Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
89.83% |
1298 / 1445 |
|
76.39% |
55 / 72 |
CRAP | |
0.00% |
0 / 1 |
| ZObjectStore | |
89.83% |
1298 / 1445 |
|
76.39% |
55 / 72 |
302.15 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNextAvailableZid | |
97.50% |
39 / 40 |
|
0.00% |
0 / 1 |
4 | |||
| getRevisionById | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| fetchZObjectByTitle | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| fetchZObject | |
63.16% |
12 / 19 |
|
0.00% |
0 / 1 |
7.80 | |||
| fetchBatchZObjects | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
3 | |||
| pushZObject | |
32.56% |
14 / 43 |
|
0.00% |
0 / 1 |
17.04 | |||
| createNewZObject | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| updateZObject | |
61.80% |
55 / 89 |
|
0.00% |
0 / 1 |
27.54 | |||
| updateZObjectAsSystemUser | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZObjectLabelsByZid | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZObjectLabelConflictsByZid | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| findZObjectLabelConflicts | |
100.00% |
27 / 27 |
|
100.00% |
1 / 1 |
4 | |||
| insertZObjectLabels | |
94.44% |
17 / 18 |
|
0.00% |
0 / 1 |
3.00 | |||
| insertZLanguageToLanguagesCache | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| insertZObjectLabelConflicts | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
3.00 | |||
| insertZObjectAliases | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
20 | |||
| synchroniseZObjectLabels | |
100.00% |
72 / 72 |
|
100.00% |
1 / 1 |
11 | |||
| synchroniseZObjectLabelConflicts | |
100.00% |
54 / 54 |
|
100.00% |
1 / 1 |
9 | |||
| makeLabelRowKey | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| makeConflictRowKey | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fetchZidsOfType | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| getCountOfTypeInstances | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| fetchAllZids | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
1 | |||
| fetchAllZLanguageObjects | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| fetchAllZLanguageCodes | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| findCodesFromZLanguage | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| fetchAllZLanguagesWithLabels | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| fetchAllInstancedTypesWithLabels | |
94.12% |
16 / 17 |
|
0.00% |
0 / 1 |
3.00 | |||
| fetchAllInstancedTypes | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| getReturnTypeQuery | |
100.00% |
26 / 26 |
|
100.00% |
1 / 1 |
1 | |||
| getPreferredLabelsQuery | |
100.00% |
39 / 39 |
|
100.00% |
1 / 1 |
4 | |||
| getStringMatchCondition | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
4 | |||
| searchFunctions | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
30 | |||
| getTestStatusQuery | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
1 | |||
| getTestStatusForLatestRevisions | |
100.00% |
56 / 56 |
|
100.00% |
1 / 1 |
2 | |||
| findZLanguageFromCode | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| findZLanguagesFromCodes | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
2 | |||
| searchZObjectLabels | |
100.00% |
56 / 56 |
|
100.00% |
1 / 1 |
11 | |||
| fetchZObjectLabels | |
100.00% |
34 / 34 |
|
100.00% |
1 / 1 |
9 | |||
| fetchZObjectLabel | |
96.55% |
28 / 29 |
|
0.00% |
0 / 1 |
7 | |||
| fetchZFunctionReturnType | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| findFirstZImplementationFunction | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
2 | |||
| findReferencedZObjectsByZFunctionIdAsList | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| findReferencedZObjectsByZFunctionId | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
2.00 | |||
| insertZFunctionReference | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZFunctionReference | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| synchroniseZFunctionReference | |
100.00% |
41 / 41 |
|
100.00% |
1 / 1 |
10 | |||
| findRelatedZObjectsByKeyAsList | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| findRelatedZObjectsByKey | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
2.00 | |||
| findFunctionsReferencingZObjectByKey | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| findFunctionsByRenderableIO | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| functionsByRenderableIOQuery | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
1 | |||
| getRenderableIOQuery | |
100.00% |
60 / 60 |
|
100.00% |
1 / 1 |
3 | |||
| findFunctionsByIOTypes | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| functionsByIOTypesQuery | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
4 | |||
| newIOTypeQuery | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
2 | |||
| insertRelatedZObjects | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
2 | |||
| deleteRelatedZObjects | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
6.01 | |||
| synchroniseRelatedZObjects | |
100.00% |
63 / 63 |
|
100.00% |
1 / 1 |
10 | |||
| makeRelatedRowKey | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| findZTesterResult | |
91.67% |
55 / 60 |
|
0.00% |
0 / 1 |
11.07 | |||
| insertZTesterResult | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
3 | |||
| isLatestRevisionTuple | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| isLatestRevision | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
2 | |||
| deleteZFunctionFromZTesterResultsCache | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZImplementationFromZTesterResultsCache | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZTesterFromZTesterResultsCache | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| deleteZLanguageFromLanguagesCache | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| synchroniseZLanguageCodes | |
100.00% |
32 / 32 |
|
100.00% |
1 / 1 |
8 | |||
| removeFunctionReferenceIfImplementationOrTester | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
5.01 | |||
| removeReferenceFromFunction | |
88.89% |
40 / 45 |
|
0.00% |
0 / 1 |
6.05 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * WikiLambda Data Access Object service |
| 4 | * |
| 5 | * @file |
| 6 | * @ingroup Extensions |
| 7 | * @copyright 2020– Abstract Wikipedia team; see AUTHORS.txt |
| 8 | * @license MIT |
| 9 | */ |
| 10 | |
| 11 | namespace MediaWiki\Extension\WikiLambda; |
| 12 | |
| 13 | use Exception; |
| 14 | use MediaWiki\Content\ContentSerializationException; |
| 15 | use MediaWiki\Context\RequestContext; |
| 16 | use MediaWiki\Extension\WikiLambda\Cache\MemcachedWrapper; |
| 17 | use MediaWiki\Extension\WikiLambda\Registry\ZErrorTypeRegistry; |
| 18 | use MediaWiki\Extension\WikiLambda\Registry\ZLangRegistry; |
| 19 | use MediaWiki\Extension\WikiLambda\Registry\ZTypeRegistry; |
| 20 | use MediaWiki\Extension\WikiLambda\ZObjectContent\ZObjectContent; |
| 21 | use MediaWiki\Extension\WikiLambda\ZObjectContent\ZObjectContentHandler; |
| 22 | use MediaWiki\Extension\WikiLambda\ZObjectContent\ZObjectPage; |
| 23 | use MediaWiki\Extension\WikiLambda\ZObjects\ZPersistentObject; |
| 24 | use MediaWiki\Extension\WikiLambda\ZObjects\ZReference; |
| 25 | use MediaWiki\Extension\WikiLambda\ZObjects\ZResponseEnvelope; |
| 26 | use MediaWiki\Extension\WikiLambda\ZObjects\ZTypedList; |
| 27 | use MediaWiki\Language\MessageLocalizer; |
| 28 | use MediaWiki\MediaWikiServices; |
| 29 | use MediaWiki\Page\WikiPage; |
| 30 | use MediaWiki\Page\WikiPageFactory; |
| 31 | use MediaWiki\Permissions\Authority; |
| 32 | use MediaWiki\Revision\RevisionRecord; |
| 33 | use MediaWiki\Revision\RevisionStore; |
| 34 | use MediaWiki\Revision\SlotRecord; |
| 35 | use MediaWiki\Title\Title; |
| 36 | use MediaWiki\Title\TitleArrayFromResult; |
| 37 | use MediaWiki\Title\TitleFactory; |
| 38 | use MediaWiki\User\User; |
| 39 | use MediaWiki\User\UserGroupManager; |
| 40 | use Psr\Log\LoggerInterface; |
| 41 | use Wikimedia\Rdbms\FakeResultWrapper; |
| 42 | use Wikimedia\Rdbms\IConnectionProvider; |
| 43 | use Wikimedia\Rdbms\IExpression; |
| 44 | use Wikimedia\Rdbms\IReadableDatabase; |
| 45 | use Wikimedia\Rdbms\IResultWrapper; |
| 46 | use Wikimedia\Rdbms\LikeValue; |
| 47 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 48 | use Wikimedia\Rdbms\Subquery; |
| 49 | |
| 50 | class ZObjectStore { |
| 51 | |
