Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 147 |
|
0.00% |
0 / 10 |
CRAP | |
0.00% |
0 / 1 |
| RefreshLinkRecommendations | |
0.00% |
0 / 141 |
|
0.00% |
0 / 10 |
1332 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
2 | |||
| checkRequiredExtensions | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| execute | |
0.00% |
0 / 58 |
|
0.00% |
0 / 1 |
240 | |||
| initGrowthConfig | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| initServices | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
2 | |||
| initConfig | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| getOresTopics | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| findArticlesInTopic | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
20 | |||
| processCandidate | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
20 | |||
| verboseLog | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\Maintenance; |
| 4 | |
| 5 | use CirrusSearch\Query\ArticleTopicFeature; |
| 6 | use Generator; |
| 7 | use GrowthExperiments\GrowthExperimentsServices; |
| 8 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationStore; |
| 9 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationUpdater; |
| 10 | use GrowthExperiments\NewcomerTasks\ConfigurationLoader\ConfigurationLoader; |
| 11 | use GrowthExperiments\NewcomerTasks\ConfigurationLoader\TopicDecorator; |
| 12 | use GrowthExperiments\NewcomerTasks\Task\TaskSetFilters; |
| 13 | use GrowthExperiments\NewcomerTasks\TaskSuggester\TaskSuggester; |
| 14 | use GrowthExperiments\NewcomerTasks\TaskType\LinkRecommendationTaskType; |
| 15 | use GrowthExperiments\NewcomerTasks\TaskType\LinkRecommendationTaskTypeHandler; |
| 16 | use GrowthExperiments\NewcomerTasks\TaskType\NullTaskTypeHandler; |
| 17 | use GrowthExperiments\WikiConfigException; |
| 18 | use Maintenance; |
| 19 | use MediaWiki\Cache\LinkBatchFactory; |
| 20 | use MediaWiki\Config\Config; |
| 21 | use MediaWiki\MediaWikiServices; |
| 22 | use MediaWiki\Status\Status; |
| 23 | use MediaWiki\Title\Title; |
| 24 | use MediaWiki\Title\TitleFactory; |
| 25 | use MediaWiki\User\User; |
| 26 | use MediaWiki\WikiMap\WikiMap; |
| 27 | use RuntimeException; |
| 28 | use StatusValue; |
| 29 | use Wikimedia\Rdbms\DBReadOnlyError; |
| 30 | |
| 31 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 32 | if ( $IP === false ) { |
| 33 | $IP = __DIR__ . '/../../..'; |
| 34 | } |
| 35 | require_once "$IP/maintenance/Maintenance.php"; |
| 36 | |
| 37 | /** |
| 38 | * Update the growthexperiments_link_recommendations table to ensure there are enough |
| 39 | * recommendations for all topics |
| 40 | */ |
| 41 | class RefreshLinkRecommendations extends Maintenance { |
| 42 | |
| 43 | /** @var Config */ |
| 44 | private $growthConfig; |
| 45 | |
| 46 | /** @var TitleFactory */ |
| 47 | private $titleFactory; |
| 48 | |
| 49 | /** @var LinkBatchFactory */ |
| 50 | private $linkBatchFactory; |
| 51 | |
| 52 | /** @var ConfigurationLoader */ |
| 53 | private $configurationLoader; |
| 54 | |
| 55 | /** @var TaskSuggester */ |
| 56 | private $taskSuggester; |
| 57 | |
| 58 | /** @var LinkRecommendationStore */ |
| 59 | private $linkRecommendationStore; |
| 60 | |
| 61 | /** @var LinkRecommendationUpdater */ |
| 62 | private $linkRecommendationUpdater; |
| 63 | |
| 64 | /** @var LinkRecommendationTaskType */ |
| 65 | private $recommendationTaskType; |
| 66 | |
| 67 | /** @var User */ |
| 68 | private $searchUser; |
| 69 | |
| 70 | public function __construct() { |
| 71 | parent::__construct(); |
