Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
78.24% |
205 / 262 |
|
27.27% |
3 / 11 |
CRAP | |
0.00% |
0 / 1 |
| ComputedUserImpactLookup | |
78.24% |
205 / 262 |
|
27.27% |
3 / 11 |
86.15 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| getUserImpact | |
94.12% |
16 / 17 |
|
0.00% |
0 / 1 |
3.00 | |||
| getExpensiveUserImpact | |
89.19% |
33 / 37 |
|
0.00% |
0 / 1 |
6.05 | |||
| getEditData | |
93.83% |
76 / 81 |
|
0.00% |
0 / 1 |
12.03 | |||
| getThanksCount | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getPageViewData | |
94.87% |
37 / 39 |
|
0.00% |
0 / 1 |
10.01 | |||
| getPageViewDataInJobContext | |
58.62% |
17 / 29 |
|
0.00% |
0 / 1 |
10.47 | |||
| getPageViewDataInWebRequestContext | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
20 | |||
| logPageDataBadStatus | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| updateToIso8601DateKeys | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| getImage | |
28.57% |
4 / 14 |
|
0.00% |
0 / 1 |
19.12 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\UserImpact; |
| 4 | |
| 5 | use ChangeTags; |
| 6 | use DateTime; |
| 7 | use DBAccessObjectUtils; |
| 8 | use ExtensionRegistry; |
| 9 | use GrowthExperiments\NewcomerTasks\ConfigurationLoader\ConfigurationLoader; |
| 10 | use GrowthExperiments\NewcomerTasks\TaskType\TaskTypeHandlerRegistry; |
| 11 | use IBufferingStatsdDataFactory; |
| 12 | use IDBAccessObject; |
| 13 | use LogicException; |
| 14 | use MediaWiki\Config\ServiceOptions; |
| 15 | use MediaWiki\Extension\PageViewInfo\PageViewService; |
| 16 | use MediaWiki\Extension\Thanks\ThanksQueryHelper; |
| 17 | use MediaWiki\MainConfigNames; |
| 18 | use MediaWiki\Revision\RevisionRecord; |
| 19 | use MediaWiki\Status\Status; |
| 20 | use MediaWiki\Storage\NameTableAccessException; |
| 21 | use MediaWiki\Storage\NameTableStore; |
| 22 | use MediaWiki\Title\MalformedTitleException; |
| 23 | use MediaWiki\Title\Title; |
| 24 | use MediaWiki\Title\TitleFactory; |
| 25 | use MediaWiki\Title\TitleFormatter; |
| 26 | use MediaWiki\Title\TitleValue; |
| 27 | use MediaWiki\User\User; |
| 28 | use MediaWiki\User\UserEditTracker; |
| 29 | use MediaWiki\User\UserFactory; |
| 30 | use MediaWiki\User\UserIdentity; |
| 31 | use MediaWiki\Utils\MWTimestamp; |
| 32 | use PageImages\PageImages; |
| 33 | use Psr\Log\LoggerInterface; |
| 34 | use Psr\Log\NullLogger; |
| 35 | use StatusValue; |
| 36 | use Wikimedia\Rdbms\IConnectionProvider; |
| 37 | |
| 38 | class ComputedUserImpactLookup implements UserImpactLookup { |
| 39 | |
| 40 | public const CONSTRUCTOR_OPTIONS = [ |
| 41 | MainConfigNames::LocalTZoffset, |
| 42 | 'GEUserImpactMaxArticlesToProcessForPageviews', |
| 43 | 'GEUserImpactMaximumProcessTimeSeconds', |
| 44 | ]; |
| 45 | |
| 46 | /** |
| 47 | * Size in pixels of the thumb image to request to PageImages. Matches the Codex |
| 48 | * thumbnail component size it is rendered in. Used in the articles list (ArticlesList.vue) |
| 49 | * in the impact module. |
| 50 | */ |
| 51 | private const THUMBNAIL_SIZE = 40; |
| 52 | |
| 53 | /** Cutoff for edit statistics. See also DATA_ROWS_LIMIT in ScoreCards.vue. */ |
| 54 | private const MAX_EDITS = 1000; |
| 55 | |
| 56 | /** Cutoff for thanks count. See also DATA_ROWS_LIMIT in ScoreCards.vue. */ |
| 57 | private const MAX_THANKS = 1000; |
| 58 | |