| 52 | private MemcachedWrapper $zObjectCache; |
| 53 | |
| 54 | // NOTE: This constant hard-codes user-provided content as starting from 'Z10000'. Now that the |
| 55 | // wiki has launched, change it will have unpredicatable effects. |
| 56 | private const PREDEFINED_TOP_LIMIT = 'Z9999'; |
| 57 | |
| 58 | public const ZOBJECT_CACHE_KEY_PREFIX = 'WikiLambdaObjectStorage'; |
| 59 | public const TESTER_RESULT_CACHE_WRITE_INSERTED = 'inserted'; |
| 60 | public const TESTER_RESULT_CACHE_WRITE_STALE = 'stale'; |
| 61 | public const TESTER_RESULT_CACHE_WRITE_FAILED = 'failed'; |
| 62 | |
| 63 | /** |
| 64 | * @param IConnectionProvider $dbProvider |
| 65 | * @param TitleFactory $titleFactory |
| 66 | * @param WikiPageFactory $wikiPageFactory |
| 67 | * @param RevisionStore $revisionStore |
| 68 | * @param UserGroupManager $userGroupManager |
| 69 | * @param LoggerInterface $logger |
| 70 | */ |
| 71 | public function __construct( |
| 72 | private readonly IConnectionProvider $dbProvider, |
| 73 | private readonly TitleFactory $titleFactory, |
| 74 | private readonly WikiPageFactory $wikiPageFactory, |
| 75 | private readonly RevisionStore $revisionStore, |
| 76 | private readonly UserGroupManager $userGroupManager, |
| 77 | private readonly LoggerInterface $logger |
| 78 | ) { |
| 79 | // Non-injected items |
| 80 | $this->zObjectCache = WikiLambdaServices::getMemcachedWrapper(); |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * Find next available ZID in the database to create a new ZObject |
| 85 | * |
| 86 | * @return string Next available ZID |
| 87 | */ |
| 88 | public function getNextAvailableZid(): string { |
| 89 | // Intentionally use DB_PRIMARY as we need the latest data here. |
| 90 | $dbr = $this->dbProvider->getPrimaryDatabase(); |
| 91 | |
| 92 | $conditions = [ |
| 93 | 'page_namespace' => NS_MAIN, |
| 94 | 'LENGTH( page_title ) > 5', |
| 95 | $dbr->expr( 'page_title', IExpression::LIKE, new LikeValue( 'Z', $dbr->anyString() ) ) |
| 96 | ]; |
| 97 | |
| 98 | // First try, get zid of highest page_id, skipping predefined |
| 99 | $latestZid = $dbr->newSelectQueryBuilder() |
| 100 | ->select( 'page_title' ) |
| 101 | ->from( 'page' ) |
| 102 | ->where( $conditions ) |
| 103 | ->orderBy( 'page_id', SelectQueryBuilder::SORT_DESC ) |
| 104 | ->caller( __METHOD__ ) |
| 105 | ->fetchField(); |
| 106 | |
| 107 | if ( $latestZid === false ) { |
| 108 | $latestZid = self::PREDEFINED_TOP_LIMIT; |
| 109 | } |
| 110 | $targetZid = 'Z' . ( max( intval( substr( $latestZid, 1 ) ) + 1, 10000 ) ); |
| 111 | |
| 112 | // Check if target zid already exists, just in case of general zid disarray |
| 113 | // where the latest non-predefined zid is not the highest one |
| 114 | $exists = $dbr->newSelectQueryBuilder() |
| 115 | ->select( 'COUNT(*)' ) |
| 116 | ->from( 'page' ) |
| 117 | ->where( $conditions ) |
| 118 | ->andWhere( [ 'page_title' => $targetZid ] ) |
| 119 | ->caller( __METHOD__ ) |
| 120 | ->fetchField(); |
| 121 | |
| 122 | // If Zid after latest added is available, return |
| 123 | if ( !$exists ) { |
| 124 | return $targetZid; |
| 125 | } |
| 126 | |
| 127 | // If Zid after latest is not available, find Zid after highest one |
| 128 | // NOTE: this operation is much more expensive, so we should only |
| 129 | // perform it in the exceptional case that latestZid fails. |
| 130 | $highestZid = $dbr->newSelectQueryBuilder() |
| 131 | ->select( 'page_title' ) |
| 132 | ->from( 'page' ) |
| 133 | ->where( $conditions ) |
| 134 | ->orderBy( 'CAST(SUBSTR(page_title, 2) AS INTEGER)', SelectQueryBuilder::SORT_DESC ) |
| 135 | ->caller( __METHOD__ ) |
| 136 | ->fetchField(); |
| 137 | |
| 138 | if ( $highestZid === false ) { |
| 139 | $highestZid = self::PREDEFINED_TOP_LIMIT; |
| 140 | } |
| 141 | |
| 142 | $this->logger->warning( |
| 143 | __METHOD__ . ' at first got "' . $targetZid . '" — exists; slower query gets: "Z' . $highestZid . '".', |
| 144 | [ 'targetZid' => $targetZid, 'highestZid' => $highestZid ] |
| 145 | ); |
| 146 | |
| 147 | $targetZid = 'Z' . ( max( intval( substr( $highestZid, 1 ) ) + 1, 10000 ) ); |
| 148 | |
| 149 | return $targetZid; |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * Load a page revision from a given revision ID number. |
| 154 | * Returns null if no such revision can be found. |
| 155 | * |
| 156 | * @param int $id Revision ID of this revision |
| 157 | * @return RevisionRecord|null |
| 158 | */ |
| 159 | public function getRevisionById( int $id ): ?RevisionRecord { |
| 160 | $revisionRecord = $this->revisionStore->getRevisionById( $id ); |
| 161 | return $revisionRecord; |
| 162 | } |
| 163 | |
| 164 | /** |
| 165 | * Fetch the ZObject given its title and return it wrapped in a ZObjectContent object |
| 166 | * |
| 167 | * When a performer is supplied the content is read for that user's audience, |
| 168 | * so a RevisionDelete'd or suppressed revision they may not see yields false; |
| 169 | * internal/system callers pass no performer and read raw. |
| 170 | * |
| 171 | * TODO: Once the security fix is deployed, make the visibility audience |
| 172 | * explicit — an $audience parameter mirroring RevisionRecord::getContent(), |
| 173 | * so the ~22 internal/system callers state RAW deliberately — rather than |
| 174 | * inferring it from whether a performer is passed. Deferred to keep this |
| 175 | * patch minimal; do it under regular code review. |
| 176 | * |
| 177 | * @param Title $title The ZObject to fetch |
| 178 | * @param int|null $requestedRevision The revision ID of the page to fetch. If unset, the latest is returned. |
| 179 | * @param Authority|null $performer User to check deletion/suppression visibility for, if any |
| 180 | * @return ZObjectContent|bool Found ZObject |
| 181 | */ |
| 182 | public function fetchZObjectByTitle( Title $title, ?int $requestedRevision = null, ?Authority $performer = null ) { |
| 183 | if ( $requestedRevision ) { |
| 184 | $revision = $this->revisionStore->getRevisionByTitle( $title, $requestedRevision, 0 ); |
| 185 | } else { |
| 186 | $revision = $this->revisionStore->getKnownLatestRevision( $title ); |
| 187 | } |
| 188 | |
| 189 | if ( !$revision ) { |
| 190 | // Return false: We should not throw exceptions that trigger more fetches (ZErrorException) |
| 191 | return false; |
| 192 | } |
| 193 | |
| 194 | // NOTE: Hard-coding use of MAIN slot; if we're going the MCR route, we may wish to change this (or not). |
| 195 | // Honour deletion/suppression for a specific user; null-fail if they may not see it. |
| 196 | $audience = $performer ? RevisionRecord::FOR_THIS_USER : RevisionRecord::RAW; |
| 197 | return $revision->getContent( SlotRecord::MAIN, $audience, $performer ) ?? false; |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * Get the current ZPersistentObject of a given ZID, from the cache if available |
| 202 | * |
| 203 | * @param string $zid |
| 204 | * @return ZObjectContent|bool Cached or persisted ZObject, false if not found or invalid |
| 205 | */ |
| 206 | public function fetchZObject( string $zid ) { |
| 207 | $cacheKey = $this->zObjectCache->makeKey( self::ZOBJECT_CACHE_KEY_PREFIX, $zid ); |
| 208 | $cachedObject = $this->zObjectCache->get( $cacheKey ); |
| 209 | |
| 210 | if ( $cachedObject ) { |
| 211 | $cachedContent = ZObjectContentHandler::makeContent( $cachedObject, null, CONTENT_MODEL_ZOBJECT ); |
| 212 | if ( $cachedContent->isValid() ) { |
| 213 | return $cachedContent; |
| 214 | } |
| 215 | // Something went wrong; |
| 216 | // the object stored in the cache is not valid, delete from cache and go on |
| 217 | $this->zObjectCache->delete( $cacheKey ); |
| 218 | } |
| 219 | |
| 220 | // Cache miss somehow; let's re-fetch the object, and stash it in the cache. |
| 221 | |
| 222 | $title = Title::newFromText( $zid ); |
| 223 | if ( !$title ) { |
| 224 | // Return false: We should not throw exceptions that trigger more fetches (ZErrorException) |
| 225 | return false; |
| 226 | } |
| 227 | |
| 228 | $zObjectContent = $this->fetchZObjectByTitle( $title ); |
| 229 | if ( !$zObjectContent || !$zObjectContent->isValid() ) { |
| 230 | // Return false: We should not throw exceptions that trigger more fetches (ZErrorException) |
| 231 | return false; |
| 232 | } |
| 233 | |
| 234 | $this->zObjectCache->set( |
| 235 | $cacheKey, |
| 236 | $zObjectContent->getText(), |
