Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 134 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| RefreshUserImpactData | |
0.00% |
0 / 128 |
|
0.00% |
0 / 7 |
1406 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 44 |
|
0.00% |
0 / 1 |
240 | |||
| initOptions | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| initServices | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| getUsers | |
0.00% |
0 / 30 |
|
0.00% |
0 / 1 |
90 | |||
| getQueryBuilder | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
30 | |||
| getTimestampFromRelativeDate | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\Maintenance; |
| 4 | |
| 5 | use DateTime; |
| 6 | use Exception; |
| 7 | use Generator; |
| 8 | use GrowthExperiments\GrowthExperimentsServices; |
| 9 | use GrowthExperiments\UserDatabaseHelper; |
| 10 | use GrowthExperiments\UserImpact\RefreshUserImpactJob; |
| 11 | use GrowthExperiments\UserImpact\UserImpactLookup; |
| 12 | use GrowthExperiments\UserImpact\UserImpactStore; |
| 13 | use Maintenance; |
| 14 | use MediaWiki\JobQueue\JobQueueGroupFactory; |
| 15 | use MediaWiki\MediaWikiServices; |
| 16 | use MediaWiki\User\ActorStore; |
| 17 | use MediaWiki\User\UserFactory; |
| 18 | use MediaWiki\User\UserIdentity; |
| 19 | use MediaWiki\User\UserSelectQueryBuilder; |
| 20 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 21 | |
| 22 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 23 | if ( $IP === false ) { |
| 24 | $IP = __DIR__ . '/../../..'; |
| 25 | } |
| 26 | require_once "$IP/maintenance/Maintenance.php"; |
| 27 | |
| 28 | class RefreshUserImpactData extends Maintenance { |
| 29 | |
| 30 | private ActorStore $actorStore; |
| 31 | private UserFactory $userFactory; |
| 32 | private UserImpactLookup $userImpactLookup; |
| 33 | private UserImpactStore $userImpactStore; |
| 34 | private UserDatabaseHelper $userDatabaseHelper; |
| 35 | |
| 36 | private JobQueueGroupFactory $jobQueueGroupFactory; |
| 37 | |
| 38 | /** @var int|null Ignore a user if they have data generated after this Unix timestamp. */ |
| 39 | private ?int $ignoreAfter = null; |
| 40 | |
| 41 | private int $totalUsers = 0; |
| 42 | |
| 43 | public function __construct() { |
| 44 | parent::__construct(); |
| 45 | $this->requireExtension( 'GrowthExperiments' ); |
| 46 | $this->addDescription( 'Update data in the growthexperiments_user_impact table.' ); |
| 47 | $this->addOption( 'editedWithin', 'Apply to users who have edited within the given time.' |
| 48 | . ' Time is a relative timestring fragment passed to DateTime, such as "30days".', false, true ); |
| 49 | $this->addOption( 'registeredWithin', 'Apply to users who have registered within the given time.' |
| 50 | . ' Time is a relative timestring fragment passed to DateTime, such as "30days".', false, true ); |
| 51 | $this->addOption( 'hasEditsAtLeast', 'Apply to users who have at least this many edits.', false, true ); |
| 52 | $this->addOption( 'ignoreIfUpdatedWithin', 'Skip cache records which were stored within the given time.' |
| 53 | . ' Time is a relative timestring fragment passed to DateTime, such as "30days".', false, true ); |
| 54 | $this->addOption( 'fromUser', 'Continue from the given user ID (exclusive).', false, true ); |
| 55 | $this->addOption( 'use-job-queue', 'If job queue should be used to refresh user impact data.' ); |
| 56 | $this->addOption( 'force', 'Run even if GERefreshUserImpactDataMaintenanceScriptEnabled is false' ); |
| 57 | $this->addOption( 'dry-run', 'When used, the script will only count the number of users it would update.' ); |
| 58 | $this->addOption( 'verbose', 'Verbose mode' ); |
| 59 | $this->setBatchSize( 100 ); |
| 60 | } |
| 61 | |
| 62 | /** @inheritDoc */ |
| 63 | public function execute() { |
| 64 | if ( !$this->getConfig()->get( 'GERefreshUserImpactDataMaintenanceScriptEnabled' ) |
| 65 | && !$this->hasOption( 'force' ) |
