Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 43 |
|
0.00% |
0 / 8 |
CRAP | |
0.00% |
0 / 1 |
| ExpensiveUserImpact | |
0.00% |
0 / 43 |
|
0.00% |
0 / 8 |
110 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| getDailyTotalViews | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDailyArticleViews | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| newEmpty | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
2 | |||
| loadFromJsonArray | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| filterViewCounts | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| jsonSerialize | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| isPageViewDataStale | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\UserImpact; |
| 4 | |
| 5 | use DateTime; |
| 6 | use Exception; |
| 7 | use MediaWiki\User\UserIdentity; |
| 8 | use MediaWiki\User\UserIdentityValue; |
| 9 | use MediaWiki\Utils\MWTimestamp; |
| 10 | use Wikimedia\Assert\Assert; |
| 11 | |
| 12 | /** |
| 13 | * Value object representing a user's impact statistics, including those which are |
| 14 | * more expensive to retrieve. |
| 15 | */ |
| 16 | class ExpensiveUserImpact extends UserImpact { |
| 17 | |
| 18 | /** @var int[] */ |
| 19 | private array $dailyTotalViews; |
| 20 | |
| 21 | /** @var array */ |
| 22 | private array $dailyArticleViews; |
| 23 | |
| 24 | /** |
| 25 | * @param UserIdentity $user |
| 26 | * @param int $receivedThanksCount Number of thanks the user has received. Might exclude |
| 27 | * thanks received a long time ago. |
| 28 | * @param int[] $editCountByNamespace Namespace ID => number of edits the user made in some |
| 29 | * namespace. Might exclude edits made a long time ago or many edits ago. |
| 30 | * @param int[] $editCountByDay Day => number of edits the user made on that day. Indexed with |
| 31 | * ISO 8601 dates, e.g. '2022-08-25'. Might exclude edits made many edits ago. |
| 32 | * @param array<string,int> $editCountByTaskType Number of newcomer task edits per task type. |
| 33 | * @param int $revertedEditCount Number of edits by the user that got reverted (determined by |
| 34 | * the mw-reverted tag). |
| 35 | * @param int $newcomerTaskEditCount Number of edits the user made which have the |
| 36 | * newcomer task tag. Might exclude edits made a long time ago or many edits ago. |
| 37 | * @param int|null $lastEditTimestamp Unix timestamp of the user's last edit. |
| 38 | * @param int[] $dailyTotalViews Day => number of total pageviews the articles edited by the user |
| 39 | * (on any day) got on that day. Indexed with ISO 8601 dates, e.g. '2022-08-25'. |
| 40 | * Might exclude edits made many days or many edits ago. |
| 41 | * @param array $dailyArticleViews @see ::getDailyArticleViews |
| 42 | * @param EditingStreak $longestEditingStreak |
| 43 | * @param int|null $totalUserEditCount Copy of user.user_editcount |
| 44 | */ |
| 45 | public function __construct( |
| 46 | UserIdentity $user, |
| 47 | int $receivedThanksCount, |
| 48 | array $editCountByNamespace, |
| 49 | array $editCountByDay, |
| 50 | array $editCountByTaskType, |
| 51 | int $revertedEditCount, |
| 52 | int $newcomerTaskEditCount, |
| 53 | ?int $lastEditTimestamp, |
| 54 | array $dailyTotalViews, |
| 55 | array $dailyArticleViews, |
| 56 | EditingStreak $longestEditingStreak, |
| 57 | ?int $totalUserEditCount |
| 58 | ) { |
| 59 | parent::__construct( $user, $receivedThanksCount, $editCountByNamespace, $editCountByDay, |
| 60 | $editCountByTaskType, $revertedEditCount, $newcomerTaskEditCount, $lastEditTimestamp, |
| 61 | $longestEditingStreak, $totalUserEditCount ); |
| 62 | $this->dailyTotalViews = $dailyTotalViews; |
| 63 | $this->dailyArticleViews = $dailyArticleViews; |
| 64 | } |
| 65 | |
| 66 | /** |
| 67 | * Day => number of total pageviews the articles edited by the user (on any day) got on that day. |
| 68 | * Indexed with ISO 8601 dates, e.g. '2022-08-25'. The list of days is contiguous, in ascending |
| 69 | * order, and ends more or less at the current day (might be a few days off to account for data |
| 70 | * collection lags). |
| 71 | * Might exclude edits made many days or many edits ago. |