| 72 | $this->requireExtension( 'GrowthExperiments' ); |
| 73 | $this->requireExtension( 'CirrusSearch' ); |
| 74 | |
| 75 | $this->addDescription( 'Update the growthexperiments_link_recommendations table to ensure ' |
| 76 | . 'there are enough recommendations for all topics.' ); |
| 77 | $this->addOption( 'topic', 'Only update articles in the given ORES topic.', false, true ); |
| 78 | $this->addOption( 'page', 'Only update a specific page.', false, true ); |
| 79 | $this->addOption( 'force', 'Generate recommendations even if they fail quality criteria.' ); |
| 80 | $this->addOption( 'verbose', 'Show debug output.' ); |
| 81 | $this->setBatchSize( 500 ); |
| 82 | } |
| 83 | |
| 84 | public function checkRequiredExtensions() { |
| 85 | // Hack: must be early enough for requireExtension to work but late enough for config |
| 86 | // to be available. |
| 87 | $growthServices = GrowthExperimentsServices::wrap( MediaWikiServices::getInstance() ); |
| 88 | if ( $growthServices->getGrowthConfig()->get( 'GELinkRecommendationsUseEventGate' ) ) { |
| 89 | $this->requireExtension( 'EventBus' ); |
| 90 | } |
| 91 | parent::checkRequiredExtensions(); |
| 92 | } |
| 93 | |
| 94 | public function execute() { |
| 95 | $this->initGrowthConfig(); |
| 96 | if ( !$this->growthConfig->get( 'GENewcomerTasksLinkRecommendationsEnabled' ) ) { |
| 97 | $this->output( "Disabled\n" ); |
| 98 | return; |
| 99 | } elseif ( $this->growthConfig->get( 'GENewcomerTasksRemoteApiUrl' ) ) { |
| 100 | $this->output( "Local tasks disabled\n" ); |
| 101 | return; |
| 102 | } |
| 103 | $this->initServices(); |
| 104 | $this->initConfig(); |
| 105 | $lockName = 'GrowthExperiments-RefreshLinkRecommendations-' . WikiMap::getCurrentWikiId(); |
| 106 | if ( !$this->linkRecommendationStore->getDB( DB_PRIMARY )->lock( $lockName, __METHOD__, 0 ) ) { |
| 107 | $this->output( "Previous invocation of the script is still running\n" ); |
| 108 | return; |
| 109 | } |
| 110 | |
| 111 | $force = $this->hasOption( 'force' ); |
| 112 | $this->output( "Refreshing link recommendations...\n" ); |
| 113 | |
| 114 | $pageName = $this->getOption( 'page' ); |
| 115 | if ( $pageName ) { |
| 116 | $title = $this->titleFactory->newFromText( $pageName ); |
| 117 | if ( $title ) { |
| 118 | $this->processCandidate( $title, $force ); |
| 119 | } else { |
| 120 | $this->fatalError( 'Invalid title: ' . $pageName ); |
| 121 | } |
| 122 | return; |
| 123 | } |
| 124 | |
| 125 | $oresTopics = $this->getOresTopics(); |
| 126 | foreach ( $oresTopics as $oresTopic ) { |
| 127 | $this->output( " processing topic $oresTopic...\n" ); |
| 128 | $suggestions = $this->taskSuggester->suggest( |
| 129 | $this->searchUser, |
| 130 | new TaskSetFilters( |
| 131 | [ LinkRecommendationTaskTypeHandler::TASK_TYPE_ID ], |
| 132 | [ $oresTopic ] |
| 133 | ), |
| 134 | 1, |
| 135 | 0, |
| 136 | // Enabling the debug flag is relatively harmless, and disables all caching, |
| 137 | // which we need here. useCache would prevent reading the cache, but would |
| 138 | // still write it, which would be just a waste of space. |
| 139 | [ 'debug' => true ] |
| 140 | ); |
| 141 | |
| 142 | // TaskSuggester::suggest() only returns StatusValue when there's an error. |
| 143 | if ( $suggestions instanceof StatusValue ) { |
| 144 | $this->error( Status::wrap( $suggestions )->getWikiText( false, false, 'en' ) ); |
| 145 | continue; |
| 146 | } |
| 147 | |
| 148 | $recommendationsNeeded = $this->recommendationTaskType->getMinimumTasksPerTopic() |
| 149 | - $suggestions->getTotalCount(); |
| 150 | |
| 151 | if ( $recommendationsNeeded <= 0 ) { |
