Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
84.48% |
98 / 116 |
|
77.27% |
17 / 22 |
CRAP | |
0.00% |
0 / 1 |
| CategoryLinksTable | |
84.48% |
98 / 116 |
|
77.27% |
17 / 22 |
51.23 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| startUpdate | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setParserOutput | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
4 | |||
| getTableName | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getFromField | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getExistingFields | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getNewLinkIDs | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| fetchExistingLinks | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| getExistingLinks | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getSavedTimestamps | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getExistingLinkIDs | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| isExisting | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| isInNewSet | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| insertLink | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| deleteLink | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| needForcedLinkRefresh | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| makePageReferenceValue | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| makeTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| deduplicateLinkIds | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| finishUpdate | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
2.01 | |||
| invalidateCategories | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| updateCategoryCounts | |
58.33% |
14 / 24 |
|
0.00% |
0 / 1 |
12.63 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Deferred\LinksUpdate; |
| 4 | |
| 5 | use Collation; |
| 6 | use MediaWiki\DAO\WikiAwareEntity; |
| 7 | use MediaWiki\Languages\LanguageConverterFactory; |
| 8 | use MediaWiki\Page\PageReferenceValue; |
| 9 | use MediaWiki\Page\WikiPageFactory; |
| 10 | use MediaWiki\Parser\ParserOutput; |
| 11 | use MediaWiki\Parser\Sanitizer; |
| 12 | use MediaWiki\Title\NamespaceInfo; |
| 13 | use MediaWiki\Title\Title; |
| 14 | use PurgeJobUtils; |
| 15 | |
| 16 | /** |
| 17 | * categorylinks |
| 18 | * |
| 19 | * Link ID format: string[] |
| 20 | * - 0: Category name |
| 21 | * - 1: User-specified sort key (cl_sortkey_prefix) |
| 22 | * |
| 23 | * @since 1.38 |
| 24 | */ |
| 25 | class CategoryLinksTable extends TitleLinksTable { |
| 26 | /** |
| 27 | * @var array Associative array of new links, with the category name in the |
| 28 | * key. The value is a list consisting of the sort key prefix and the sort |
| 29 | * key. |
| 30 | */ |
| 31 | private $newLinks = []; |
| 32 | |
| 33 | /** |
| 34 | * @var array|null Associative array of existing links, or null if it has |
| 35 | * not been loaded yet |
| 36 | */ |
| 37 | private $existingLinks; |
| 38 | |
| 39 | /** |
| 40 | * @var array Associative array of saved timestamps, if there is a force |
| 41 | * refresh due to a page move |
| 42 | */ |
| 43 | private $savedTimestamps = null; |
| 44 | |
| 45 | /** @var \ILanguageConverter */ |
| 46 | private $languageConverter; |
| 47 | |
| 48 | /** @var \Collation */ |
| 49 | private $collation; |
| 50 | |
| 51 | /** @var string The collation name for cl_collation */ |
| 52 | private $collationName; |
| 53 | |
| 54 | /** @var string The table name */ |
| 55 | private $tableName = 'categorylinks'; |
| 56 | |
| 57 | /** @var bool */ |
| 58 | private $isTempTable; |
| 59 | |
| 60 | /** @var string The category type, which depends on the source page */ |