| 59 | /** How many articles to use for $priorityTitles in getPageViewData(). */ |
| 60 | private const PRIORITY_ARTICLES_LIMIT = 5; |
| 61 | |
| 62 | /** How many days of pageview data to get. PageViewInfo supports up to 60. */ |
| 63 | public const PAGEVIEW_DAYS = 60; |
| 64 | |
| 65 | private ServiceOptions $config; |
| 66 | private IConnectionProvider $connectionProvider; |
| 67 | private NameTableStore $changeTagDefStore; |
| 68 | private UserFactory $userFactory; |
| 69 | private UserEditTracker $userEditTracker; |
| 70 | private TitleFormatter $titleFormatter; |
| 71 | private TitleFactory $titleFactory; |
| 72 | private IBufferingStatsdDataFactory $statsdDataFactory; |
| 73 | private ?LoggerInterface $logger; |
| 74 | private ?PageViewService $pageViewService; |
| 75 | private ?ThanksQueryHelper $thanksQueryHelper; |
| 76 | private TaskTypeHandlerRegistry $taskTypeHandlerRegistry; |
| 77 | private ConfigurationLoader $configurationLoader; |
| 78 | |
| 79 | /** |
| 80 | * @param ServiceOptions $config |
| 81 | * @param IConnectionProvider $connectionProvider |
| 82 | * @param NameTableStore $changeTagDefStore |
| 83 | * @param UserFactory $userFactory |
| 84 | * @param UserEditTracker $userEditTracker |
| 85 | * @param TitleFormatter $titleFormatter |
| 86 | * @param TitleFactory $titleFactory |
| 87 | * @param IBufferingStatsdDataFactory $statsdDataFactory |
| 88 | * @param TaskTypeHandlerRegistry $taskTypeHandlerRegistry |
| 89 | * @param ConfigurationLoader $configurationLoader |
| 90 | * @param LoggerInterface|null $loggerFactory |
| 91 | * @param PageViewService|null $pageViewService |
| 92 | * @param ThanksQueryHelper|null $thanksQueryHelper |
| 93 | */ |
| 94 | public function __construct( |
| 95 | ServiceOptions $config, |
| 96 | IConnectionProvider $connectionProvider, |
| 97 | NameTableStore $changeTagDefStore, |
| 98 | UserFactory $userFactory, |
| 99 | UserEditTracker $userEditTracker, |
| 100 | TitleFormatter $titleFormatter, |
| 101 | TitleFactory $titleFactory, |
| 102 | IBufferingStatsdDataFactory $statsdDataFactory, |
| 103 | TaskTypeHandlerRegistry $taskTypeHandlerRegistry, |
| 104 | ConfigurationLoader $configurationLoader, |
| 105 | ?LoggerInterface $loggerFactory, |
| 106 | ?PageViewService $pageViewService, |
| 107 | ?ThanksQueryHelper $thanksQueryHelper |
| 108 | ) { |
| 109 | $this->config = $config; |
| 110 | $this->connectionProvider = $connectionProvider; |
| 111 | $this->changeTagDefStore = $changeTagDefStore; |
| 112 | $this->userFactory = $userFactory; |
| 113 | $this->userEditTracker = $userEditTracker; |
| 114 | $this->titleFormatter = $titleFormatter; |
| 115 | $this->titleFactory = $titleFactory; |
| 116 | $this->statsdDataFactory = $statsdDataFactory; |
| 117 | $this->logger = $loggerFactory ?? new NullLogger(); |
| 118 | $this->pageViewService = $pageViewService; |
| 119 | $this->thanksQueryHelper = $thanksQueryHelper; |
| 120 | $this->taskTypeHandlerRegistry = $taskTypeHandlerRegistry; |
| 121 | $this->configurationLoader = $configurationLoader; |
| 122 | } |
| 123 | |
| 124 | /** @inheritDoc */ |
| 125 | public function getUserImpact( UserIdentity $user, int $flags = IDBAccessObject::READ_NORMAL ): ?UserImpact { |
| 126 | $user = $this->userFactory->newFromUserIdentity( $user ); |
| 127 | if ( !$user->isNamed() || $user->isHidden() ) { |