| 237 | $this->zObjectCache::TTL_MONTH |
| 238 | ); |
| 239 | |
| 240 | return $zObjectContent; |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * Returns an array of ZPersistentObjects fetched from the DB given an array of their Zids |
| 245 | * |
| 246 | * Note that this will only fetch the latest revision of a ZObject; if you want a specific |
| 247 | * revision, you need to use ZObjectStore::fetchZObjectByTitle() instead. |
| 248 | * |
| 249 | * @param string[] $zids |
| 250 | * @return ZPersistentObject[] |
| 251 | */ |
| 252 | public function fetchBatchZObjects( $zids ): array { |
| 253 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 254 | $query = WikiPage::getQueryInfo(); |
| 255 | |
| 256 | $res = $dbr->newSelectQueryBuilder() |
| 257 | ->select( $query['fields'] ) |
| 258 | ->from( 'page' ) |
| 259 | ->where( [ |
| 260 | 'page_namespace' => NS_MAIN, |
| 261 | 'page_title' => $zids |
| 262 | ] ) |
| 263 | ->caller( __METHOD__ ) |
| 264 | ->fetchResultSet(); |
| 265 | |
| 266 | $titleArray = new TitleArrayFromResult( $res ); |
| 267 | |
| 268 | $dataArray = []; |
| 269 | foreach ( $titleArray as $title ) { |
| 270 | // TODO (T300521): Handle error from fetchZObjectByTitle |
| 271 | $content = $this->fetchZObjectByTitle( $title ); |
| 272 | if ( $content->isValid() ) { |
| 273 | $dataArray[ $title->getBaseText() ] = $content->getZObject(); |
| 274 | } |
| 275 | } |
| 276 | return $dataArray; |
| 277 | } |
| 278 | |
| 279 | /** |
| 280 | * Push a given Object into the Database, without validation. |
| 281 | * |
| 282 | * @param string $zid |
| 283 | * @param string $data |
| 284 | * @param string $summary |
| 285 | * @return bool True if a new revision was created; false on a null edit (no change) |
| 286 | * @throws Exception |
| 287 | */ |
| 288 | public function pushZObject( string $zid, string $data, string $summary ) { |
| 289 | $title = $this->titleFactory->newFromText( $zid, NS_MAIN ); |
| 290 | // Error: Failed creating title due to invalid format |
| 291 | if ( !$title ) { |
| 292 | throw new ZErrorException( |
| 293 | ZErrorFactory::createZErrorInstance( |
| 294 | ZErrorTypeRegistry::Z_ERROR_INVALID_TITLE, |
| 295 | [ 'title' => $zid ] |
| 296 | ) |
| 297 | ); |
| 298 | } |
| 299 | |
| 300 | $page = $this->wikiPageFactory->newFromTitle( $title ); |
| 301 | $flags = $title->exists() ? EDIT_UPDATE : EDIT_NEW; |
| 302 | |
| 303 | // Create system user: |
| 304 | $creatingUserName = wfMessage( 'wikilambda-systemuser' )->inLanguage( 'en' )->text(); |
| 305 | $user = User::newSystemUser( $creatingUserName, [ 'steal' => true ] ); |
| 306 | $this->userGroupManager->addUserToGroup( $user, 'sysop' ); |
| 307 | $this->userGroupManager->addUserToGroup( $user, 'functionmaintainer' ); |
| 308 | $this->userGroupManager->addUserToGroup( $user, 'functioneer' ); |
| 309 | $this->userGroupManager->addUserToGroup( $user, 'wikifunctions-staff' ); |
| 310 | |
| 311 | try { |
| 312 | $content = ZObjectContentHandler::makeContent( $data, $title ); |
| 313 | } catch ( ZErrorException $e ) { |
| 314 | // Error: Make content will only fail if JSON is invalid |
| 315 | $this->logger->warning( |
| 316 | __METHOD__ . ' triggered an error on creating content for page "' . $zid . '"', |
| 317 | [ 'responseError' => $e ] |
| 318 | ); |
| 319 | throw $e; |
| 320 | } |
| 321 | |
| 322 | try { |
| 323 | $status = $page->doUserEditContent( $content, $user, $summary, $flags ); |
| 324 | } catch ( Exception $e ) { |
| 325 | // Error: Database or MediaWiki exception |
| 326 | $this->logger->warning( |
| 327 | __METHOD__ . ' triggered an error on publish for page "' . $zid . '"', |
| 328 | [ 'responseError' => $e ] |
| 329 | ); |
| 330 | throw $e; |
| 331 | } |
| 332 | |
| 333 | // Error: Other doUserEditContent related errors |
| 334 | if ( !$status->isOK() ) { |
| 335 | $statusMessage = $status->getMessage(); |
| 336 | $this->logger->info( |
| 337 | __METHOD__ . ' got a non-OK Status on publish, for page "' . $zid . '"', |
| 338 | [ 'responseStatus' => var_export( $statusMessage, true ) ] |
| 339 | ); |
| 340 | throw new ZErrorException( |
| 341 | ZErrorFactory::createZErrorInstance( |
| 342 | ZErrorTypeRegistry::Z_ERROR_UNKNOWN, |
| 343 | [ 'message' => (string)$statusMessage ] |
| 344 | ) |
| 345 | ); |
| 346 | } |
| 347 | |
| 348 | return $status->wasRevisionCreated(); |
| 349 | } |
| 350 | |
| 351 | /** |
| 352 | * Create a new ZObject, with a newly assigned ZID, and store it in the Database |
| 353 | * |
| 354 | * @param MessageLocalizer $context The context of the action operation, for localisation of messages |
| 355 | * @param string $data |
| 356 | * @param string $summary |
| 357 | * @param User $user |
| 358 | * @return ZObjectPage |
| 359 | */ |
| 360 | public function createNewZObject( MessageLocalizer $context, string $data, string $summary, User $user ) { |
| 361 | // Find all placeholder ZIDs and ZKeys and replace those with the next available ZID |
| 362 | $zid = $this->getNextAvailableZid(); |
| 363 | $zObjectString = ZObjectUtils::replaceNullReferencePlaceholder( $data, $zid ); |
| 364 | |
| 365 | return $this->updateZObject( $context, $zid, $zObjectString, $summary, $user, EDIT_NEW ); |
| 366 | } |
| 367 | |
| 368 | /** |
| 369 | * Create or update a ZObject it in the Database |
| 370 | * |
| 371 | * @param MessageLocalizer $context The context of the action operation, for localisation of messages |
| 372 | * @param string $zid The ZID of the page to create/update, e.g. 'Z12345' |
| 373 | * @param string $data The ZObject's JSON to store, in string form, i.e. "{ Z1K1: "Z2", Z2K1: … }" |
| 374 | * @param string $summary An edit summary to display in the page's history, Recent Changes, watchlists, etc. |
| 375 | * @param User $user The user making the edit. |
| 376 | * @param int $flags Either EDIT_UPDATE (default) if editing or EDIT_NEW if creating a page |
| 377 | * @return ZObjectPage |
| 378 | */ |
| 379 | public function updateZObject( |
| 380 | MessageLocalizer $context, string $zid, string $data, string $summary, User $user, int $flags = EDIT_UPDATE |
| 381 | ) { |
| 382 | $title = $this->titleFactory->newFromText( $zid, NS_MAIN ); |
| 383 | |
| 384 | // ERROR: Title is empty or invalid |
| 385 | if ( !( $title instanceof Title ) ) { |
| 386 | $error = ZErrorFactory::createZErrorInstance( |
| 387 | ZErrorTypeRegistry::Z_ERROR_INVALID_TITLE, |
| 388 | [ 'title' => $zid ] |
| 389 | ); |
| 390 | return ZObjectPage::newFatal( $error ); |
| 391 | } |
| 392 | |
| 393 | // If edit flag or title did not exist, we are creating a new object |
| 394 | $creating = ( $flags === EDIT_NEW ) || !( $title->exists() ); |
| 395 | |
| 396 | try { |
| 397 | $content = ZObjectContentHandler::makeContent( $data, $title ); |
| 398 | } catch ( ZErrorException $e ) { |
| 399 | $this->logger->info( |
| 400 | __METHOD__ . ': makeContent threw ZErrorException for {zid}: {message}', |
| 401 | [ 'zid' => $zid, 'message' => $e->getMessage() ] |
| 402 | ); |
| 403 | return ZObjectPage::newFatal( $e->getZError() ); |
| 404 | } catch ( ContentSerializationException $mwe ) { |
| 405 | $this->logger->info( |
| 406 | __METHOD__ . ': makeContent threw ContentSerializationException for {zid}: {message}', |
| 407 | [ 'zid' => $zid, 'message' => $mwe->getMessage() ] |
| 408 | ); |
| 409 | return ZObjectPage::newFatal( |
| 410 | // We can't cleanly recover the inner ZErrorException (if indeed it was even thrown by us), so |
| 411 | // for now just pass this down as a Z500, perhaps with the ErrorType as the message. |
| 412 | ZErrorFactory::createZErrorInstance( |
| 413 | ZErrorTypeRegistry::Z_ERROR_UNKNOWN, |
| 414 | [ 'message' => $mwe->getMessage() ] |
| 415 | ) |
| 416 | ); |
| 417 | } |
| 418 | |
| 419 | // Error: ZObject validation errors. |
| 420 | if ( !( $content->isValid() ) ) { |
| 421 | return ZObjectPage::newFatal( $content->getErrors() ); |
| 422 | } |
| 423 | |
| 424 | // Validate that $zid and zObject[Z2K1] are the same |
| 425 | $zObjectId = $content->getZid(); |
| 426 | |
| 427 | if ( $zObjectId !== $zid ) { |
| 428 | $error = ZErrorFactory::createZErrorInstance( |
| 429 | ZErrorTypeRegistry::Z_ERROR_UNMATCHING_ZID, |
| 430 | [ |
| 431 | 'zid' => $zObjectId, |
| 432 | 'title' => $zid |
| 433 | ] |
| 434 | ); |
| 435 | return ZObjectPage::newFatal( $error ); |
| 436 | } |
| 437 | |