| 61 | private $categoryType; |
| 62 | |
| 63 | /** @var NamespaceInfo */ |
| 64 | private $namespaceInfo; |
| 65 | |
| 66 | /** @var WikiPageFactory */ |
| 67 | private $wikiPageFactory; |
| 68 | |
| 69 | /** |
| 70 | * @param LanguageConverterFactory $converterFactory |
| 71 | * @param NamespaceInfo $namespaceInfo |
| 72 | * @param WikiPageFactory $wikiPageFactory |
| 73 | * @param Collation $collation |
| 74 | * @param string $collationName |
| 75 | * @param string $tableName |
| 76 | * @param bool $isTempTable |
| 77 | */ |
| 78 | public function __construct( |
| 79 | LanguageConverterFactory $converterFactory, |
| 80 | NamespaceInfo $namespaceInfo, |
| 81 | WikiPageFactory $wikiPageFactory, |
| 82 | Collation $collation, |
| 83 | $collationName, |
| 84 | $tableName, |
| 85 | $isTempTable |
| 86 | ) { |
| 87 | $this->languageConverter = $converterFactory->getLanguageConverter(); |
| 88 | $this->namespaceInfo = $namespaceInfo; |
| 89 | $this->wikiPageFactory = $wikiPageFactory; |
| 90 | $this->collation = $collation; |
| 91 | $this->collationName = $collationName; |
| 92 | $this->tableName = $tableName; |
| 93 | $this->isTempTable = $isTempTable; |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * Cache the category type after the source page has been set |
| 98 | */ |
| 99 | public function startUpdate() { |
| 100 | $this->categoryType = $this->namespaceInfo |
| 101 | ->getCategoryLinkType( $this->getSourcePage()->getNamespace() ); |
| 102 | } |
| 103 | |
| 104 | public function setParserOutput( ParserOutput $parserOutput ) { |
| 105 | $this->newLinks = []; |
| 106 | $sourceTitle = Title::castFromPageIdentity( $this->getSourcePage() ); |
| 107 | $sortKeyInputs = []; |
| 108 | foreach ( $parserOutput->getCategoryNames() as $name ) { |
| 109 | $sortKey = $parserOutput->getCategorySortKey( $name ); |
| 110 | '@phan-var string $sortKey'; // sort key will never be null |
| 111 | |
| 112 | if ( $sortKey == '' ) { |
| 113 | $sortKey = $parserOutput->getPageProperty( "defaultsort" ) ?? ''; |
| 114 | } |
| 115 | $sortKey = $this->languageConverter->convertCategoryKey( $sortKey ); |
| 116 | |
| 117 | // Clean up the sort key, regardless of source |
| 118 | $sortKey = Sanitizer::decodeCharReferences( $sortKey ); |
| 119 | $sortKey = str_replace( "\n", '', $sortKey ); |
| 120 | |
| 121 | // If the sort key is longer then 255 bytes, it is truncated by DB, |
| 122 | // and then doesn't match when comparing existing vs current |
| 123 | // categories, causing T27254. |
| 124 | $sortKeyPrefix = mb_strcut( $sortKey, 0, 255 ); |
| 125 | |
| 126 | $targetTitle = Title::makeTitle( NS_CATEGORY, $name ); |
| 127 | $this->languageConverter->findVariantLink( $name, $targetTitle, true ); |
| 128 | // Ignore the returned text, DB key should be used for links (T328477). |
| 129 | $name = $targetTitle->getDBKey(); |
| 130 | |
| 131 | // Treat custom sort keys as a prefix, so that if multiple |
| 132 | // things are forced to sort as '*' or something, they'll |
| 133 | // sort properly in the category rather than in page_id |
| 134 | // order or such. |
| 135 | $sortKeyInputs[$name] = $sourceTitle->getCategorySortkey( $sortKeyPrefix ); |
| 136 | $this->newLinks[$name] = [ $sortKeyPrefix ]; |
| 137 | } |
| 138 | $sortKeys = $this->collation->getSortKeys( $sortKeyInputs ); |
| 139 | foreach ( $sortKeys as $name => $sortKey ) { |
| 140 | $this->newLinks[$name][1] = $sortKey; |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | protected function getTableName() { |
| 145 | return $this->tableName; |
| 146 | } |
| 147 | |
| 148 | protected function getFromField() { |