| 128 | return null; |
| 129 | } |
| 130 | |
| 131 | $editData = $this->getEditData( $user, $flags ); |
| 132 | $thanksCount = $this->getThanksCount( $user, $flags ); |
| 133 | |
| 134 | return new UserImpact( |
| 135 | $user, |
| 136 | $thanksCount, |
| 137 | $editData->getEditCountByNamespace(), |
| 138 | $editData->getEditCountByDay(), |
| 139 | $editData->getEditCountByTaskType(), |
| 140 | $editData->getRevertedEditCount(), |
| 141 | $editData->getNewcomerTaskEditCount(), |
| 142 | wfTimestampOrNull( TS_UNIX, $editData->getLastEditTimestamp() ), |
| 143 | ComputeEditingStreaks::getLongestEditingStreak( $editData->getEditCountByDay() ), |
| 144 | $this->userEditTracker->getUserEditCount( $user ) |
| 145 | ); |
| 146 | } |
| 147 | |
| 148 | /** @inheritDoc */ |
| 149 | public function getExpensiveUserImpact( |
| 150 | UserIdentity $user, |
| 151 | int $flags = IDBAccessObject::READ_NORMAL, |
| 152 | array $priorityArticles = [] |
| 153 | ): ?ExpensiveUserImpact { |
| 154 | $start = microtime( true ); |
| 155 | if ( !$this->pageViewService ) { |
| 156 | return null; |
| 157 | } |
| 158 | $user = $this->userFactory->newFromUserIdentity( $user ); |
| 159 | if ( !$user->isNamed() || $user->isHidden() ) { |
| 160 | return null; |
| 161 | } |
| 162 | |
| 163 | $editData = $this->getEditData( $user, $flags ); |
| 164 | $thanksCount = $this->getThanksCount( $user, $flags ); |
| 165 | // Use priority articles if known, otherwise make use of the last edited articles |
| 166 | // as "top articles" . |
| 167 | // This won't exclude retrieving data for other articles, but ensures that we fetch page |
| 168 | // view data for priority (as defined by the caller) articles first. |
| 169 | if ( $priorityArticles ) { |
| 170 | $priorityArticles = array_intersect_key( $editData->getEditedArticles(), $priorityArticles ); |
| 171 | } else { |
| 172 | $priorityArticles = $editData->getEditedArticles(); |
| 173 | } |
| 174 | $pageViewData = $this->getPageViewData( |
| 175 | $user, |
| 176 | $editData->getEditedArticles(), |
| 177 | array_slice( $priorityArticles, 0, self::PRIORITY_ARTICLES_LIMIT, true ), |
| 178 | self::PAGEVIEW_DAYS |
| 179 | ); |
| 180 | if ( $pageViewData === null ) { |
| 181 | return null; |
| 182 | } |
| 183 | |
| 184 | $expensiveUserImpact = new ExpensiveUserImpact( |
| 185 | $user, |
| 186 | $thanksCount, |
| 187 | $editData->getEditCountByNamespace(), |
| 188 | $editData->getEditCountByDay(), |
| 189 | $editData->getEditCountByTaskType(), |
| 190 | $editData->getRevertedEditCount(), |
| 191 | $editData->getNewcomerTaskEditCount(), |
| 192 | wfTimestampOrNull( TS_UNIX, $editData->getLastEditTimestamp() ), |
| 193 | $pageViewData['dailyTotalViews'], |
| 194 | $pageViewData['dailyArticleViews'], |
| 195 | ComputeEditingStreaks::getLongestEditingStreak( $editData->getEditCountByDay() ), |
| 196 | $this->userEditTracker->getUserEditCount( $user ) |
| 197 | ); |
| 198 | $this->statsdDataFactory->timing( |
| 199 | 'timing.growthExperiments.ComputedUserImpactLookup.getExpensiveUserImpact', microtime( true ) - $start |
| 200 | ); |
| 201 | return $expensiveUserImpact; |
| 202 | } |
| 203 | |
| 204 | /** |
| 205 | * Run a SQL query to fetch edit data for the user. |
| 206 | * |
| 207 | * @param User $user |
| 208 | * @param int $flags |
| 209 | * @return EditData |