| 152 | $this->output( " no new tasks needed\n" ); |
| 153 | continue; |
| 154 | } |
| 155 | $this->output( " $recommendationsNeeded new tasks needed\n" ); |
| 156 | foreach ( $this->findArticlesInTopic( $oresTopic ) as $titleBatch ) { |
| 157 | $recommendationsFound = 0; |
| 158 | foreach ( $titleBatch as $title ) { |
| 159 | // TODO filter out protected pages. Needs to be batched. Or wait for T259346. |
| 160 | $success = $this->processCandidate( $title, $force ); |
| 161 | if ( $success ) { |
| 162 | $recommendationsFound++; |
| 163 | $recommendationsNeeded--; |
| 164 | if ( $recommendationsNeeded <= 0 ) { |
| 165 | break 2; |
| 166 | } |
| 167 | } |
| 168 | } |
| 169 | $this->waitForReplication(); |
| 170 | // findArticlesInTopic() picks articles at random, so we need to abort the loop |
| 171 | // at some point. Do it when no new tasks were generated from the current batch. |
| 172 | if ( $recommendationsFound === 0 ) { |
| 173 | break; |
| 174 | } |
| 175 | } |
| 176 | $this->output( ( $recommendationsNeeded === 0 ) ? " task pool filled\n" |
| 177 | : " topic exhausted, $recommendationsNeeded tasks still needed\n" ); |
| 178 | } |
| 179 | } |
| 180 | |
| 181 | protected function initGrowthConfig(): void { |
| 182 | // Needs to be separate from initServices/initConfig as checking whether the script |
| 183 | // should run on a given wiki relies on this, but initServices/initConfig will break |
| 184 | // on some wikis where the script is not supposed to run and the task configuration |
| 185 | // is missing. |
| 186 | $services = MediaWikiServices::getInstance(); |
| 187 | $growthServices = GrowthExperimentsServices::wrap( $services ); |
| 188 | $this->growthConfig = $growthServices->getGrowthConfig(); |
| 189 | } |
| 190 | |
| 191 | protected function initServices(): void { |
| 192 | // Extend the task type configuration with a custom "candidate" task type, which |
| 193 | // finds articles which do not have link recommendations. |
| 194 | $linkRecommendationCandidateTaskType = NullTaskTypeHandler::getNullTaskType( |
| 195 | '_nolinkrecommendations', '-hasrecommendation:link' ); |
| 196 | |
| 197 | $services = MediaWikiServices::getInstance(); |
| 198 | $growthServices = GrowthExperimentsServices::wrap( MediaWikiServices::getInstance() ); |
| 199 | $newcomerTaskConfigurationLoader = $growthServices->getNewcomerTasksConfigurationLoader(); |
| 200 | $this->configurationLoader = new TopicDecorator( |
| 201 | $newcomerTaskConfigurationLoader, |
| 202 | true, |
| 203 | [ $linkRecommendationCandidateTaskType ] |
| 204 | ); |
| 205 | $this->titleFactory = $services->getTitleFactory(); |
| 206 | $this->linkBatchFactory = $services->getLinkBatchFactory(); |
| 207 | $this->taskSuggester = $growthServices->getTaskSuggesterFactory()->create( $this->configurationLoader ); |
| 208 | $this->linkRecommendationStore = $growthServices->getLinkRecommendationStore(); |
| 209 | $this->linkRecommendationUpdater = $growthServices->getLinkRecommendationUpdater(); |
| 210 | } |
| 211 | |
| 212 | protected function initConfig(): void { |
| 213 | $taskTypes = $this->configurationLoader->getTaskTypes(); |
| 214 | $taskType = $taskTypes[LinkRecommendationTaskTypeHandler::TASK_TYPE_ID] ?? null; |
| 215 | if ( !$taskType || !$taskType instanceof LinkRecommendationTaskType ) { |
| 216 | $this->fatalError( sprintf( "'%s' is not a link recommendation task type", |
| 217 | LinkRecommendationTaskTypeHandler::TASK_TYPE_ID ) ); |
| 218 | } else { |
| 219 | $this->recommendationTaskType = $taskType; |