| 438 | $ztype = $content->getZType(); |
| 439 | |
| 440 | // Stop from creating and editing any types form DISALLOWED_ROOT_ZOBJECT |
| 441 | if ( in_array( $ztype, ZTypeRegistry::DISALLOWED_ROOT_ZOBJECTS ) ) { |
| 442 | $error = ZErrorFactory::createZErrorInstance( |
| 443 | ZErrorTypeRegistry::Z_ERROR_DISALLOWED_ROOT_ZOBJECT, |
| 444 | [ 'data' => $content->getZType() ] |
| 445 | ); |
| 446 | return ZObjectPage::newFatal( $error ); |
| 447 | } |
| 448 | |
| 449 | // Find the label conflicts. |
| 450 | $labels = $content->getLabels()->getValueAsList(); |
| 451 | $clashes = $this->findZObjectLabelConflicts( $zid, $ztype, $labels ); |
| 452 | |
| 453 | if ( count( $clashes ) > 0 ) { |
| 454 | $error = ZErrorFactory::createLabelClashZErrors( $clashes ); |
| 455 | return ZObjectPage::newFatal( $error ); |
| 456 | } |
| 457 | |
| 458 | // Use ZObjectAuthorization service to check that the user has the required permissions |
| 459 | // while creating or editing an object |
| 460 | $fromContent = null; |
| 461 | if ( !$creating ) { |
| 462 | $currentRevision = $this->revisionStore->getKnownLatestRevision( $title ); |
| 463 | $fromContent = $currentRevision->getSlots()->getContent( SlotRecord::MAIN ); |
| 464 | '@phan-var ZObjectContent $fromContent'; |
| 465 | } |
| 466 | $authorizationService = WikiLambdaServices::getZObjectAuthorization(); |
| 467 | $status = $authorizationService->authorize( $fromContent, $content, $user, $title ); |
| 468 | |
| 469 | // Return AuthorizationStatus->error if authorization service failed |
| 470 | if ( !$status->isValid() ) { |
| 471 | return ZObjectPage::newFatal( $status->getErrors() ); |
| 472 | } |
| 473 | |
| 474 | // Run ZObjectContent field validation |
| 475 | try { |
| 476 | $content->validateFields( $context ); |
| 477 | } catch ( ZErrorException $e ) { |
| 478 | return ZObjectPage::newFatal( $e->getZError() ); |
| 479 | } |
| 480 | |
| 481 | // We prepare the content to be saved |
| 482 | $page = $this->wikiPageFactory->newFromTitle( $title ); |
| 483 | try { |
| 484 | $status = $page->doUserEditContent( $content, $user, $summary, $flags ); |
| 485 | } catch ( Exception $e ) { |
| 486 | // Error: Database or a deeper MediaWiki error, e.g. a general editing rate limit |
| 487 | |
| 488 | $this->logger->warning( |
| 489 | __METHOD__ . ' triggered an error on publish, e.g. rate limited, for page "' . $zid . '"', |
| 490 | [ 'responseError' => $e ] |
| 491 | ); |
| 492 | |
| 493 | if ( $e instanceof ZErrorException ) { |
| 494 | // TODO (T362236): Add the rendering language as a parameter, don't default to English |
| 495 | $errorMessage = $e->getZError()->getMessage( 'en' ); |
| 496 | } else { |
| 497 | $errorMessage = $e->getMessage(); |
| 498 | } |
| 499 | |
| 500 | $error = ZErrorFactory::createZErrorInstance( |
| 501 | ZErrorTypeRegistry::Z_ERROR_UNKNOWN, |
| 502 | [ 'message' => $errorMessage ] |
| 503 | ); |
| 504 | return ZObjectPage::newFatal( $error ); |
| 505 | } |
| 506 | |
| 507 | if ( !$status->isOK() ) { |
| 508 | |
| 509 | // TODO (T362246): Dependency-inject |
| 510 | $statusFormatter = MediaWikiServices::getInstance()->getFormatterFactory() |
| 511 | ->getStatusFormatter( $context ); |
| 512 | |
| 513 | // Error: Other doUserEditContent related errors |
| 514 | |
| 515 | $this->logger->info( |
| 516 | __METHOD__ . ' got a non-OK Status on publish, for page "' . $zid . '"', |
| 517 | [ 'responseStatus' => var_export( $status, true ) ] |
| 518 | ); |
| 519 | |
| 520 | $error = ZErrorFactory::createZErrorInstance( |
| 521 | ZErrorTypeRegistry::Z_ERROR_UNKNOWN, |
| 522 | [ 'message' => $statusFormatter->getMessage( $status )->plain() ] |
| 523 | ); |
| 524 | return ZObjectPage::newFatal( $error ); |
| 525 | } |
| 526 | |
| 527 | // Success: return WikiPage |
| 528 | return ZObjectPage::newSuccess( $page ); |
| 529 | } |
| 530 | |
| 531 | /** |
| 532 | * Create or update a ZObject it in the Database as a System User |
| 533 | * |
| 534 | * @param MessageLocalizer $context The context of the action operation, for localisation of messages |
| 535 | * @param string $zid |
| 536 | * @param string $data |
| 537 | * @param string $summary |
| 538 | * @param int $flags |
| 539 | * @return ZObjectPage |
| 540 | */ |
| 541 | public function updateZObjectAsSystemUser( |
| 542 | MessageLocalizer $context, string $zid, string $data, string $summary, int $flags = EDIT_UPDATE |
| 543 | ) { |
| 544 | $creatingUserName = wfMessage( 'wikilambda-systemuser' )->inLanguage( 'en' )->text(); |
| 545 | // System user must belong to all privileged groups in order to |
| 546 | // perform all zobject creation and editing actions: |
| 547 | $user = User::newSystemUser( $creatingUserName, [ 'steal' => true ] ); |
| 548 | $this->userGroupManager->addUserToGroup( $user, 'sysop' ); |
| 549 | $this->userGroupManager->addUserToGroup( $user, 'functionmaintainer' ); |
| 550 | $this->userGroupManager->addUserToGroup( $user, 'functioneer' ); |
| 551 | $this->userGroupManager->addUserToGroup( $user, 'wikifunctions-staff' ); |
| 552 | // Make sure the edit will be marked as a bot edit |
| 553 | $flags |= EDIT_FORCE_BOT; |
| 554 | return $this->updateZObject( $context, $zid, $data, $summary, $user, $flags ); |
| 555 | } |
| 556 | |
| 557 | /** |
| 558 | * Delete the labels from the wikilambda_zobject_labels database that correspond |
| 559 | * to the given ZID. |
| 560 | * |
| 561 | * @param string $zid |
| 562 | */ |
| 563 | public function deleteZObjectLabelsByZid( string $zid ): void { |
| 564 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 565 | |
| 566 | $dbw->newDeleteQueryBuilder() |
| 567 | ->deleteFrom( 'wikilambda_zobject_labels' ) |
| 568 | ->where( [ 'wlzl_zobject_zid' => $zid ] ) |
| 569 | ->caller( __METHOD__ )->execute(); |
| 570 | } |
| 571 | |
| 572 | /** |
| 573 | * Delete the label conflicts from the wikilambda_zobject_label_conflicts database |
| 574 | * that correspond to the given ZID. |
| 575 | * |
| 576 | * @param string $zid |
| 577 | */ |
| 578 | public function deleteZObjectLabelConflictsByZid( string $zid ): void { |
| 579 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 580 | |
| 581 | $dbw->newDeleteQueryBuilder() |
| 582 | ->deleteFrom( 'wikilambda_zobject_label_conflicts' ) |
| 583 | ->where( |
| 584 | $dbw->expr( 'wlzlc_existing_zid', '=', $zid ) |
| 585 | ->or( 'wlzlc_conflicting_zid', '=', $zid ) |
| 586 | ) |
| 587 | ->caller( __METHOD__ )->execute(); |
| 588 | } |
| 589 | |
| 590 | /** |
| 591 | * Query the wikilambda_zobject_labels database for primary labels that have |
| 592 | * the same combination of language code and value for a different ZID |
| 593 | * than the given in the parameters. These will be considered conflicting labels. |
| 594 | * |
| 595 | * @param string $zid |
| 596 | * @param string $ztype |
| 597 | * @param array<string,string> $labels Array of labels, where the key is the language code and the value |
| 598 | * is the string representation of the label in that language |
| 599 | * @return array Conflicts found in the wikilambda_zobject_labels database |
| 600 | */ |
| 601 | public function findZObjectLabelConflicts( $zid, $ztype, $labels ): array { |
| 602 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 603 | |
| 604 | // remove labels with an undefined value |
| 605 | $labels = array_filter( |
| 606 | $labels, static function ( $value ) { |
| 607 | return $value !== ""; |
| 608 | } ); |
| 609 | |
| 610 | if ( $labels === [] ) { |
| 611 | return []; |
| 612 | } |
| 613 | |
| 614 | $labelConflictConditions = []; |
| 615 | foreach ( $labels as $language => $value ) { |
| 616 | $labelConflictConditions[] = $dbr->andExpr( [ |
| 617 | 'wlzl_language' => $language, |
| 618 | 'wlzl_label' => $value, |
| 619 | 'wlzl_label_primary' => true |
| 620 | ] ); |
| 621 | } |
| 622 | |
| 623 | $res = $dbr->newSelectQueryBuilder() |
| 624 | ->select( [ 'wlzl_zobject_zid', 'wlzl_language' ] ) |
| 625 | ->from( 'wikilambda_zobject_labels' ) |
| 626 | ->where( [ |
| 627 | $dbr->expr( 'wlzl_zobject_zid', '!=', $zid ), |
| 628 | // TODO (T357552): Check against type, once we properly implement that. |
| 629 | // 'wlzl_type' => $ztype, |
| 630 | $dbr->orExpr( $labelConflictConditions ) |
| 631 | ] ) |
| 632 | ->caller( __METHOD__ ) |
| 633 | ->fetchResultSet(); |
| 634 | |
| 635 | $conflicts = []; |
| 636 | foreach ( $res as $row ) { |