| 66 | ) { |
| 67 | $this->output( |
| 68 | 'GERefreshUserImpactDataMaintenanceScriptEnabled is set to false on this wiki.' . |
| 69 | PHP_EOL |
| 70 | ); |
| 71 | return; |
| 72 | } |
| 73 | $this->initOptions(); |
| 74 | $this->initServices(); |
| 75 | |
| 76 | $users = []; |
| 77 | foreach ( $this->getUsers() as $user ) { |
| 78 | $realUser = $this->userFactory->newFromUserIdentity( $user ); |
| 79 | if ( $realUser->isHidden() ) { |
| 80 | // do not update impact data for hidden users (T337845) |
| 81 | $this->output( " ...skipping user {$user->getId()}, hidden.\n" ); |
| 82 | continue; |
| 83 | } |
| 84 | if ( $realUser->isBot() ) { |
| 85 | // do not update impact data for bots (T351898) |
| 86 | $this->output( " ...skipping user {$user->getId()}, bot.\n" ); |
| 87 | continue; |
| 88 | } |
| 89 | |
| 90 | if ( $this->hasOption( 'dry-run' ) ) { |
| 91 | if ( $this->hasOption( 'verbose' ) ) { |
| 92 | $this->output( " ...would refresh user impact for user {$user->getId()}\n" ); |
| 93 | } |
| 94 | continue; |
| 95 | } elseif ( $this->hasOption( 'use-job-queue' ) ) { |
| 96 | $users[$user->getId()] = null; |
| 97 | if ( count( $users ) >= $this->getBatchSize() ) { |
| 98 | if ( $this->hasOption( 'verbose' ) ) { |
| 99 | $usersText = implode( ', ', array_keys( $users ) ); |
| 100 | $this->output( " ... enqueueing refreshUserImpactJob for users $usersText\n" ); |
| 101 | } |
| 102 | $this->jobQueueGroupFactory->makeJobQueueGroup()->lazyPush( |
| 103 | new RefreshUserImpactJob( [ |
| 104 | 'impactDataBatch' => $users, |
| 105 | 'staleBefore' => $this->ignoreAfter, |
| 106 | ] ) |
| 107 | ); |
| 108 | $users = []; |
| 109 | } |
| 110 | } else { |
| 111 | $userImpact = $this->userImpactLookup->getExpensiveUserImpact( $user ); |
| 112 | if ( $userImpact ) { |
| 113 | if ( $this->hasOption( 'verbose' ) ) { |
| 114 | $this->output( " ...refreshing user impact for user {$user->getId()}\n" ); |
| 115 | } |
| 116 | $this->userImpactStore->setUserImpact( $userImpact ); |
| 117 | } elseif ( $this->hasOption( 'verbose' ) ) { |
| 118 | $this->output( " ...could not generate user impact for user {$user->getId()}\n" ); |
| 119 | } |
| 120 | } |
| 121 | } |
| 122 | |
| 123 | if ( $this->totalUsers ) { |
| 124 | $this->output( "Done. Processed $this->totalUsers users.\n" ); |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | private function initOptions(): void { |
| 129 | if ( !$this->hasOption( 'editedWithin' ) && !$this->hasOption( 'registeredWithin' ) ) { |
| 130 | $this->fatalError( 'must use at least one of --editedWithin and --registeredWithin' ); |
| 131 | } |
| 132 | |
| 133 | $ignoreIfUpdatedWithin = $this->getOption( 'ignoreIfUpdatedWithin' ); |
| 134 | if ( $ignoreIfUpdatedWithin ) { |
| 135 | $this->ignoreAfter = $this->getTimestampFromRelativeDate( $ignoreIfUpdatedWithin ); |
| 136 | } |
| 137 | } |
| 138 | |
| 139 | private function initServices(): void { |
| 140 | $services = MediaWikiServices::getInstance(); |
| 141 | $growthServices = GrowthExperimentsServices::wrap( $services ); |
| 142 | $this->actorStore = $services->getActorStore(); |
| 143 | $this->userFactory = $services->getUserFactory(); |
| 144 | $this->jobQueueGroupFactory = $services->getJobQueueGroupFactory(); |
| 145 | $this->userImpactLookup = $growthServices->getUncachedUserImpactLookup(); |
| 146 | $this->userImpactStore = $growthServices->getUserImpactStore(); |
| 147 | $this->userDatabaseHelper = $growthServices->getUserDatabaseHelper(); |
| 148 | } |
| 149 | |
| 150 | /** |
| 151 | * @return Generator<UserIdentity> |
| 152 | */ |
| 153 | private function getUsers(): Generator { |
| 154 | $queryBuilder = $this->getQueryBuilder(); |
| 155 | $queryBuilder->select( 'actor_user' ); |
| 156 | $queryBuilder->limit( $this->getBatchSize() ); |
| 157 | $queryBuilder->orderByUserId( SelectQueryBuilder::SORT_ASC ); |
| 158 | $lastUserId = (int)$this->getOption( 'fromUser', 0 ); |
