Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
98.53% |
67 / 68 |
|
83.33% |
5 / 6 |
CRAP | |
0.00% |
0 / 1 |
| UserImpactFormatter | |
98.53% |
67 / 68 |
|
83.33% |
5 / 6 |
17 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| format | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
2.01 | |||
| sortAndFilter | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
6 | |||
| getModifiedDailyArticleViews | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| fillDailyArticleViewsWithPageViewToolsUrl | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| getPageViewToolsUrl | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\UserImpact; |
| 4 | |
| 5 | use DateTime; |
| 6 | use Exception; |
| 7 | use MediaWiki\Utils\MWTimestamp; |
| 8 | use stdClass; |
| 9 | |
| 10 | /** |
| 11 | * Formats an ExpensiveUserImpact object to be more suitable for frontend use. |
| 12 | */ |
| 13 | class UserImpactFormatter { |
| 14 | |
| 15 | private const PAGEVIEW_TOOL_BASE_URL = 'https://pageviews.wmcloud.org/'; |
| 16 | |
| 17 | private stdClass $AQSConfig; |
| 18 | |
| 19 | /** |
| 20 | * @param stdClass $AQSConfig |
| 21 | */ |
| 22 | public function __construct( |
| 23 | stdClass $AQSConfig |
| 24 | ) { |
| 25 | $this->AQSConfig = $AQSConfig; |
| 26 | } |
| 27 | |
| 28 | /** |
| 29 | * Create a new UserImpactFormatter from a serialized ExpensiveUserImpact. |
| 30 | * @param array|ExpensiveUserImpact $userImpact |
| 31 | * @param string $languageCode The requesting user's language code to use in the pageviews url construction. |
| 32 | * @return array |
| 33 | * @throws Exception |
| 34 | */ |
| 35 | public function format( $userImpact, string $languageCode ): array { |
| 36 | if ( $userImpact instanceof ExpensiveUserImpact ) { |
| 37 | $jsonData = $userImpact->jsonSerialize(); |
| 38 | } else { |
| 39 | $jsonData = $userImpact; |
| 40 | } |
| 41 | $jsonData += $this->sortAndFilter( $jsonData ); |
| 42 | unset( $jsonData['dailyArticleViews'] ); |
| 43 | $this->fillDailyArticleViewsWithPageViewToolsUrl( $jsonData, $languageCode ); |
| 44 | return $jsonData; |
| 45 | } |
| 46 | |
| 47 | /** |
| 48 | * Calculate the topViewedArticles, topViewedArticlesCount and recentEditsWithoutPageviews |
| 49 | * fields by sorting and filtering the daily article views data: |
| 50 | * - Get the top 5 most viewed articles, in descending order of views |
| 51 | * - Get up to 5 of the most recently edited articles with no page view |
| 52 | * data available yet, in descending order of recency |
| 53 | * |
| 54 | * Note that, in some situations, views/viewsCount information doesn't exist, because this |
| 55 | * method removed those fields from data, and that data was used as the basis for constructing |
| 56 | * a new ExpensiveUserImpact object. We can work around that with a few checks for the |
| 57 | * existence of the properties, before using them. |
| 58 | * |
| 59 | * @param array $jsonData |
| 60 | * @return array |
| 61 | */ |
| 62 | public function sortAndFilter( array $jsonData ): array { |
| 63 | $topViewedArticles = $recentEditsWithoutPageviews = []; |
| 64 | foreach ( $this->getModifiedDailyArticleViews( $jsonData ) as $title => $data ) { |
| 65 | $lastDayWithPageViewData = array_key_last( $data['views'] ?? [] ); |
| 66 | // See if we have pageview data for the page. |
| 67 | $noPageviewDataYet = $data['firstEditDate'] > $lastDayWithPageViewData |
| 68 | // The last day actually might or might not have data (T217286) so allow equality |
| 69 | // if there's no data for that day. (We can't differentiate between no data and |
| 70 | // legitimately 0 views, but it's not really possible to edit a page and not |
| 71 | // generate any pageviews.) |
| 72 | || ( $data['firstEditDate'] === $lastDayWithPageViewData |
| 73 | && $data['views'][$lastDayWithPageViewData] === 0 ); |
| 74 | |
| 75 | if ( $noPageviewDataYet ) { |
| 76 | $recentEditsWithoutPageviews[$title] = $data; |
| 77 | } else { |
| 78 | $topViewedArticles[$title] = $data; |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | // Order the articles by most views to fewest |
| 83 | uasort( $topViewedArticles, static function ( $a, $b ) { |
| 84 | return $b['viewsCount'] <=> $a['viewsCount']; |
| 85 | } ); |
| 86 | // Get the top five articles in the list that have page views |
| 87 | $topViewedArticles = array_slice( array_filter( $topViewedArticles, static function ( $item ) { |
| 88 | return ( $item['viewsCount'] ?? 0 ) > 0; |
| 89 | } ), 0, 5, true ); |
| 90 | |
| 91 | $topViewedArticlesCount = array_sum( array_column( $topViewedArticles, 'viewsCount' ) ); |