| 637 | // TODO (T362247): What if more than one conflicts with us on each language? |
| 638 | $conflicts[ $row->wlzl_language ] = $row->wlzl_zobject_zid; |
| 639 | } |
| 640 | |
| 641 | return $conflicts; |
| 642 | } |
| 643 | |
| 644 | /** |
| 645 | * Insert labels into the wikilambda_zobject_labels database for a given ZID and Type |
| 646 | * |
| 647 | * @param string $zid |
| 648 | * @param string $ztype |
| 649 | * @param array $labels Array of labels, where the key is the language code and the value |
| 650 | * is the string representation of the label in that language |
| 651 | * @param string|null $returnType |
| 652 | */ |
| 653 | public function insertZObjectLabels( $zid, $ztype, $labels, $returnType = null ): void { |
| 654 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 655 | |
| 656 | $updates = []; |
| 657 | foreach ( $labels as $language => $value ) { |
| 658 | $updates[] = [ |
| 659 | 'wlzl_zobject_zid' => $zid, |
| 660 | 'wlzl_language' => $language, |
| 661 | 'wlzl_type' => $ztype, |
| 662 | 'wlzl_label' => $value, |
| 663 | 'wlzl_label_normalised' => ZObjectUtils::comparableString( $value ), |
| 664 | 'wlzl_label_primary' => true, |
| 665 | 'wlzl_return_type' => $returnType |
| 666 | ]; |
| 667 | } |
| 668 | |
| 669 | // Exit early if there are no updates to make. |
| 670 | if ( count( $updates ) === 0 ) { |
| 671 | return; |
| 672 | } |
| 673 | |
| 674 | $dbw->newInsertQueryBuilder() |
| 675 | ->insertInto( 'wikilambda_zobject_labels' ) |
| 676 | ->rows( $updates ) |
| 677 | ->caller( __METHOD__ )->execute(); |
| 678 | } |
| 679 | |
| 680 | /** |
| 681 | * Insert language code into the wikilambda_zlanguages database for a given ZID |
| 682 | * |
| 683 | * @param string $zid |
| 684 | * @param string $languageCode |
| 685 | */ |
| 686 | public function insertZLanguageToLanguagesCache( string $zid, string $languageCode ): void { |
| 687 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 688 | |
| 689 | $dbw->newInsertQueryBuilder() |
| 690 | ->insertInto( 'wikilambda_zlanguages' ) |
| 691 | ->row( [ 'wlzlangs_zid' => $zid, 'wlzlangs_language' => $languageCode ] ) |
| 692 | ->caller( __METHOD__ )->execute(); |
| 693 | } |
| 694 | |
| 695 | /** |
| 696 | * Insert label conflicts into the wikilambda_zobject_label_conflicts database for a given ZID |
| 697 | * |
| 698 | * @param string $zid |
| 699 | * @param array $conflicts Array of labels, where the key is the language code and the value |
| 700 | * is the other ZID for which this label is repeated |
| 701 | */ |
| 702 | public function insertZObjectLabelConflicts( $zid, $conflicts ): void { |
| 703 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 704 | |
| 705 | $updates = []; |
| 706 | foreach ( $conflicts as $language => $existingZid ) { |
| 707 | $updates[] = [ |
| 708 | 'wlzlc_existing_zid' => $existingZid, |
| 709 | 'wlzlc_conflicting_zid' => $zid, |
| 710 | 'wlzlc_language' => $language, |
| 711 | ]; |
| 712 | } |
| 713 | |
| 714 | // Exit early if there are no updates to make. |
| 715 | if ( count( $updates ) === 0 ) { |
| 716 | return; |
| 717 | } |
| 718 | |
| 719 | $dbw->newInsertQueryBuilder() |
| 720 | ->insertInto( 'wikilambda_zobject_label_conflicts' ) |
| 721 | ->rows( $updates ) |
| 722 | ->caller( __METHOD__ )->execute(); |
| 723 | } |
| 724 | |
| 725 | /** |
| 726 | * Insert alias (secondary labels) into the wikilambda_zobject_labels database for a given ZID and Type |
| 727 | * |
| 728 | * @param string $zid |
| 729 | * @param string $ztype |
| 730 | * @param array $aliases Set of labels, where the key is the language code |
| 731 | * and the value is an array of strings |
| 732 | * @param string|null $returnType |
| 733 | */ |
| 734 | public function insertZObjectAliases( $zid, $ztype, $aliases, $returnType = null ): void { |
| 735 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 736 | |
| 737 | $updates = []; |
| 738 | foreach ( $aliases as $language => $stringset ) { |
| 739 | foreach ( $stringset as $value ) { |
| 740 | $updates[] = [ |
| 741 | 'wlzl_zobject_zid' => $zid, |
| 742 | 'wlzl_language' => $language, |
| 743 | 'wlzl_type' => $ztype, |
| 744 | 'wlzl_label' => $value, |
| 745 | 'wlzl_label_normalised' => ZObjectUtils::comparableString( $value ), |
| 746 | 'wlzl_label_primary' => false, |
| 747 | 'wlzl_return_type' => $returnType |
| 748 | ]; |
| 749 | } |
| 750 | } |
| 751 | |
| 752 | if ( count( $updates ) === 0 ) { |
| 753 | return; |
| 754 | } |
| 755 | |
| 756 | $dbw->newInsertQueryBuilder() |
| 757 | ->insertInto( 'wikilambda_zobject_labels' ) |
| 758 | ->rows( $updates ) |
| 759 | ->caller( __METHOD__ )->execute(); |
| 760 | } |
| 761 | |
| 762 | /** |
| 763 | * Reconcile the rows in wikilambda_zobject_labels for a given ZID with the desired |
| 764 | * set of primary labels and aliases, writing only the differences. |
| 765 | * |
| 766 | * Existing rows whose content tuple matches a desired row are preserved (their |
| 767 | * wlzl_id is re-used); unmatched existing rows are deleted by id; desired rows |
| 768 | * with no match are inserted. When the desired set is identical to the existing |
| 769 | * set, no writes are issued at all. (T300522) |
| 770 | * |
| 771 | * Eventually we could replace this with a simpler upsert(), but that would need a |
| 772 | * DB schema change. |
| 773 | * |
| 774 | * @param string $zid |
| 775 | * @param string $ztype |
| 776 | * @param array<string,string> $primaryLabels Labels keyed by language code |
| 777 | * @param array<string,string[]> $aliases Aliases keyed by language code, with a list of strings each |
| 778 | * @param string|null $returnType |
| 779 | */ |
| 780 | public function synchroniseZObjectLabels( |
| 781 | string $zid, |
| 782 | string $ztype, |
| 783 | array $primaryLabels, |
| 784 | array $aliases, |
| 785 | ?string $returnType = null |
| 786 | ): void { |
| 787 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 788 | |
| 789 | // Build the desired row payloads from primary labels and aliases. |
| 790 | $desiredRows = []; |
| 791 | foreach ( $primaryLabels as $language => $value ) { |
| 792 | $desiredRows[] = [ |
| 793 | 'wlzl_zobject_zid' => $zid, |
| 794 | 'wlzl_language' => $language, |
| 795 | 'wlzl_type' => $ztype, |
| 796 | 'wlzl_label' => $value, |
| 797 | 'wlzl_label_normalised' => ZObjectUtils::comparableString( $value ), |
| 798 | 'wlzl_label_primary' => true, |
| 799 | 'wlzl_return_type' => $returnType, |
| 800 | ]; |
| 801 | } |
| 802 | foreach ( $aliases as $language => $stringset ) { |
| 803 | foreach ( $stringset as $value ) { |
| 804 | $desiredRows[] = [ |
| 805 | 'wlzl_zobject_zid' => $zid, |
| 806 | 'wlzl_language' => $language, |
| 807 | 'wlzl_type' => $ztype, |
| 808 | 'wlzl_label' => $value, |
| 809 | 'wlzl_label_normalised' => ZObjectUtils::comparableString( $value ), |
| 810 | 'wlzl_label_primary' => false, |
| 811 | 'wlzl_return_type' => $returnType, |
| 812 | ]; |
| 813 | } |
| 814 | } |
| 815 | |
| 816 | // Read existing rows from the primary DB: replica lag could otherwise |
| 817 | // produce spurious inserts (existing row not yet visible) or deletes. |
| 818 | $existingRows = $dbw->newSelectQueryBuilder() |
| 819 | ->select( [ |
| 820 | 'wlzl_id', |
| 821 | 'wlzl_language', |
| 822 | 'wlzl_type', |
| 823 | 'wlzl_label', |
| 824 | 'wlzl_label_primary', |
| 825 | 'wlzl_return_type', |
| 826 | ] ) |
| 827 | ->from( 'wikilambda_zobject_labels' ) |
| 828 | ->where( [ 'wlzl_zobject_zid' => $zid ] ) |
| 829 | ->caller( __METHOD__ ) |
| 830 | ->fetchResultSet(); |
| 831 | |
| 832 | // Bucket existing rows by their content tuple so each desired row can |
| 833 | // claim one matching existing row (preserving its wlzl_id). |
| 834 | $existingByKey = []; |
| 835 | foreach ( $existingRows as $row ) { |
| 836 | $key = self::makeLabelRowKey( |
| 837 | $row->wlzl_language, |
| 838 | $row->wlzl_type, |
| 839 | $row->wlzl_label, |
| 840 | (bool)$row->wlzl_label_primary, |
| 841 | $row->wlzl_return_type |
| 842 | ); |
| 843 | $existingByKey[$key][] = (int)$row->wlzl_id; |
| 844 | } |
| 845 | |
| 846 | $toInsert = []; |
| 847 | foreach ( $desiredRows as $row ) { |
| 848 | $key = self::makeLabelRowKey( |
| 849 | $row['wlzl_language'], |
| 850 | $row['wlzl_type'], |
| 851 | $row['wlzl_label'], |
| 852 | $row['wlzl_label_primary'], |
| 853 | $row['wlzl_return_type'] |
| 854 | ); |