| 159 | $dbr = $this->getDB( DB_REPLICA ); |
| 160 | do { |
| 161 | $this->output( "processing {$this->getBatchSize()} users starting with $lastUserId\n" ); |
| 162 | $batchQueryBuilder = clone $queryBuilder; |
| 163 | $batchQueryBuilder->where( $dbr->expr( 'actor_user', '>', $lastUserId ) ); |
| 164 | $userIds = $batchQueryBuilder->fetchFieldValues(); |
| 165 | if ( $userIds ) { |
| 166 | $users = $this->actorStore->newSelectQueryBuilder( $dbr ) |
| 167 | ->whereUserIds( $userIds ) |
| 168 | ->fetchUserIdentities(); |
| 169 | } else { |
| 170 | $users = []; |
| 171 | } |
| 172 | foreach ( $users as $user ) { |
| 173 | $lastUserId = $user->getId(); |
| 174 | // Do staleness check, if we are not using the job queue. Jobs can run after |
| 175 | // significant delays and multiple updates for the same user might get queued, |
| 176 | // so we do the check when the job runs. |
| 177 | if ( $this->ignoreAfter && !$this->hasOption( 'use-job-queue' ) ) { |
| 178 | $cachedUserImpact = $this->userImpactStore->getExpensiveUserImpact( $user ); |
| 179 | if ( $cachedUserImpact && $cachedUserImpact->getGeneratedAt() >= $this->ignoreAfter ) { |
| 180 | if ( $this->hasOption( 'verbose' ) ) { |
| 181 | $this->output( " ...skipping user {$user->getId()}, has recent cached entry\n" ); |
| 182 | } |
| 183 | continue; |
| 184 | } |
| 185 | } |
| 186 | yield $user; |
| 187 | } |
| 188 | $this->waitForReplication(); |
| 189 | $usersProcessedInThisBatch = count( $userIds ); |
| 190 | $this->totalUsers += $usersProcessedInThisBatch; |
| 191 | if ( $usersProcessedInThisBatch > 0 ) { |
| 192 | $this->output( " processed $usersProcessedInThisBatch users\n" ); |
| 193 | } |
| 194 | } while ( $usersProcessedInThisBatch === $this->getBatchSize() ); |
| 195 | } |
| 196 | |
| 197 | private function getQueryBuilder(): UserSelectQueryBuilder { |
| 198 | $editedWithin = $this->getOption( 'editedWithin' ); |
| 199 | $registeredWithin = $this->getOption( 'registeredWithin' ); |
| 200 | $hasEditsAtLeast = $this->getOption( 'hasEditsAtLeast' ); |
| 201 | |
| 202 | $dbr = $this->getDB( DB_REPLICA ); |
| 203 | $queryBuilder = $this->actorStore->newSelectQueryBuilder( $dbr ); |
| 204 | if ( $editedWithin ) { |
| 205 | $timestamp = $dbr->timestamp( $this->getTimestampFromRelativeDate( $editedWithin ) ); |
| 206 | $queryBuilder->join( 'revision', null, [ 'rev_actor = actor_id' ] ); |
| 207 | $queryBuilder->where( $dbr->expr( 'rev_timestamp', '>=', $timestamp ) ); |
| 208 | $queryBuilder->groupBy( [ 'actor_user' ] ); |
| 209 | } |
| 210 | if ( $registeredWithin ) { |
| 211 | $firstUserId = $this->userDatabaseHelper->findFirstUserIdForRegistrationTimestamp( |
| 212 | $this->getTimestampFromRelativeDate( $registeredWithin ) |
| 213 | ); |
| 214 | if ( $firstUserId ) { |
| 215 | $queryBuilder->where( $dbr->expr( 'actor_user', '>=', $firstUserId ) ); |
| 216 | } else { |
| 217 | $queryBuilder->where( '0 = 1' ); |
| 218 | } |
| 219 | } |
| 220 | if ( $hasEditsAtLeast ) { |
| 221 | $queryBuilder->join( 'user', null, [ 'user_id = actor_user' ] ); |
| 222 | $queryBuilder->where( $dbr->expr( 'user_editcount', '>=', (int)$hasEditsAtLeast ) ); |
| 223 | } |
| 224 | return $queryBuilder; |
| 225 | } |
| 226 | |
| 227 | /** |
| 228 | * @param string $relativeDate A relative date string fragment that will be prefixed with a |
| 229 | * minus sign and passed to the DateTime constructor. |
| 230 | * @return int |
| 231 | */ |
| 232 | private function getTimestampFromRelativeDate( string $relativeDate ): int { |
| 233 | try { |
| 234 | $dateTime = new DateTime( 'now - ' . $relativeDate ); |
| 235 | } catch ( Exception $e ) { |
| 236 | $this->fatalError( $e->getMessage() ); |
| 237 | } |
| 238 | return $dateTime->getTimestamp(); |
| 239 | } |
| 240 | |
| 241 | } |
| 242 | |
| 243 | $maintClass = RefreshUserImpactData::class; |
| 244 | require_once RUN_MAINTENANCE_IF_MAIN; |