| 72 | * @return int[] |
| 73 | */ |
| 74 | public function getDailyTotalViews(): array { |
| 75 | return $this->dailyTotalViews; |
| 76 | } |
| 77 | |
| 78 | /** |
| 79 | * An array with the title's prefixed DBkey as the key. The values contain: |
| 80 | * - firstEditDate: The date the user first edited the article, in Y-m-d format. |
| 81 | * If the user made a very high number of total edits, it might just be some edit the |
| 82 | * user made to the article, not necessarily the first. |
| 83 | * - newestEdit: The TS_MW timestamp of the newest edit by the user to the article. |
| 84 | * - views: An array of page view counts. Each row in the array has a key with the date in |
| 85 | * Y-m-d format and the value is the page view count total for that day. |
| 86 | * The list of days is contiguous, in ascending order, and ends more or less at the current |
| 87 | * day (might be a few days off to account for data collection lags). |
| 88 | * - imageUrl: URL of a thumbnail of the article's main image, or null if there's no main image. |
| 89 | * @return array[] |
| 90 | * @phan-return array<string,array{views:array<string,int>,firstEditDate:string,newestEdit:string,imageUrl:?string}> |
| 91 | */ |
| 92 | public function getDailyArticleViews(): array { |
| 93 | return $this->dailyArticleViews; |
| 94 | } |
| 95 | |
| 96 | /** @inheritDoc */ |
| 97 | protected static function newEmpty(): self { |
| 98 | return new ExpensiveUserImpact( |
| 99 | new UserIdentityValue( 0, '' ), |
| 100 | 0, |
| 101 | [], |
| 102 | [], |
| 103 | [], |
| 104 | 0, |
| 105 | 0, |
| 106 | 0, |
| 107 | [], |
| 108 | [], |
| 109 | new EditingStreak(), |
| 110 | 0 |
| 111 | ); |
| 112 | } |
| 113 | |
| 114 | /** @inheritDoc */ |
| 115 | protected function loadFromJsonArray( array $json ): void { |
| 116 | parent::loadFromJsonArray( $json ); |
| 117 | |
| 118 | Assert::parameterKeyType( 'string', $json['dailyTotalViews'], '$json[\'dailyTotalViews\']' ); |
| 119 | Assert::parameterElementType( 'integer', $json['dailyTotalViews'], '$json[\'dailyTotalViews\']' ); |
| 120 | Assert::parameterElementType( 'array', $json['dailyArticleViews'], '$json[\'dailyArticleViews\']' ); |
| 121 | foreach ( $json['dailyArticleViews'] as $title => $views ) { |
| 122 | Assert::parameterKeyType( 'string', $views, '$json[\'dailyArticleViews\'][\'' . $title . '\']' ); |
| 123 | } |
| 124 | |
| 125 | $this->dailyTotalViews = $json['dailyTotalViews']; |
| 126 | $this->dailyArticleViews = $json['dailyArticleViews']; |
| 127 | } |
| 128 | |
| 129 | /** |
| 130 | * Filter view counts to only include dates with non-zero counts |
| 131 | * |
| 132 | * @note This is a safeguard filtering. ComputedUserImpactLookup should not generate impact |
| 133 | * objects with zeros in the first place. |
| 134 | * @see https://phabricator.wikimedia.org/T351898 |
| 135 | * @param array $views |
| 136 | * @return array |
| 137 | */ |
| 138 | private function filterViewCounts( array $views ): array { |
| 139 | return array_filter( $views, static function ( $el ) { |
| 140 | return $el > 0; |
| 141 | } ); |
| 142 | } |
| 143 | |
| 144 | /** @inheritDoc */ |
| 145 | public function jsonSerialize(): array { |
| 146 | return parent::jsonSerialize() + [ |
| 147 | 'dailyTotalViews' => $this->filterViewCounts( $this->dailyTotalViews ), |
| 148 | 'dailyArticleViews' => $this->filterViewCounts( $this->dailyArticleViews ), |
| 149 | ]; |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * @return bool |
| 154 | * @throws Exception |
| 155 | */ |
| 156 | public function isPageViewDataStale(): bool { |
| 157 | $dailyTotalViews = $this->getDailyTotalViews(); |
| 158 | if ( $dailyTotalViews === [] ) { |
| 159 | return true; |
| 160 | } |
| 161 | |
| 162 | $latestPageViewsDateTime = new DateTime( array_key_last( $dailyTotalViews ) ); |
| 163 | $now = MWTimestamp::getInstance(); |
| 164 | $diff = $now->timestamp->diff( $latestPageViewsDateTime ); |
| 165 | // Page view data generation can lag by 24-48 hours. |
| 166 | // Consider the data stale if it's from before (UTC) yesterday. |
| 167 | return $diff->days > 1; |
| 168 | } |
| 169 | |
| 170 | } |