| 210 | * @throws \Exception |
| 211 | */ |
| 212 | private function getEditData( User $user, int $flags ): EditData { |
| 213 | $db = DBAccessObjectUtils::getDBFromRecency( $this->connectionProvider, $flags ); |
| 214 | |
| 215 | $queryBuilder = $db->newSelectQueryBuilder() |
| 216 | ->table( 'revision' ) |
| 217 | ->join( 'page', null, 'rev_page = page_id' ); |
| 218 | |
| 219 | $taskChangeTagNames = $this->taskTypeHandlerRegistry->getUniqueChangeTags(); |
| 220 | $additionalChangeTagNames = [ |
| 221 | ChangeTags::TAG_REVERTED |
| 222 | ]; |
| 223 | |
| 224 | $changeTagNames = array_merge( $taskChangeTagNames, $additionalChangeTagNames ); |
| 225 | $changeTagIds = []; |
| 226 | $changeTagIdToName = []; |
| 227 | foreach ( $changeTagNames as $changeTagName ) { |
| 228 | try { |
| 229 | // Presume the tag is not related to a task; set $taskTypeId to the task type ID |
| 230 | // if it is. |
| 231 | $taskTypeId = null; |
| 232 | if ( in_array( $changeTagName, $taskChangeTagNames ) ) { |
| 233 | $taskTypeHandlerId = $this->taskTypeHandlerRegistry->getTaskTypeHandlerIdByChangeTagName( |
| 234 | $changeTagName |
| 235 | ); |
| 236 | if ( !$taskTypeHandlerId ) { |
| 237 | // In theory shouldn't be possible, given that the change tag names originate from the |
| 238 | // task type handler registry. Adding this to make phan happy. |
| 239 | throw new LogicException( |
| 240 | "Unable to find task type handler ID for change tag \"$changeTagName\"" |
| 241 | ); |
| 242 | } |
| 243 | $taskTypeHandler = $this->taskTypeHandlerRegistry->get( $taskTypeHandlerId ); |
| 244 | $taskTypeId = $taskTypeHandler->getTaskTypeIdByChangeTagName( $changeTagName ); |
| 245 | } |
| 246 | |
| 247 | $tagId = $this->changeTagDefStore->getId( $changeTagName ); |
| 248 | $changeTagIds[$tagId] = $taskTypeId; |
| 249 | $changeTagIdToName[$tagId] = $changeTagName; |
| 250 | } catch ( NameTableAccessException $nameTableAccessException ) { |
| 251 | // Some tags won't exist in test scenarios, and possibly in some small wikis where |
| 252 | // no suggested edits have been done yet. We can safely ignore the exception, |
| 253 | // it will mean that 'newcomerTaskEditCount' is 0 in the result. |
| 254 | } |
| 255 | } |
| 256 | |
| 257 | if ( $changeTagIds ) { |
| 258 | $queryBuilder->leftJoin( 'change_tag', null, [ |
| 259 | 'rev_id = ct_rev_id', |
| 260 | 'ct_tag_id' => array_keys( $changeTagIds ), |
| 261 | ] ); |
| 262 | $queryBuilder->field( 'ct_tag_id' ); |
| 263 | } |
| 264 | |
| 265 | $queryBuilder->fields( [ 'page_namespace', 'page_title', 'rev_timestamp' ] ); |
| 266 | $queryBuilder->where( [ 'rev_actor' => $user->getActorId() ] ); |
| 267 | $queryBuilder->where( $db->bitAnd( 'rev_deleted', RevisionRecord::DELETED_USER ) . ' = 0' ); |
| 268 | // hopefully able to use the rev_actor_timestamp index for an efficient query |
| 269 | $queryBuilder->orderBy( 'rev_timestamp', 'DESC' ); |
| 270 | $queryBuilder->limit( self::MAX_EDITS ); |
| 271 | $queryBuilder->recency( $flags ); |
| 272 | $queryBuilder->caller( __METHOD__ ); |
| 273 | // T331264 |
| 274 | $queryBuilder->straightJoinOption(); |
| 275 | |
| 276 | $editCountByNamespace = []; |
| 277 | $editCountByDay = []; |
| 278 | $revertedEditCount = 0; |
| 279 | $editCountByTaskType = array_fill_keys( array_keys( $this->configurationLoader->getTaskTypes() ), 0 ); |