| 855 | if ( !empty( $existingByKey[$key] ) ) { |
| 856 | array_shift( $existingByKey[$key] ); |
| 857 | } else { |
| 858 | $toInsert[] = $row; |
| 859 | } |
| 860 | } |
| 861 | |
| 862 | $toDeleteIds = []; |
| 863 | foreach ( $existingByKey as $ids ) { |
| 864 | foreach ( $ids as $id ) { |
| 865 | $toDeleteIds[] = $id; |
| 866 | } |
| 867 | } |
| 868 | |
| 869 | if ( $toDeleteIds !== [] ) { |
| 870 | $dbw->newDeleteQueryBuilder() |
| 871 | ->deleteFrom( 'wikilambda_zobject_labels' ) |
| 872 | ->where( [ 'wlzl_id' => $toDeleteIds ] ) |
| 873 | ->caller( __METHOD__ )->execute(); |
| 874 | } |
| 875 | |
| 876 | if ( $toInsert !== [] ) { |
| 877 | $dbw->newInsertQueryBuilder() |
| 878 | ->insertInto( 'wikilambda_zobject_labels' ) |
| 879 | ->rows( $toInsert ) |
| 880 | ->caller( __METHOD__ )->execute(); |
| 881 | } |
| 882 | } |
| 883 | |
| 884 | /** |
| 885 | * Reconcile the rows in wikilambda_zobject_label_conflicts touching the given ZID |
| 886 | * with the desired set of conflicts, writing only the differences. (T300522) |
| 887 | * |
| 888 | * The scope mirrors deleteZObjectLabelConflictsByZid: rows where the ZID appears |
| 889 | * as either side of the conflict. Desired rows are those where this ZID is the |
| 890 | * conflicting (incoming) side. |
| 891 | * |
| 892 | * @param string $zid |
| 893 | * @param array<string,string> $conflicts Map of language code to existing ZID that owns the label |
| 894 | */ |
| 895 | public function synchroniseZObjectLabelConflicts( string $zid, array $conflicts ): void { |
| 896 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 897 | |
| 898 | $desiredRows = []; |
| 899 | foreach ( $conflicts as $language => $existingZid ) { |
| 900 | $desiredRows[] = [ |
| 901 | 'wlzlc_existing_zid' => $existingZid, |
| 902 | 'wlzlc_conflicting_zid' => $zid, |
| 903 | 'wlzlc_language' => $language, |
| 904 | ]; |
| 905 | } |
| 906 | |
| 907 | $existingRows = $dbw->newSelectQueryBuilder() |
| 908 | ->select( [ |
| 909 | 'wlzlc_id', |
| 910 | 'wlzlc_existing_zid', |
| 911 | 'wlzlc_conflicting_zid', |
| 912 | 'wlzlc_language', |
| 913 | ] ) |
| 914 | ->from( 'wikilambda_zobject_label_conflicts' ) |
| 915 | ->where( |
| 916 | $dbw->expr( 'wlzlc_existing_zid', '=', $zid ) |
| 917 | ->or( 'wlzlc_conflicting_zid', '=', $zid ) |
| 918 | ) |
| 919 | ->caller( __METHOD__ ) |
| 920 | ->fetchResultSet(); |
| 921 | |
| 922 | $existingByKey = []; |
| 923 | foreach ( $existingRows as $row ) { |
| 924 | $key = self::makeConflictRowKey( |
| 925 | $row->wlzlc_existing_zid, |
| 926 | $row->wlzlc_conflicting_zid, |
| 927 | $row->wlzlc_language |
| 928 | ); |
| 929 | $existingByKey[$key][] = (int)$row->wlzlc_id; |
| 930 | } |
| 931 | |
| 932 | $toInsert = []; |
| 933 | foreach ( $desiredRows as $row ) { |
| 934 | $key = self::makeConflictRowKey( |
| 935 | $row['wlzlc_existing_zid'], |
| 936 | $row['wlzlc_conflicting_zid'], |
| 937 | $row['wlzlc_language'] |
| 938 | ); |
| 939 | if ( !empty( $existingByKey[$key] ) ) { |
| 940 | array_shift( $existingByKey[$key] ); |
| 941 | } else { |
| 942 | $toInsert[] = $row; |
| 943 | } |
| 944 | } |
| 945 | |
| 946 | $toDeleteIds = []; |
| 947 | foreach ( $existingByKey as $ids ) { |
| 948 | foreach ( $ids as $id ) { |
| 949 | $toDeleteIds[] = $id; |
| 950 | } |
| 951 | } |
| 952 | |
| 953 | if ( $toDeleteIds !== [] ) { |
| 954 | $dbw->newDeleteQueryBuilder() |
| 955 | ->deleteFrom( 'wikilambda_zobject_label_conflicts' ) |
| 956 | ->where( [ 'wlzlc_id' => $toDeleteIds ] ) |
| 957 | ->caller( __METHOD__ )->execute(); |
| 958 | } |
| 959 | |
| 960 | if ( $toInsert !== [] ) { |
| 961 | $dbw->newInsertQueryBuilder() |
| 962 | ->insertInto( 'wikilambda_zobject_label_conflicts' ) |
| 963 | ->rows( $toInsert ) |
| 964 | ->caller( __METHOD__ )->execute(); |
| 965 | } |
| 966 | } |
| 967 | |
| 968 | /** |
| 969 | * Build a content-tuple key for a wikilambda_zobject_labels row, used to |
| 970 | * detect rows that should be preserved across a label update. |
| 971 | * |
| 972 | * @param string $language |
| 973 | * @param string $type |
| 974 | * @param string $label |
| 975 | * @param bool $isPrimary |
| 976 | * @param string|null $returnType |
| 977 | * @return string |
| 978 | */ |
| 979 | private static function makeLabelRowKey( |
| 980 | string $language, |
| 981 | string $type, |
| 982 | string $label, |
| 983 | bool $isPrimary, |
| 984 | ?string $returnType |
| 985 | ): string { |
| 986 | // \x1f (Unit Separator) is non-printable and cannot occur in any of these columns. |
| 987 | return implode( "\x1f", [ |
| 988 | $language, |
| 989 | $type, |
| 990 | $label, |
| 991 | $isPrimary ? '1' : '0', |
| 992 | $returnType ?? '', |
| 993 | ] ); |
| 994 | } |
| 995 | |
| 996 | /** |
| 997 | * Build a content-tuple key for a wikilambda_zobject_label_conflicts row. |
| 998 | * |
| 999 | * @param string $existingZid |
| 1000 | * @param string $conflictingZid |
| 1001 | * @param string $language |
| 1002 | * @return string |
| 1003 | */ |
| 1004 | private static function makeConflictRowKey( |
| 1005 | string $existingZid, |
| 1006 | string $conflictingZid, |
| 1007 | string $language |
| 1008 | ): string { |
| 1009 | return implode( "\x1f", [ $existingZid, $conflictingZid, $language ] ); |
| 1010 | } |
| 1011 | |
| 1012 | /** |
| 1013 | * Gets from the secondary database a list of all Zids belonging to a given type |
| 1014 | * |
| 1015 | * @param string $ztype |
| 1016 | * @return string[] |
| 1017 | */ |
| 1018 | public function fetchZidsOfType( $ztype ): array { |
| 1019 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1020 | return $dbr->newSelectQueryBuilder() |
| 1021 | ->select( 'wlzl_zobject_zid' ) |
| 1022 | ->distinct() |
| 1023 | ->from( 'wikilambda_zobject_labels' ) |
| 1024 | ->where( [ |
| 1025 | 'wlzl_type' => $ztype |
| 1026 | ] ) |
| 1027 | ->orderBy( 'wlzl_zobject_zid', SelectQueryBuilder::SORT_ASC ) |
| 1028 | ->caller( __METHOD__ ) |
| 1029 | ->fetchFieldValues(); |
| 1030 | } |
| 1031 | |
| 1032 | /** |
| 1033 | * Gets from the secondary database the number of all Zids belonging to a given type |
| 1034 | * |
| 1035 | * @param string $ztype |
| 1036 | * @return int |
| 1037 | */ |
| 1038 | public function getCountOfTypeInstances( string $ztype ): int { |
| 1039 | // Special case for all ZObjects |
| 1040 | if ( $ztype === ZTypeRegistry::Z_OBJECT ) { |
| 1041 | return count( $this->fetchAllZids() ); |
| 1042 | } |
| 1043 | return count( $this->fetchZidsOfType( $ztype ) ); |
| 1044 | } |
| 1045 | |
| 1046 | /** |
| 1047 | * Get a list of all Zids persisted in the database |
| 1048 | * |
| 1049 | * @return string[] All persisted Zids |
| 1050 | */ |
| 1051 | public function fetchAllZids(): array { |
| 1052 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1053 | $zids = $dbr->newSelectQueryBuilder() |
| 1054 | ->select( 'page_title' ) |
| 1055 | ->from( 'page' ) |
| 1056 | ->where( [ |
| 1057 | 'page_namespace' => NS_MAIN |
| 1058 | ] ) |
| 1059 | ->caller( __METHOD__ ) |
| 1060 | ->fetchFieldValues(); |
| 1061 | return array_filter( $zids, ZObjectUtils::isValidZObjectReference( ... ) ); |
| 1062 | } |
| 1063 | |
| 1064 | /** |
| 1065 | * Gets from the secondary database a list of all natural language ZIDs, |
| 1066 | * mapping from BCP47 (or MediaWiki) language code to ZID (one zid can map |
| 1067 | * to multiple BCP47 codes) |
| 1068 | * |
| 1069 | * @return array<string,string> |
| 1070 | */ |
| 1071 | public function fetchAllZLanguageObjects(): array { |
| 1072 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1073 | $res = $dbr->newSelectQueryBuilder() |
| 1074 | ->select( [ 'wlzlangs_zid', 'wlzlangs_language' ] ) |
| 1075 | ->distinct() |
| 1076 | ->from( 'wikilambda_zlanguages' ) |
| 1077 | ->orderBy( 'wlzlangs_zid', SelectQueryBuilder::SORT_ASC ) |
| 1078 | ->caller( __METHOD__ ) |
| 1079 | ->fetchResultSet(); |
| 1080 | |
| 1081 | $languages = []; |
| 1082 | foreach ( $res as $row ) { |
| 1083 | $languages[ $row->wlzlangs_language ] = $row->wlzlangs_zid; |
| 1084 | } |
| 1085 | return $languages; |
| 1086 | } |
| 1087 | |
| 1088 | /** |
| 1089 | * Gets from the secondary database a list of all supported natural |
| 1090 | * BCP47 (or MediaWiki) language codes. |
| 1091 | * |
| 1092 | * @return array<string> |
| 1093 | */ |
| 1094 | public function fetchAllZLanguageCodes(): array { |