| 149 | return 'cl_from'; |
| 150 | } |
| 151 | |
| 152 | protected function getExistingFields() { |
| 153 | $fields = [ 'cl_to', 'cl_sortkey_prefix' ]; |
| 154 | if ( $this->needForcedLinkRefresh() ) { |
| 155 | $fields[] = 'cl_timestamp'; |
| 156 | } |
| 157 | return $fields; |
| 158 | } |
| 159 | |
| 160 | /** |
| 161 | * Get the new link IDs. The link ID is a list with the name in the first |
| 162 | * element and the sort key prefix in the second element. |
| 163 | * |
| 164 | * @return iterable<array> |
| 165 | */ |
| 166 | protected function getNewLinkIDs() { |
| 167 | foreach ( $this->newLinks as $name => [ $prefix, ] ) { |
| 168 | yield [ (string)$name, $prefix ]; |
| 169 | } |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * Get the existing links from the database |
| 174 | */ |
| 175 | private function fetchExistingLinks() { |
| 176 | $this->existingLinks = []; |
| 177 | $this->savedTimestamps = []; |
| 178 | $force = $this->needForcedLinkRefresh(); |
| 179 | foreach ( $this->fetchExistingRows() as $row ) { |
| 180 | $this->existingLinks[$row->cl_to] = $row->cl_sortkey_prefix; |
| 181 | if ( $force ) { |
| 182 | $this->savedTimestamps[$row->cl_to] = $row->cl_timestamp; |
| 183 | } |
| 184 | } |
| 185 | } |
| 186 | |
| 187 | /** |
| 188 | * Get the existing links as an associative array, with the category name |
| 189 | * in the key and the sort key prefix in the value. |
| 190 | * |
| 191 | * @return array |
| 192 | */ |
| 193 | private function getExistingLinks() { |
| 194 | if ( $this->existingLinks === null ) { |
| 195 | $this->fetchExistingLinks(); |
| 196 | } |
| 197 | return $this->existingLinks; |
| 198 | } |
| 199 | |
| 200 | private function getSavedTimestamps() { |
| 201 | if ( $this->savedTimestamps === null ) { |
| 202 | $this->fetchExistingLinks(); |
| 203 | } |
| 204 | return $this->savedTimestamps; |
| 205 | } |
| 206 | |
| 207 | /** |
| 208 | * @return \Generator |
| 209 | */ |
| 210 | protected function getExistingLinkIDs() { |
| 211 | foreach ( $this->getExistingLinks() as $name => $sortkey ) { |
| 212 | yield [ (string)$name, $sortkey ]; |
| 213 | } |
| 214 | } |
| 215 | |
| 216 | protected function isExisting( $linkId ) { |
| 217 | $links = $this->getExistingLinks(); |
| 218 | [ $name, $prefix ] = $linkId; |
| 219 | return \array_key_exists( $name, $links ) && $links[$name] === $prefix; |
| 220 | } |
| 221 | |
| 222 | protected function isInNewSet( $linkId ) { |
| 223 | [ $name, $prefix ] = $linkId; |
| 224 | return \array_key_exists( $name, $this->newLinks ) |
| 225 | && $this->newLinks[$name][0] === $prefix; |
| 226 | } |
| 227 | |
| 228 | protected function insertLink( $linkId ) { |
| 229 | [ $name, $prefix ] = $linkId; |
| 230 | $sortKey = $this->newLinks[$name][1]; |
| 231 | $savedTimestamps = $this->getSavedTimestamps(); |
| 232 | |
| 233 | // Preserve cl_timestamp in the case of a forced refresh |
| 234 | $timestamp = $this->getDB()->timestamp( $savedTimestamps[$name] ?? 0 ); |
| 235 | |
| 236 | $this->insertRow( [ |
| 237 | 'cl_to' => $name, |
| 238 | 'cl_sortkey' => $sortKey, |
| 239 | 'cl_timestamp' => $timestamp, |
| 240 | 'cl_sortkey_prefix' => $prefix, |
| 241 | 'cl_collation' => $this->collationName, |
| 242 | 'cl_type' => $this->categoryType, |
| 243 | ] ); |
| 244 | } |
| 245 | |
| 246 | protected function deleteLink( $linkId ) { |
| 247 | $this->deleteRow( [ 'cl_to' => $linkId[0] ] ); |
| 248 | } |
| 249 | |
| 250 | protected function needForcedLinkRefresh() { |
| 251 | // cl_sortkey and possibly cl_type will change if it is a page move |
| 252 | return $this->isMove(); |
| 253 | } |
| 254 | |