| 280 | $newcomerTaskEditCount = 0; |
| 281 | $lastEditTimestamp = null; |
| 282 | $editedArticles = []; |
| 283 | |
| 284 | foreach ( $queryBuilder->fetchResultSet() as $row ) { |
| 285 | $linkTarget = new TitleValue( (int)$row->page_namespace, $row->page_title ); |
| 286 | $titleDbKey = $this->titleFormatter->getPrefixedDBkey( $linkTarget ); |
| 287 | $editTime = new MWTimestamp( $row->rev_timestamp ); |
| 288 | // Avoid using registered user timezone preference which can be used to de-anonymize users. |
| 289 | // Use anonymous UserIdentity instead which will fall back to use the wiki's default |
| 290 | // timezone and local tz offset. |
| 291 | $editTime->offsetForUser( $this->userFactory->newAnonymous() ); |
| 292 | $day = $editTime->format( 'Ymd' ); |
| 293 | |
| 294 | $editCountByNamespace[$row->page_namespace] |
| 295 | = ( $editCountByNamespace[$row->page_namespace] ?? 0 ) + 1; |
| 296 | $editCountByDay[$day] = ( $editCountByDay[$day] ?? 0 ) + 1; |
| 297 | if ( $row->ct_tag_id ?? null ) { |
| 298 | $taskTypeId = $changeTagIds[$row->ct_tag_id]; |
| 299 | if ( $taskTypeId ) { |
| 300 | $newcomerTaskEditCount++; |
| 301 | if ( !isset( $editCountByTaskType[$taskTypeId] ) ) { |
| 302 | $editCountByTaskType[$taskTypeId] = 0; |
| 303 | } |
| 304 | $editCountByTaskType[$taskTypeId]++; |
| 305 | } |
| 306 | |
| 307 | $changeTagName = $changeTagIdToName[$row->ct_tag_id]; |
| 308 | if ( $changeTagName === ChangeTags::TAG_REVERTED ) { |
| 309 | $revertedEditCount++; |
| 310 | } |
| 311 | } |
| 312 | $lastEditTimestamp ??= $row->rev_timestamp; |
| 313 | // Computed values $editCountByNamespace, $editCountByDay, $newcomerTaskEditCount and $lastEditTimestamp |
| 314 | // use data from all namespaces. Filter out non-article pages from the collection of returned articles |
| 315 | // ($editedArticles) since they are not relevant for the user article list of recent edits. |
| 316 | if ( (int)$row->page_namespace !== NS_MAIN ) { |
| 317 | continue; |
| 318 | } |
| 319 | // We're iterating over the result set, newest edits to oldest edits in descending order. The same |
| 320 | // article can have been edited multiple times. We'll stash the revision timestamp of the oldest |
| 321 | // edit made by the user to the article; we will use that later to calculate the "start date" |
| 322 | // for the impact of the user for a particular article, e.g. when making a pageviews tool URL |
| 323 | // or choosing the date range for page view data to display for an article. |
| 324 | $editedArticles[$titleDbKey]['oldestEdit'] = $row->rev_timestamp; |
| 325 | $editedArticles[$titleDbKey]['newestEdit'] ??= $row->rev_timestamp; |
| 326 | } |
| 327 | |
| 328 | return new EditData( |
| 329 | $editCountByNamespace, |
| 330 | array_reverse( $this->updateToIso8601DateKeys( $editCountByDay ) ), |
| 331 | $editCountByTaskType, |
| 332 | $revertedEditCount, |
| 333 | $newcomerTaskEditCount, |
| 334 | $lastEditTimestamp, |
| 335 | $editedArticles |
| 336 | ); |
| 337 | } |
| 338 | |
| 339 | /** |
| 340 | * @param User $user |
| 341 | * @param int $flags |
| 342 | * @return int Number of thanks received for the user ID |
| 343 | */ |
| 344 | private function getThanksCount( User $user, int $flags ): int { |
| 345 | return $this->thanksQueryHelper |
| 346 | ? $this->thanksQueryHelper->getThanksReceivedCount( $user, self::MAX_THANKS, $flags ) |
