Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
80.00% |
48 / 60 |
|
25.00% |
1 / 4 |
CRAP | |
0.00% |
0 / 1 |
| RefreshUserImpactJob | |
80.00% |
48 / 60 |
|
25.00% |
1 / 4 |
21.89 | |
0.00% |
0 / 1 |
| __construct | |
93.33% |
14 / 15 |
|
0.00% |
0 / 1 |
2.00 | |||
| run | |
96.43% |
27 / 28 |
|
0.00% |
0 / 1 |
13 | |||
| isFresh | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| computeUserImpact | |
37.50% |
6 / 16 |
|
0.00% |
0 / 1 |
5.20 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\UserImpact; |
| 4 | |
| 5 | use GenericParameterJob; |
| 6 | use GrowthExperiments\GrowthExperimentsServices; |
| 7 | use Job; |
| 8 | use MediaWiki\Logger\LoggerFactory; |
| 9 | use MediaWiki\MediaWikiServices; |
| 10 | use MediaWiki\User\UserFactory; |
| 11 | use MediaWiki\User\UserIdentityLookup; |
| 12 | use MediaWiki\Utils\MWTimestamp; |
| 13 | use Psr\Log\LoggerInterface; |
| 14 | use Wikimedia\Assert\Assert; |
| 15 | use Wikimedia\Assert\ParameterAssertionException; |
| 16 | use Wikimedia\LightweightObjectStore\ExpirationAwareness; |
| 17 | |
| 18 | /** |
| 19 | * Job for computing and caching expensive user impact data. Can also be used to refresh the cache |
| 20 | * with an already computed value. |
| 21 | */ |
| 22 | class RefreshUserImpactJob extends Job implements GenericParameterJob { |
| 23 | |
| 24 | private UserImpactStore $userImpactStore; |
| 25 | private UserImpactLookup $userImpactLookup; |
| 26 | private UserFactory $userFactory; |
| 27 | private UserImpactFormatter $userImpactFormatter; |
| 28 | private UserIdentityLookup $userIdentityLookup; |
| 29 | private LoggerInterface $logger; |
| 30 | |
| 31 | /** |
| 32 | * Map of user ID => impact data as JSON string, or null to generate in the job |
| 33 | * @var (string|null)[] |
| 34 | */ |
| 35 | private array $impactDataBatch; |
| 36 | |
| 37 | /** |
| 38 | * Cached objects generated before this UNIX timestamp are considered stale and recomputed. |
| 39 | * Only used when no impact data is provided for the given user. |
| 40 | * @var int |
| 41 | */ |
| 42 | private int $staleBefore; |
| 43 | |
| 44 | /** |
| 45 | * @inheritDoc |
| 46 | * Parameters: |
| 47 | * - impactDataBatch: user impact data to write/compute, see self::$impactDataBatch |
| 48 | * - staleBefore: staleness limit, see self::$staleBefore; optional, defaults to 1 day ago |
| 49 | * - userId: user to refresh data for (deprecated, required if impactDataBatch not present) |
| 50 | * - impactData: impact data for userId (deprecated) |
| 51 | */ |
| 52 | public function __construct( $params = null ) { |
| 53 | parent::__construct( 'refreshUserImpactJob', $params ); |
| 54 | |
| 55 | $services = MediaWikiServices::getInstance(); |
| 56 | $growthServices = GrowthExperimentsServices::wrap( $services ); |
| 57 | $this->userImpactStore = $growthServices->getUserImpactStore(); |
| 58 | $this->userImpactLookup = $growthServices->getUserImpactLookup(); |
| 59 | $this->userImpactFormatter = $growthServices->getUserImpactFormatter(); |
| 60 | $this->userIdentityLookup = $services->getUserIdentityLookup(); |
| 61 | $this->userFactory = $services->getUserFactory(); |
| 62 | $this->logger = LoggerFactory::getInstance( 'GrowthExperiments' ); |
| 63 | |
| 64 | $this->impactDataBatch = $params['impactDataBatch'] |
| 65 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 66 | ?? [ $params['userId'] => $params['impactData'] ?? null ]; |
| 67 | $this->staleBefore = $params['staleBefore'] ?? MWTimestamp::time() - ExpirationAwareness::TTL_DAY; |
| 68 | // Prevent accidental use of TS_MW or some other non-TS_UNIX format but don't require int type |
| 69 | // as e.g. wfTimestamp( TS_UNIX ) returns a string. |
| 70 | Assert::parameter( is_numeric( $this->staleBefore ) && $this->staleBefore < 2147483647, |
| 71 | 'staleBefore', 'must be a UNIX timestamp' ); |
| 72 | } |
| 73 | |
| 74 | /** @inheritDoc */ |
| 75 | public function run() { |