| 1095 | return array_keys( $this->fetchAllZLanguageObjects() ); |
| 1096 | } |
| 1097 | |
| 1098 | /** |
| 1099 | * Gets from the secondary database the matching BCP47 (or MediaWiki) language code(s) |
| 1100 | * for a given ZID |
| 1101 | * |
| 1102 | * @param string $zid The ZID of the matching ZLanguage object for which to search. |
| 1103 | * @return string[] Any BCP47 (or MediaWiki) language code(s) if found; unordered. |
| 1104 | */ |
| 1105 | public function findCodesFromZLanguage( string $zid ): array { |
| 1106 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1107 | return $dbr->newSelectQueryBuilder() |
| 1108 | ->select( 'wlzlangs_language' ) |
| 1109 | ->from( 'wikilambda_zlanguages' ) |
| 1110 | ->where( [ 'wlzlangs_zid' => $zid ], ) |
| 1111 | ->caller( __METHOD__ ) |
| 1112 | ->fetchFieldValues(); |
| 1113 | } |
| 1114 | |
| 1115 | /** |
| 1116 | * Fetch all ZLanguages stored in the language cache table, and |
| 1117 | * for each one, return its zid, its language code, and its label |
| 1118 | * in the user language or the closest available fallback. |
| 1119 | * |
| 1120 | * @param string $userLang - User language BCP47 code |
| 1121 | * @return IResultWrapper |
| 1122 | */ |
| 1123 | public function fetchAllZLanguagesWithLabels( $userLang ) { |
| 1124 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1125 | |
| 1126 | // TODO (T362246): Dependency-inject |
| 1127 | $zLangRegistry = ZLangRegistry::singleton(); |
| 1128 | $languageFallback = MediaWikiServices::getInstance()->getLanguageFallback(); |
| 1129 | $languages = $zLangRegistry->getListOfFallbackLanguageZids( $languageFallback, $userLang ); |
| 1130 | |
| 1131 | // Returns table with unique zids and the most preferred primary label |
| 1132 | $subquery = $this->getPreferredLabelsQuery( $languages )->getSQL(); |
| 1133 | return $dbr->newSelectQueryBuilder() |
| 1134 | ->select( [ 'wlzl_zobject_zid', 'wlzl_label', 'wlzlangs_language' ] ) |
| 1135 | ->rawTables( [ 'preferred_labels' => new Subquery( $subquery ) ] ) |
| 1136 | ->join( 'wikilambda_zlanguages', null, [ 'wlzl_zobject_zid = wlzlangs_zid' ] ) |
| 1137 | ->orderBy( 'wlzl_label', SelectQueryBuilder::SORT_ASC ) |
| 1138 | ->caller( __METHOD__ ) |
| 1139 | ->fetchResultSet(); |
| 1140 | } |
| 1141 | |
| 1142 | /** |
| 1143 | * Fetch all ZTypes that have persisted instances, with their |
| 1144 | * label in the user language or the closest available fallback. |
| 1145 | * |
| 1146 | * @param string $userLang - User language BCP47 code |
| 1147 | * @param bool $includeZ1 - Whether to include Z1/ZObject in the results |
| 1148 | * @return IResultWrapper |
| 1149 | */ |
| 1150 | public function fetchAllInstancedTypesWithLabels( string $userLang, bool $includeZ1 = false ) { |
| 1151 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1152 | |
| 1153 | // TODO (T362246): Dependency-inject |
| 1154 | // Returns table with unique zids and the most preferred primary label |
| 1155 | $zLangRegistry = ZLangRegistry::singleton(); |
| 1156 | $languageFallback = MediaWikiServices::getInstance()->getLanguageFallback(); |
| 1157 | $languages = $zLangRegistry->getListOfFallbackLanguageZids( $languageFallback, $userLang ); |
| 1158 | $subquery = $this->getPreferredLabelsQuery( $languages )->getSQL(); |
| 1159 | |
| 1160 | // Fetch only those types that have instances |
| 1161 | $zids = $this->fetchAllInstancedTypes(); |
| 1162 | |
| 1163 | // If Z1 should be included, add it to the list of ZIDs |
| 1164 | if ( $includeZ1 ) { |
| 1165 | $zids[] = 'Z1'; |
| 1166 | } |
| 1167 | |
| 1168 | // If there are no Types with instances (e.g. in CI), just return a fake empty result. |
| 1169 | if ( $zids === [] ) { |
| 1170 | return new FakeResultWrapper( $zids ); |
| 1171 | } |
| 1172 | |
| 1173 | return $dbr->newSelectQueryBuilder() |
| 1174 | ->select( [ 'wlzl_zobject_zid', 'wlzl_label' ] ) |
| 1175 | ->rawTables( [ 'preferred_labels' => new Subquery( $subquery ) ] ) |
| 1176 | ->where( [ 'wlzl_zobject_zid' => $zids ] ) |
| 1177 | ->orderBy( 'wlzl_label', SelectQueryBuilder::SORT_ASC ) |
| 1178 | ->caller( __METHOD__ ) |
| 1179 | ->fetchResultSet(); |
| 1180 | } |
| 1181 | |
| 1182 | /** |
| 1183 | * Returns a list of distinct type Zids that have persisted instances. |
| 1184 | * |
| 1185 | * @return string[] |
| 1186 | */ |
| 1187 | public function fetchAllInstancedTypes(): array { |
| 1188 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1189 | |
| 1190 | return $dbr->newSelectQueryBuilder() |
| 1191 | ->select( [ 'wlzl_type' ] ) |
| 1192 | ->distinct() |
| 1193 | ->from( 'wikilambda_zobject_labels' ) |
| 1194 | ->caller( __METHOD__ ) |
| 1195 | ->fetchFieldValues(); |
| 1196 | } |
| 1197 | |
| 1198 | /** |
| 1199 | * Generates a query to get the return types for all stored functions |
| 1200 | * and function calls. |
| 1201 | * |
| 1202 | * @return SelectQueryBuilder |
| 1203 | */ |
| 1204 | public function getReturnTypeQuery() { |
| 1205 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1206 | |
| 1207 | $caseReturnTypes = 'CASE ' |
| 1208 | . 'WHEN a.wlzo_main_type =' . $dbr->addQuotes( 'Z7' ) . ' THEN a.wlzo_related_zobject ' |
| 1209 | . 'WHEN a.wlzo_main_type =' . $dbr->addQuotes( 'Z8' ) . ' THEN a.wlzo_main_zid ' |
| 1210 | . 'END'; |
| 1211 | |
| 1212 | $joinConditions = [ |
| 1213 | 'rt.wlzo_key' => 'Z8K2', |
| 1214 | 'rt.wlzo_main_zid = ' . $caseReturnTypes |
| 1215 | ]; |
| 1216 | |
| 1217 | $filterConditions = $dbr->orExpr( [ |
| 1218 | $dbr->andExpr( [ 'a.wlzo_main_type' => 'Z7' ] ), |
| 1219 | $dbr->andExpr( [ |
| 1220 | 'a.wlzo_main_type' => 'Z8', |
| 1221 | 'a.wlzo_key' => 'Z8K2' |
| 1222 | ] ) |
| 1223 | ] ); |
| 1224 | |
| 1225 | $returnTypeQueryBuilder = $dbr->newSelectQueryBuilder() |
| 1226 | ->select( [ |
| 1227 | 'a.wlzo_main_zid', |
| 1228 | 'a.wlzo_main_type', |
| 1229 | 'return_type' => 'rt.wlzo_related_zobject' |
| 1230 | ] ) |
| 1231 | ->from( 'wikilambda_zobject_join', 'a' ) |
| 1232 | ->leftJoin( 'wikilambda_zobject_join', 'rt', $joinConditions ) |
| 1233 | ->where( $filterConditions ); |
| 1234 | |
| 1235 | return $returnTypeQueryBuilder; |
| 1236 | } |
| 1237 | |
| 1238 | /** |
| 1239 | * Generates a query that filters to the preferred label in the |
| 1240 | * labels table, depending on the user's language fallback chain |
| 1241 | * passed as parameter. |
| 1242 | * |
| 1243 | * @param string[] $languageChain List of language zids in order of preference |
| 1244 | * @param string|null $type Optional type zid to restrict the query to; callers |
| 1245 | * that only want objects of one type should pass it here rather than filter |
| 1246 | * the returned query, so that only that type's rows are ranked (T430853) |
| 1247 | * @return SelectQueryBuilder |
| 1248 | */ |
| 1249 | public function getPreferredLabelsQuery( $languageChain, ?string $type = null ) { |
| 1250 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1251 | |
| 1252 | // Build the CASE expression to assign each label row a language preference |
| 1253 | // index: the position of its language in the fallback chain if it is that |
| 1254 | // language's primary label, otherwise a shared worst index. |
| 1255 | $caseParts = []; |
| 1256 | foreach ( $languageChain as $index => $langZid ) { |
| 1257 | $caseParts[] = "WHEN l1.wlzl_language = " . |
| 1258 | $dbr->addQuotes( $langZid ) . " AND l1.wlzl_label_primary = " . |
| 1259 | $dbr->addQuotes( true ) . " THEN " . ( $index + 1 ); |
| 1260 | } |
| 1261 | $caseExpr = "CASE " . implode( " ", $caseParts ) . " ELSE " . ( count( $languageChain ) + 1 ) . " END"; |
| 1262 | |
| 1263 | // Rank each label row within its zid and keep only the single best one. |
| 1264 | // The tiebreakers after the preference index make the choice deterministic |
| 1265 | // even when no fallback-chain language has a primary label (in which case |
| 1266 | // every row shares the ELSE index): prefer a primary label, then the |
| 1267 | // earliest language, then the row id (wlzl_id, the table's PK). Without |
| 1268 | // this a zid with no preferred label was emitted once per label (T430847). |
| 1269 | $rankExpr = "ROW_NUMBER() OVER ( PARTITION BY l1.wlzl_zobject_zid ORDER BY " . |
| 1270 | "$caseExpr, l1.wlzl_label_primary DESC, l1.wlzl_language, l1.wlzl_id )"; |
| 1271 | |
| 1272 | $rankedQuery = $dbr->newSelectQueryBuilder() |
| 1273 | ->select( [ 'l1.wlzl_id', 'row_num' => $rankExpr ] ) |