| 347 | : 0; |
| 348 | } |
| 349 | |
| 350 | /** |
| 351 | * Returns page views and other data, or null on error during data fetching. |
| 352 | * Must not be called when $this->pageViewService is null. |
| 353 | * @param User $user |
| 354 | * @param array[] $titles Data about edited articles. See {@see EditData::getEditedArticles()} |
| 355 | * for format. |
| 356 | * @param array[] $priorityTitles A subset of $titles that should get priority treatment |
| 357 | * (in case not all the pageview data can be retrieved due to resource limits). |
| 358 | * @param int $days How many days to query. No more than 60. |
| 359 | * @return array|null |
| 360 | * - dailyTotalViews: (array<string, int>) daily number of total views of articles in $titles, |
| 361 | * keyed by ISO 8601 date. |
| 362 | * - dailyArticleViews: (array[]) Daily article views and other data. Keyed by |
| 363 | * prefixed DBkey; values are arrays with the following fields: |
| 364 | * - views: (int[]) daily article views, keyed by ISO 8601 date. Might be 0 for the last day |
| 365 | * if it's still being processed. |
| 366 | * - firstEditDate: (string) ISO 8601 date of the user's first edit to the article. |
| 367 | * If the user made a very high number of total edits, it might just be some edit the |
| 368 | * user made to the article, not necessarily the first. |
| 369 | * - newestEdit: (string) MW_TS timestamp of the user's most recent edit. |
| 370 | * - imageUrl: (string|null) URL of a thumbnail of the article's main image. |
| 371 | * @phan-return array{dailyTotalViews:array<string,int>,dailyArticleViews:array<string,array{views:array<string,int>,firstEditDate:string,newestEdit:string,imageUrl:?string}>}|null |
| 372 | * @throws MalformedTitleException |
| 373 | */ |
| 374 | private function getPageViewData( User $user, array $titles, array $priorityTitles, int $days ): ?array { |
| 375 | // Short-circuit if the user has no edits. |
| 376 | if ( !$titles ) { |
| 377 | return [ |
| 378 | 'dailyTotalViews' => [], |
| 379 | 'dailyArticleViews' => [], |
| 380 | ]; |
| 381 | } |
| 382 | |
| 383 | // $priorityTitles is a subset of $titles but putting it to the front makes sure the data |
| 384 | // for those titles is fetched even if PageViewInfo cuts off the list of titles at some |
| 385 | // point, which it is allowed to do. |
| 386 | $allTitles = $priorityTitles + $titles; |
| 387 | $allTitleObjects = []; |
| 388 | |
| 389 | foreach ( $allTitles as $title => $data ) { |
| 390 | $allTitleObjects[$title] = [ |
| 391 | 'title' => $this->titleFactory->newFromTextThrow( $title ), |
| 392 | // rev_timestamp is in TS_MW format (e.g. 20210406200220), we only want |
| 393 | // the first 8 characters for comparison with Ymd format date strings. |
| 394 | 'rev_timestamp' => substr( $data['oldestEdit'], 0, 8 ), |
| 395 | 'newestEdit' => $data['newestEdit'], |
| 396 | 'oldestEdit' => $data['oldestEdit'] |
| 397 | ]; |
| 398 | } |
| 399 | if ( defined( 'MEDIAWIKI_JOB_RUNNER' ) || MW_ENTRY_POINT === 'cli' ) { |
| 400 | $pageViewData = $this->getPageViewDataInJobContext( $allTitleObjects, $user, $days ); |
| 401 | } else { |
| 402 | $pageViewData = $this->getPageViewDataInWebRequestContext( $allTitleObjects, $user, $days ); |
| 403 | } |
| 404 | |
| 405 | $dailyTotalViews = []; |