| 76 | $preloadedUserImpacts = []; |
| 77 | if ( $this->userImpactStore instanceof DatabaseUserImpactStore ) { |
| 78 | $preloadedUserImpacts = $this->userImpactStore->batchGetUserImpact( |
| 79 | array_keys( $this->impactDataBatch ) |
| 80 | ); |
| 81 | } |
| 82 | foreach ( $this->impactDataBatch as $userId => $impactJson ) { |
| 83 | if ( $this->userFactory->newFromId( $userId )->isHidden() ) { |
| 84 | // do not update impact data for hidden users (T337845) |
| 85 | continue; |
| 86 | } |
| 87 | |
| 88 | $userImpact = null; |
| 89 | /** @var UserImpact $preloadedUserImpact */ |
| 90 | $preloadedUserImpact = $preloadedUserImpacts[$userId] ?? null; |
| 91 | if ( $impactJson ) { |
| 92 | try { |
| 93 | $userImpact = UserImpact::newFromJsonArray( json_decode( $impactJson, true ) ); |
| 94 | // Do not update the cache if it is already more recent. |
| 95 | if ( $preloadedUserImpact |
| 96 | && $preloadedUserImpact->getGeneratedAt() > $userImpact->getGeneratedAt() |
| 97 | ) { |
| 98 | continue; |
| 99 | } |
| 100 | } catch ( ParameterAssertionException $parameterAssertionException ) { |
| 101 | // Invalid cache format used, recalculate from scratch. |
| 102 | } |
| 103 | } elseif ( $preloadedUserImpact && $this->isFresh( $preloadedUserImpact ) ) { |
| 104 | // We haven't been explicitly told to save new data, and the existing data |
| 105 | // is still usable, nothing to do. |
| 106 | continue; |
| 107 | } |
| 108 | |
| 109 | if ( !$userImpact || !$this->isFresh( $userImpact ) ) { |
| 110 | $userImpact = $this->computeUserImpact( $userId ); |
| 111 | } |
| 112 | |
| 113 | if ( $userImpact ) { |
| 114 | // We don't want to cache all page view data captured by ::computeUserImpact; in a job queue |
| 115 | // context, this can contain up to 1000 articles of PageViewData (configured via |
| 116 | // GEUserImpactMaxArticlesToProcessForPageviews). Call |
| 117 | // the formatter to get just the data we need, and replace the dailyArticleViews with just the |
| 118 | // top entries. |
| 119 | $jsonData = $userImpact->jsonSerialize(); |
| 120 | $sortedAndFiltered = $this->userImpactFormatter->sortAndFilter( $jsonData ); |
| 121 | $jsonData['dailyArticleViews'] = |
| 122 | // Make sure dailyArticleViews includes both the top viewed articles and recently edited |
| 123 | // articles without page views. Those will both be used by UserImpactFormatter again when |
| 124 | // fetching the data to display. |
| 125 | $sortedAndFiltered['topViewedArticles'] + $sortedAndFiltered['recentEditsWithoutPageviews']; |
| 126 | $userImpact = UserImpact::newFromJsonArray( $jsonData ); |
| 127 | $this->userImpactStore->setUserImpact( $userImpact ); |
| 128 | } |
| 129 | } |
| 130 | return true; |
| 131 | } |
| 132 | |
| 133 | /** |
| 134 | * @param UserImpact $impact |
| 135 | * @return bool |
| 136 | */ |
| 137 | private function isFresh( UserImpact $impact ): bool { |
| 138 | return $impact->getGeneratedAt() >= $this->staleBefore; |
| 139 | } |
| 140 | |
| 141 | /** |
| 142 | * @param int $userId |
| 143 | * @return ExpensiveUserImpact|null |
| 144 | */ |
| 145 | private function computeUserImpact( int $userId ): ?ExpensiveUserImpact { |
| 146 | $loggerParams = [ 'userId' => $userId ]; |
| 147 | $userIdentity = $this->userIdentityLookup->getUserIdentityByUserId( $userId ); |
| 148 | if ( !$userIdentity ) { |
| 149 | $this->logger->error( |
| 150 | 'Unable to get user identity in RefreshUserImpactJob.', |
| 151 | $loggerParams |
| 152 | ); |
| 153 | return null; |
| 154 | } |
| 155 | |
| 156 | $userImpact = $this->userImpactLookup->getExpensiveUserImpact( $userIdentity ); |
| 157 | if ( !$userImpact ) { |
| 158 | $this->logger->error( |
| 159 | 'Unable to generate user impact for user in RefreshUserImpactJob.', |
| 160 | $loggerParams |
| 161 | ); |
| 162 | return null; |
| 163 | } |
| 164 | |
| 165 | return $userImpact; |
| 166 | } |
| 167 | |
| 168 | } |