| 1274 | ->from( 'wikilambda_zobject_labels', 'l1' ); |
| 1275 | |
| 1276 | // Restricting by type before ranking cannot change any per-zid winner: |
| 1277 | // a zid has exactly one wlzl_type across all of its label rows, so the |
| 1278 | // filter keeps all-or-none of each partition. Applying it here rather |
| 1279 | // than on the outer query prunes the window sort to one type's rows. |
| 1280 | if ( $type !== null ) { |
| 1281 | $rankedQuery->where( [ 'l1.wlzl_type' => $type ] ); |
| 1282 | } |
| 1283 | |
| 1284 | // Subquery yielding exactly one winning label row id per zid |
| 1285 | $prefQuery = $dbr->newSelectQueryBuilder() |
| 1286 | ->select( [ 'wlzl_id' ] ) |
| 1287 | ->rawTables( [ 'ranked' => new Subquery( $rankedQuery->getSQL() ) ] ) |
| 1288 | ->where( [ 'row_num' => 1 ] ); |
| 1289 | |
| 1290 | $pageJoinConditions = [ |
| 1291 | 'p.page_title = l1.wlzl_zobject_zid', |
| 1292 | 'p.page_namespace = ' . NS_MAIN |
| 1293 | ]; |
| 1294 | |
| 1295 | // Create query to select the preferred label for each zid |
| 1296 | // - Inner join on the winning label row id to keep one row per zid |
| 1297 | // - Join with zobject entries in the page table to order objects by creation date |
| 1298 | $queryBuilder = $dbr->newSelectQueryBuilder() |
| 1299 | ->select( [ |
| 1300 | 'p.page_id', |
| 1301 | 'l1.wlzl_zobject_zid', |
| 1302 | 'l1.wlzl_label', |
| 1303 | 'l1.wlzl_language', |
| 1304 | 'l1.wlzl_type', |
| 1305 | 'l1.wlzl_label_normalised', |
| 1306 | 'l1.wlzl_label_primary', |
| 1307 | 'l1.wlzl_return_type' |
| 1308 | ] ) |
| 1309 | ->from( 'wikilambda_zobject_labels', 'l1' ) |
| 1310 | ->join( $prefQuery, 'pref', 'l1.wlzl_id = pref.wlzl_id' ) |
| 1311 | ->leftJoin( 'page', 'p', $pageJoinConditions ); |
| 1312 | |
| 1313 | // Semantically redundant with the ranked-subquery filter above (the |
| 1314 | // join on the winning wlzl_id already restricts rows to this type), |
| 1315 | // but restated here so the optimiser can prune this scan too rather |
| 1316 | // than reading and sorting every label row. |
| 1317 | if ( $type !== null ) { |
| 1318 | $queryBuilder->where( [ 'l1.wlzl_type' => $type ] ); |
| 1319 | } |
| 1320 | |
| 1321 | return $queryBuilder; |
| 1322 | } |
| 1323 | |
| 1324 | /** |
| 1325 | * Create OR aggregated LIKE statements for the whole (trimmed) substring, and |
| 1326 | * for each large token (>2 chars) from the searchTerm. |
| 1327 | * The search term is tokenized by splitting by whitespace characters. |
| 1328 | * |
| 1329 | * @param string $searchColumn |
| 1330 | * @param string $searchTerm |
| 1331 | * @return IExpression|null |
| 1332 | */ |
| 1333 | private function getStringMatchCondition( $searchColumn, $searchTerm ) { |
| 1334 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1335 | $tokens = preg_split( '/\s+/', trim( $searchTerm ), -1, PREG_SPLIT_NO_EMPTY ); |
| 1336 | |
| 1337 | // Return null if there are no string match conditions |
| 1338 | if ( count( $tokens ) === 0 ) { |
| 1339 | return null; |
| 1340 | } |
| 1341 | |
| 1342 | $conditions = []; |
| 1343 | |
| 1344 | // Add condition for whole substring (trimmed) |
| 1345 | $conditions[] = $dbr->expr( |
| 1346 | $searchColumn, |
| 1347 | IExpression::LIKE, |
| 1348 | new LikeValue( $dbr->anyString(), trim( $searchTerm ), $dbr->anyString() ) |
| 1349 | ); |
| 1350 | |
| 1351 | // Add additional conditions for largest tokens |
| 1352 | foreach ( $tokens as $token ) { |
| 1353 | if ( strlen( $token ) > 2 ) { |
| 1354 | // NOTE: length this is not a good method to determine stop words and is not |
| 1355 | // applicable to all languages, but the whole substring is already part of the |
| 1356 | // search and this method effective to filter out most of the bad matches. |
| 1357 | $conditions[] = $dbr->expr( |
| 1358 | $searchColumn, |
| 1359 | IExpression::LIKE, |
| 1360 | new LikeValue( $dbr->anyString(), $token, $dbr->anyString() ) |
| 1361 | ); |
| 1362 | } |
| 1363 | } |
| 1364 | |
| 1365 | // Use OR to aggregate results |
| 1366 | return $dbr->orExpr( $conditions ); |
| 1367 | } |
| 1368 | |
| 1369 | /** |
| 1370 | * Search functions that match a series of conditions: |
| 1371 | * * are running functions (they must have connected implementations) |
| 1372 | * * have a label that partially matches the given searchTerm, and |
| 1373 | * * either have fully renderable inputs and outputs (to be used by WP integration) |
| 1374 | * * or have the specified input and output types |
| 1375 | * |
| 1376 | * @param string $searchTerm Term to search in the label database |
| 1377 | * @param string[] $languages List of language Zids to filter by |
| 1378 | * @param bool $renderable Whether to only filter through functions with renderable IOs |
| 1379 | * @param string[] $inputTypes List of input types to match the functions |
| 1380 | * @param string|null $outputType List of output type to match the functions |
| 1381 | * @return IResultWrapper |
| 1382 | */ |
| 1383 | public function searchFunctions( |
| 1384 | $searchTerm, |
| 1385 | $languages, |
| 1386 | $renderable = false, |
| 1387 | $inputTypes = [], |
| 1388 | $outputType = null |
| 1389 | ) { |
| 1390 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 1391 | |
| 1392 | // Subquery: Available function zids |
| 1393 | // * if renderable is true, set conditions to only renderable types |
| 1394 | // * else, set conditions to inputTypes and outputType |
| 1395 | // If no function filtering is required, $functionsQuery will be null |
| 1396 | $functionsQuery = $renderable ? |
| 1397 | $this->functionsByRenderableIOQuery() : |
| 1398 | $this->functionsByIOTypesQuery( $inputTypes, $outputType ); |
| 1399 | |
| 1400 | // Subquery: Preferred labels for all functions |
| 1401 | // * Only rank and return labels of functions |
| 1402 | // * Only return necessary fields |
| 1403 | $preferredLabelsQuery = $this->getPreferredLabelsQuery( $languages, ZTypeRegistry::Z_FUNCTION ); |
| 1404 | $preferredLabelsQuery |
| 1405 | ->clearFields() |
| 1406 | ->fields( [ 'p.page_id', 'l1.wlzl_zobject_zid', 'l1.wlzl_label', 'l1.wlzl_language' ] ); |
| 1407 | |
| 1408 | // String match condition: match tokenized searchTerm substrings (if any) |
| 1409 | $searchTerm = ZObjectUtils::comparableString( $searchTerm ); |
| 1410 | $searchColumn = ZObjectUtils::isValidZObjectReference( $searchTerm ) ? |
| 1411 | 'wlzl_zobject_zid' : 'wlzl_label_normalised'; |
| 1412 | |
| 1413 | // Create main query builder |
| 1414 | $queryBuilder = $dbr->newSelectQueryBuilder() |
| 1415 | ->select( [ |
| 1416 | 'page_id', |
| 1417 | 'lb.wlzl_zobject_zid', |
| 1418 | 'wlzl_type', |
| 1419 | 'wlzl_return_type', |
| 1420 | 'lb.wlzl_language', |
| 1421 | 'lb.wlzl_label', |
| 1422 | 'wlzl_label_primary', |
| 1423 | 'wlzl_id', |
| 1424 | 'preferred_label' => 'pl.wlzl_label', |
| 1425 | 'preferred_language' => 'pl.wlzl_language' |
| 1426 | ] ) |
| 1427 | ->from( 'wikilambda_zobject_labels', 'lb' ) |
| 1428 | // Inner join with preferred labels Subquery |
| 1429 | ->join( $preferredLabelsQuery, 'pl', 'lb.wlzl_zobject_zid = pl.wlzl_zobject_zid' ) |
| 1430 | ->where( [ 'wlzl_type' => 'Z8' ] ); |
| 1431 | |
| 1432 | // If searchTerm is not empty, add string match conditions |
| 1433 | $stringMatchCondition = $this->getStringMatchCondition( $searchColumn, $searchTerm ); |
| 1434 | if ( $stringMatchCondition ) { |
| 1435 | $queryBuilder->andWhere( $stringMatchCondition ); |
| 1436 | } |
| 1437 | |
| 1438 | // If functionsQuery is not null, Left join and filter by wlzo_main_zid IS NOT NULL condition |
| 1439 | if ( $functionsQuery ) { |
| 1440 | $queryBuilder |
| 1441 | ->leftJoin( $functionsQuery, 'fn', 'lb.wlzl_zobject_zid = fn.wlzo_main_zid' ) |
| 1442 | ->andWhere( 'fn.wlzo_main_zid IS NOT NULL' ); |
| 1443 | } |
| 1444 | |
| 1445 | // Return result set: |
| 1446 | return $queryBuilder |
| 1447 | ->caller( __METHOD__ ) |
| 1448 | ->fetchResultSet(); |
| 1449 | } |
| 1450 | |
| 1451 | /** |
| 1452 | * Generates a query that returns a table with all the existing |
| 1453 | * test objects, their function, and their status: |
| 1454 | * * is_connected: whether the test is connected to the function |
| 1455 | * * is_passing: whether the test is passing against all the |
| 1456 | * function's connected implementations |
| 1457 | * |
| 1458 | * @return string |
| 1459 |