ) |
| 251 | ->ignore() |
| 252 | ->rows( $pagesToInsert ) |
| 253 | ->caller( __METHOD__ ) |
| 254 | ->execute(); |
| 255 | } |
| 256 | |
| 257 | /** |
| 258 | * @param LinkTarget $linkTarget |
| 259 | * @param string $site |
| 260 | * @param array[] &$pagesToInsert |
| 261 | * @param array[] &$titlesToInsert |
| 262 | * |
| 263 | * @return array[] 0 => $pagesToInsert, 1 => $titleToInsert |
| 264 | */ |
| 265 | private function buildRows( |
| 266 | LinkTarget $linkTarget, |
| 267 | $site, |
| 268 | array &$pagesToInsert = [], |
| 269 | array &$titlesToInsert = [] |
| 270 | ) { |
| 271 | $pagesToInsert[] = [ |
| 272 | 'cgpa_site' => $this->getStringHash( $site ), |
| 273 | 'cgpa_namespace' => $linkTarget->getNamespace(), |
| 274 | 'cgpa_title' => $this->getStringHash( $linkTarget->getDBkey() ), |
| 275 | ]; |
| 276 | $titlesToInsert[] = [ |
| 277 | 'cgti_raw' => $linkTarget->getDBkey(), |
| 278 | 'cgti_raw_key' => $this->getStringHash( $linkTarget->getDBkey() ), |
| 279 | 'cgti_normalized_key' => $this->getNormalizedStringHash( $linkTarget->getDBkey() ), |
| 280 | ]; |
| 281 | |
| 282 | return [ $pagesToInsert, $titlesToInsert ]; |
| 283 | } |
| 284 | |
| 285 | /** |
| 286 | * @param string[] $sites keys of site dbname => values of interwiki prefix |
| 287 | * e.g. 'enwiktionary' => 'en' |
| 288 | * |
| 289 | * @throws RuntimeException |
| 290 | */ |
| 291 | public function insertSites( array $sites ) { |
| 292 | if ( !defined( 'RUN_MAINTENANCE_IF_MAIN' ) && !defined( 'MW_PHPUNIT_TEST' ) ) { |
| 293 | throw new RuntimeException( __METHOD__ . ' can only be used for maintenance or tests.' ); |
| 294 | } |
| 295 | |
| 296 | if ( !$sites ) { |
| 297 | return; |
| 298 | } |
| 299 | |
| 300 | $toInsert = []; |
| 301 | foreach ( $sites as $dbname => $interwikiPrefix ) { |
| 302 | $toInsert[] = [ |
| 303 | 'cgsi_key' => $this->getStringHash( $dbname ), |
| 304 | 'cgsi_dbname' => $dbname, |
| 305 | 'cgsi_interwiki' => $interwikiPrefix, |
| 306 | ]; |
| 307 | } |
| 308 | |
| 309 | $dbw = $this->connectionProvider->getPrimaryDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 310 | $dbw->newInsertQueryBuilder() |
| 311 | ->insertInto( 'cognate_sites' ) |
| 312 | ->ignore() |
| 313 | ->rows( $toInsert ) |
| 314 | ->caller( __METHOD__ ) |
| 315 | ->execute(); |
| 316 | } |
| 317 | |
| 318 | /** |
| 319 | * Delete all entries from the cognate_pages table for the given site. |
| 320 | * |
| 321 | * @param string $dbName The dbname of the site to delete pages for. |
| 322 | * |
| 323 | * @throws RuntimeException if not run in a maintenance or test scope |
| 324 | */ |
| 325 | public function deletePagesForSite( $dbName ) { |
| 326 | if ( !defined( 'RUN_MAINTENANCE_IF_MAIN' ) && !defined( 'MW_PHPUNIT_TEST' ) ) { |
| 327 | throw new RuntimeException( __METHOD__ . ' can only be used for maintenance or tests.' ); |
| 328 | } |
| 329 | |
| 330 | $dbw = $this->connectionProvider->getPrimaryDatabase( CognateServices::VIRTUAL_DOMAIN ); |
| 331 | $dbw->newDeleteQueryBuilder() |
| 332 | ->deleteFrom( 'cognate_pages' ) |
| 333 | ->where( [ |
| 334 | 'cgpa_site' => $this->getStringHash( $dbName ), |
| 335 | ] ) |
| 336 | ->caller( __METHOD__ ) |
| 337 | ->execute(); |
| 338 | } |
| 339 | |
| 340 | /** |
| 341 | * @param string $string |
| 342 | * |
| 343 | * @return int |
| 344 | */ |
| 345 | private function getStringHash( $string ) { |
| 346 | return $this->stringHasher->hash( $string ); |
| 347 | } |
| 348 | |
| 349 | /** |
| 350 | * @param string $string |
| 351 | * |
| 352 | * @return int |
| 353 | */ |
| 354 | private function getNormalizedStringHash( $string ) { |
| 355 | return $this->stringHasher->hash( |
| 356 | $this->stringNormalizer->normalize( $string ) |
| 357 | ); |
| 358 | } |
| 359 | |
| 360 | /** |
| 361 | * @throws DBReadOnlyError |
| 362 | * @return never |
| 363 | */ |
| 364 | private function throwReadOnlyException() { |
| 365 | throw new DBReadOnlyError( null, 'Cognate is in Read Only mode' ); |
| 366 | } |
| 367 | |
| 368 | } |