| 406 | $dailyArticleViews = []; |
| 407 | foreach ( $pageViewData as $title => $days ) { |
| 408 | // Normalize titles as PageViewInfo does not define which title format it uses :( |
| 409 | $title = str_replace( ' ', '_', $title ); |
| 410 | $mwTitle = $this->titleFactory->newFromTextThrow( $title ); |
| 411 | $imageUrl = $this->getImage( $mwTitle ); |
| 412 | if ( $imageUrl ) { |
| 413 | $dailyArticleViews[$title]['imageUrl'] = $imageUrl; |
| 414 | } |
| 415 | $firstEditDate = new DateTime( $allTitleObjects[$title]['rev_timestamp'] ); |
| 416 | $dailyArticleViews[$title]['firstEditDate'] = $firstEditDate->format( 'Y-m-d' ); |
| 417 | $dailyArticleViews[$title]['newestEdit'] = $allTitleObjects[$title]['newestEdit']; |
| 418 | |
| 419 | foreach ( $days as $day => $views ) { |
| 420 | // NOTE: Do not insert the data if it is a zero due to JSON blob size issues (T351898) |
| 421 | |
| 422 | $todayTotalViews = ( ( $dailyTotalViews[$day] ?? 0 ) + $views ); |
| 423 | if ( $todayTotalViews > 0 ) { |
| 424 | $dailyTotalViews[$day] = $todayTotalViews; |
| 425 | } |
| 426 | |
| 427 | $todayArticleViews = ( $views ?? 0 ); |
| 428 | if ( $todayArticleViews > 0 ) { |
| 429 | $dailyArticleViews[$title]['views'][$day] = $todayArticleViews; |
| 430 | } |
| 431 | } |
| 432 | } |
| 433 | |
| 434 | return [ |
| 435 | 'dailyTotalViews' => $dailyTotalViews, |
| 436 | 'dailyArticleViews' => $dailyArticleViews, |
| 437 | ]; |
| 438 | } |
| 439 | |
| 440 | private function getPageViewDataInJobContext( array $allTitleObjects, UserIdentity $user, int $days ): array { |
| 441 | $pageViewData = []; |
| 442 | $titleObjects = $allTitleObjects; |
| 443 | $loopStartTime = microtime( true ); |
| 444 | while ( count( $titleObjects ) ) { |
| 445 | $titleObjectsCount = count( $titleObjects ); |
| 446 | if ( count( $pageViewData ) > $this->config->get( 'GEUserImpactMaxArticlesToProcessForPageviews' ) ) { |
| 447 | $this->logger->info( |
| 448 | 'Reached article count limit while fetching page view data for {count} titles for user {user}.', |
| 449 | [ 'user' => $user->getName(), 'count' => count( $allTitleObjects ) ] |
| 450 | ); |
| 451 | break; |
| 452 | } |
| 453 | if ( microtime( true ) - $loopStartTime > $this->config->get( 'GEUserImpactMaximumProcessTimeSeconds' ) ) { |
| 454 | $this->logger->info( |
| 455 | "Reached maximum process time while fetching page view data for {count} titles for user {user}", |
| 456 | [ 'user' => $user->getName(), 'count' => count( $allTitleObjects ) ] |
| 457 | ); |
| 458 | break; |
| 459 | } |
| 460 | $pageDataStatus = $this->pageViewService->getPageData( |
| 461 | array_column( $titleObjects, 'title' ), $days |
| 462 | ); |
| 463 | if ( !$pageDataStatus->isGood() ) { |
| 464 | $this->logPageDataBadStatus( $pageDataStatus ); |
| 465 | } |
| 466 | if ( $pageDataStatus->isOK() ) { |
| 467 | $successful = array_filter( $pageDataStatus->success ); |
| 468 | $pageViewData += array_intersect_key( $pageDataStatus->getValue(), $successful ); |
| 469 | } |
| 470 | $titleObjects = array_diff_key( $titleObjects, $pageViewData ); |
| 471 | if ( count( $titleObjects ) === $titleObjectsCount ) { |
| 472 | // Received no new data. Abort to avoid a loop - errors are cached for a short time |
| 473 | // so re-requesting them wouldn't help. |
| 474 | return $pageViewData; |
| 475 | } |
| 476 | } |