| 92 | $totalPageviewsCount = array_sum( |
| 93 | array_column( $this->getModifiedDailyArticleViews( $jsonData ), 'viewsCount' ) |
| 94 | ); |
| 95 | |
| 96 | // Order the articles by date, most recent edit to oldest, and get the most recent 5. |
| 97 | uasort( $recentEditsWithoutPageviews, static function ( $a, $b ) { |
| 98 | return $b['newestEdit'] <=> $a['newestEdit']; |
| 99 | } ); |
| 100 | $recentEditsWithoutPageviews = array_slice( $recentEditsWithoutPageviews, 0, 5, true ); |
| 101 | // Remove the 'viewsCount' key - the frontend will show this as still waiting for pageview data. |
| 102 | // Also unset 'views' as a micro-optimization. |
| 103 | foreach ( $recentEditsWithoutPageviews as $title => $_ ) { |
| 104 | unset( $recentEditsWithoutPageviews[$title]['viewsCount'] ); |
| 105 | unset( $recentEditsWithoutPageviews[$title]['views'] ); |
| 106 | } |
| 107 | |
| 108 | return [ |
| 109 | 'recentEditsWithoutPageviews' => $recentEditsWithoutPageviews, |
| 110 | 'topViewedArticles' => $topViewedArticles, |
| 111 | 'topViewedArticlesCount' => $topViewedArticlesCount, |
| 112 | 'totalPageviewsCount' => $totalPageviewsCount |
| 113 | ]; |
| 114 | } |
| 115 | |
| 116 | /** |
| 117 | * Returns dailyArticleViews field, with views set to 0 on days before the user's first edit |
| 118 | * to the article, and a 'viewsCount' field added to each title with the sum of the |
| 119 | * (remaining) views. |
| 120 | * @param array $jsonData |
| 121 | * @return array |
| 122 | */ |
| 123 | private function getModifiedDailyArticleViews( array $jsonData ): array { |
| 124 | $dailyArticleViews = $jsonData['dailyArticleViews']; |
| 125 | foreach ( $dailyArticleViews as $title => $data ) { |
| 126 | foreach ( $data['views'] ?? [] as $date => $dailyViews ) { |
| 127 | if ( $date < $data['firstEditDate'] ) { |
| 128 | // Note this is unreliable for established users, as we look at the user's |
| 129 | // last 1000 edits to determine firstEditDate. We ignore that issue here. |
| 130 | $dailyArticleViews[$title]['views'][$date] = 0; |
| 131 | } |
| 132 | } |
| 133 | $dailyArticleViews[$title]['viewsCount'] = array_sum( $dailyArticleViews[$title]['views'] ?? [] ); |
| 134 | } |
| 135 | return $dailyArticleViews; |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * @param array &$jsonData |
| 140 | * @param string $languageCode The requesting user's language code to use in the pageviews url |
| 141 | * @return void |
| 142 | * @throws Exception |
| 143 | */ |
| 144 | private function fillDailyArticleViewsWithPageViewToolsUrl( |
| 145 | array &$jsonData, |
| 146 | string $languageCode |
| 147 | ): void { |
| 148 | foreach ( $jsonData['topViewedArticles'] as $title => $articleData ) { |
| 149 | $latestPageViewDate = array_key_last( $articleData['views'] ); |
| 150 | $jsonData['topViewedArticles'][$title]['pageviewsUrl'] = $this->getPageViewToolsUrl( |
| 151 | $title, $articleData['firstEditDate'], $latestPageViewDate, $languageCode |
| 152 | ); |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | /** |
| 157 | * @param string $title |
| 158 | * @param string $firstEditDate Date of the first edit to the article in Y-m-d format. |
| 159 | * @param string $latestPageViewDate Date of the most last page view data entry available for this article. |
| 160 | * Used for constructing the 'end' parameter for the URL, to avoid confusion with timezones and what "latest" |
| 161 | * means in the context of the pageviews application and Analytics Query Service. |
| 162 | * @param string $languageCode The requesting user's language code to use in the pageviews url |
| 163 | * @return string Full URL for the PageViews tool for the given title and start date |
| 164 | * @throws Exception |
| 165 | */ |
| 166 | private function getPageViewToolsUrl( |
| 167 | string $title, |
| 168 | string $firstEditDate, |
| 169 | string $latestPageViewDate, |
| 170 | string $languageCode |
| 171 | ): string { |
| 172 | $daysAgo = ComputedUserImpactLookup::PAGEVIEW_DAYS; |
| 173 | $dtiAgo = new DateTime( '@' . strtotime( "-$daysAgo days", MWTimestamp::time() ) ); |
| 174 | $startDate = $dtiAgo->format( 'Y-m-d' ); |
| 175 | if ( $firstEditDate > $startDate ) { |
| 176 | $startDate = $firstEditDate; |
| 177 | } |
| 178 | return wfAppendQuery( self::PAGEVIEW_TOOL_BASE_URL, [ |
| 179 | 'project' => $this->AQSConfig->project, |
| 180 | 'userlang' => $languageCode, |
| 181 | 'start' => $startDate, |
| 182 | 'end' => $latestPageViewDate, |
| 183 | 'pages' => $title, |
| 184 | ] ); |
| 185 | } |
| 186 | |
| 187 | } |