| 255 | protected function makePageReferenceValue( $linkId ): PageReferenceValue { |
| 256 | return new PageReferenceValue( NS_CATEGORY, $linkId[0], WikiAwareEntity::LOCAL ); |
| 257 | } |
| 258 | |
| 259 | protected function makeTitle( $linkId ): Title { |
| 260 | return Title::makeTitle( NS_CATEGORY, $linkId[0] ); |
| 261 | } |
| 262 | |
| 263 | protected function deduplicateLinkIds( $linkIds ) { |
| 264 | $seen = []; |
| 265 | foreach ( $linkIds as $linkId ) { |
| 266 | if ( !\array_key_exists( $linkId[0], $seen ) ) { |
| 267 | $seen[$linkId[0]] = true; |
| 268 | yield $linkId; |
| 269 | } |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | protected function finishUpdate() { |
| 274 | if ( $this->isTempTable ) { |
| 275 | // Don't do invalidations for temporary collations |
| 276 | return; |
| 277 | } |
| 278 | |
| 279 | // A update of sortkey on move is detected as insert + delete, |
| 280 | // but the categories does not need to update the counters or invalidate caches |
| 281 | $allInsertedLinks = array_column( $this->insertedLinks, 0 ); |
| 282 | $allDeletedLinks = array_column( $this->deletedLinks, 0 ); |
| 283 | $insertedLinks = array_diff( $allInsertedLinks, $allDeletedLinks ); |
| 284 | $deletedLinks = array_diff( $allDeletedLinks, $allInsertedLinks ); |
| 285 | |
| 286 | $this->invalidateCategories( $insertedLinks, $deletedLinks ); |
| 287 | $this->updateCategoryCounts( $insertedLinks, $deletedLinks ); |
| 288 | } |
| 289 | |
| 290 | private function invalidateCategories( array $insertedLinks, array $deletedLinks ) { |
| 291 | $changedCategoryNames = array_merge( |
| 292 | $insertedLinks, |
| 293 | $deletedLinks |
| 294 | ); |
| 295 | PurgeJobUtils::invalidatePages( |
| 296 | $this->getDB(), NS_CATEGORY, $changedCategoryNames ); |
| 297 | } |
| 298 | |
| 299 | /** |
| 300 | * Update all the appropriate counts in the category table. |
| 301 | * @param array $insertedLinks |
| 302 | * @param array $deletedLinks |
| 303 | */ |
| 304 | private function updateCategoryCounts( array $insertedLinks, array $deletedLinks ) { |
| 305 | if ( !$insertedLinks && !$deletedLinks ) { |
| 306 | return; |
| 307 | } |
| 308 | |
| 309 | $domainId = $this->getDB()->getDomainID(); |
| 310 | $wp = $this->wikiPageFactory->newFromTitle( $this->getSourcePage() ); |
| 311 | $lbf = $this->getLBFactory(); |
| 312 | $size = $this->getBatchSize(); |
| 313 | // T163801: try to release any row locks to reduce contention |
| 314 | $lbf->commitAndWaitForReplication( __METHOD__, $this->getTransactionTicket() ); |
| 315 | |
| 316 | if ( count( $insertedLinks ) + count( $deletedLinks ) < $size ) { |
| 317 | $wp->updateCategoryCounts( |
| 318 | $insertedLinks, |
| 319 | $deletedLinks, |
| 320 | $this->getSourcePageId() |
| 321 | ); |
| 322 | $lbf->commitAndWaitForReplication( __METHOD__, $this->getTransactionTicket() ); |
| 323 | } else { |
| 324 | $addedChunks = array_chunk( $insertedLinks, $size ); |
| 325 | foreach ( $addedChunks as $chunk ) { |
| 326 | $wp->updateCategoryCounts( $chunk, [], $this->getSourcePageId() ); |
| 327 | if ( count( $addedChunks ) > 1 ) { |
| 328 | $lbf->commitAndWaitForReplication( __METHOD__, $this->getTransactionTicket() ); |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | $deletedChunks = array_chunk( $deletedLinks, $size ); |
| 333 | foreach ( $deletedChunks as $chunk ) { |
| 334 | $wp->updateCategoryCounts( [], $chunk, $this->getSourcePageId() ); |
| 335 | if ( count( $deletedChunks ) > 1 ) { |
| 336 | $lbf->commitAndWaitForReplication( __METHOD__, $this->getTransactionTicket() ); |
| 337 | } |
| 338 | } |
| 339 | |
| 340 | } |
| 341 | } |
| 342 | } |