| 477 | return $pageViewData; |
| 478 | } |
| 479 | |
| 480 | private function getPageViewDataInWebRequestContext( |
| 481 | array $allTitleObjects, UserIdentity $user, int $days |
| 482 | ): array { |
| 483 | $status = $this->pageViewService->getPageData( array_column( $allTitleObjects, 'title' ), $days ); |
| 484 | if ( !$status->isGood() ) { |
| 485 | $this->logPageDataBadStatus( $status ); |
| 486 | if ( !$status->isOK() ) { |
| 487 | return []; |
| 488 | } |
| 489 | } elseif ( $status->successCount < count( $allTitleObjects ) ) { |
| 490 | $failedTitles = array_keys( array_diff_key( $allTitleObjects, $status->success ) ); |
| 491 | $this->logger->info( "Failed to get page view data for {count} titles for user {user}", |
| 492 | [ |
| 493 | 'user' => $user->getName(), |
| 494 | 'count' => count( $failedTitles ), |
| 495 | 'failedTitles' => substr( implode( ',', $failedTitles ), 0, 250 ), |
| 496 | ] |
| 497 | ); |
| 498 | } |
| 499 | return $status->getValue(); |
| 500 | } |
| 501 | |
| 502 | /** |
| 503 | * Don't log pvi-cached-error-title messages (T328945) but track it in statsd, |
| 504 | * and log any other message that occurs. |
| 505 | * |
| 506 | * @param StatusValue $status |
| 507 | * @return void |
| 508 | */ |
| 509 | private function logPageDataBadStatus( StatusValue $status ) { |
| 510 | if ( $status->hasMessagesExcept( 'pvi-cached-error-title' ) ) { |
| 511 | $this->logger->error( |
| 512 | Status::wrap( $status )->getWikiText( false, false, 'en' ) |
| 513 | ); |
| 514 | } else { |
| 515 | $this->statsdDataFactory->updateCount( |
| 516 | 'GrowthExperiments.ComputedUserImpactLookup.PviCachedErrorTitle', |
| 517 | $status->failCount |
| 518 | ); |
| 519 | } |
| 520 | } |
| 521 | |
| 522 | /** |
| 523 | * Change array keys from MW_TS date prefixes to ISO 8601 dates. |
| 524 | * @param array $mwTsArray |
| 525 | * @return array |
| 526 | */ |
| 527 | private function updateToIso8601DateKeys( array $mwTsArray ): array { |
| 528 | $iso8601Array = []; |
| 529 | foreach ( $mwTsArray as $mwTsKey => $value ) { |
| 530 | $iso8601Key = substr( $mwTsKey, 0, 4 ) . '-' . substr( $mwTsKey, 4, 2 ) |
| 531 | . '-' . substr( $mwTsKey, 6, 2 ); |
| 532 | $iso8601Array[$iso8601Key] = $value; |
| 533 | } |
| 534 | return $iso8601Array; |
| 535 | } |
| 536 | |
| 537 | /** |
| 538 | * Get image URL for a page |
| 539 | * Depends on the PageImages extension. |
| 540 | * |
| 541 | * @param Title $title |
| 542 | * @return ?string |
| 543 | */ |
| 544 | private function getImage( Title $title ): ?string { |
| 545 | if ( !ExtensionRegistry::getInstance()->isLoaded( 'PageImages' ) ) { |
| 546 | return null; |
| 547 | } |
| 548 | |
| 549 | $imageFile = PageImages::getPageImage( $title ); |
| 550 | if ( $imageFile ) { |
| 551 | $ratio = $imageFile->getWidth() / $imageFile->getHeight(); |
| 552 | $options = [ |
| 553 | 'width' => $ratio > 1 ? |
| 554 | // Avoid decimals in the width because it makes the thumb url construction fail |
| 555 | floor( self::THUMBNAIL_SIZE / $imageFile->getHeight() * $imageFile->getWidth() ) : |
| 556 | self::THUMBNAIL_SIZE |
| 557 | ]; |
| 558 | |
| 559 | $thumb = $imageFile->transform( $options ); |
| 560 | if ( $thumb ) { |
| 561 | return $thumb->getUrl() ?: null; |
| 562 | } |
| 563 | } |
| 564 | |
| 565 | return null; |
| 566 | } |
| 567 | |
| 568 | } |