| 220 | } |
| 221 | $this->searchUser = User::newSystemUser( 'Maintenance script', [ 'steal' => true ] ); |
| 222 | } |
| 223 | |
| 224 | /** |
| 225 | * @return string[] |
| 226 | */ |
| 227 | private function getOresTopics(): array { |
| 228 | $topic = $this->getOption( 'topic' ); |
| 229 | $oresTopics = array_keys( ArticleTopicFeature::TERMS_TO_LABELS ); |
| 230 | if ( $topic ) { |
| 231 | $oresTopics = array_intersect( $oresTopics, [ $topic ] ); |
| 232 | if ( !$oresTopics ) { |
| 233 | $this->fatalError( "invalid topic $topic" ); |
| 234 | } |
| 235 | } |
| 236 | return $oresTopics; |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * @param string $oresTopic |
| 241 | * @return Generator<Title[]> |
| 242 | */ |
| 243 | private function findArticlesInTopic( $oresTopic ) { |
| 244 | $batchSize = $this->getBatchSize(); |
| 245 | do { |
| 246 | $this->output( " fetching $batchSize tasks...\n" ); |
| 247 | $candidates = $this->taskSuggester->suggest( |
| 248 | $this->searchUser, |
| 249 | new TaskSetFilters( |
| 250 | [ '_nolinkrecommendations' ], |
| 251 | [ $oresTopic ] |
| 252 | ), |
| 253 | $batchSize, |
| 254 | null, |
| 255 | [ 'debug' => true ] |
| 256 | ); |
| 257 | if ( $candidates instanceof StatusValue ) { |
| 258 | // FIXME exiting will make the cronjob unreliable. Not exiting might result |
| 259 | // in an infinite error loop. Neither looks like a great option. |
| 260 | throw new RuntimeException( 'Search error: ' |
| 261 | . Status::wrap( $candidates )->getWikiText( false, false, 'en' ) ); |
| 262 | } |
| 263 | |
| 264 | $linkTargets = $titles = []; |
| 265 | foreach ( $candidates as $candidate ) { |
| 266 | $linkTargets[] = $candidate->getTitle(); |
| 267 | } |
| 268 | $this->linkBatchFactory->newLinkBatch( $linkTargets )->execute(); |
| 269 | foreach ( $linkTargets as $linkTarget ) { |
| 270 | $titles[] = $this->titleFactory->newFromLinkTarget( $linkTarget ); |
| 271 | } |
| 272 | yield $titles; |
| 273 | } while ( $candidates->count() ); |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Evaluate a task candidate and potentially generate the task. |
| 278 | * @param Title $title |
| 279 | * @param bool $force Ignore all failed conditions that can be safely ignored. |
| 280 | * @return bool Whether a new task was generated. |
| 281 | */ |
| 282 | private function processCandidate( Title $title, bool $force = false ): bool { |
| 283 | $this->verboseLog( " checking candidate " . $title->getPrefixedDBkey() . "... " ); |
| 284 | try { |
| 285 | $status = $this->linkRecommendationUpdater->processCandidate( $title, $force ); |
| 286 | if ( $status->isOK() ) { |
| 287 | $this->verboseLog( "success, updating index\n" ); |
| 288 | return true; |
| 289 | } else { |
| 290 | $error = Status::wrap( $status )->getWikiText( false, false, 'en' ); |
| 291 | $this->verboseLog( "$error\n" ); |
| 292 | } |
| 293 | } catch ( DBReadOnlyError $e ) { |
| 294 | // This is a long-running script, read-only state can change in the middle. |
| 295 | // It's run frequently so just do the easy thing and abort. |
| 296 | $this->fatalError( 'DB is readonly, aborting' ); |
| 297 | } catch ( WikiConfigException $e ) { |
| 298 | // Link recommendations are not configured correctly. |
| 299 | $this->fatalError( $e->getMessage() ); |
| 300 | } |
| 301 | return false; |
| 302 | } |
| 303 | |
| 304 | private function verboseLog( string $message ): void { |
| 305 | if ( $this->hasOption( 'verbose' ) ) { |
| 306 | $this->output( $message ); |
| 307 | } |
| 308 | } |
| 309 | |
| 310 | } |
| 311 | |
| 312 | $maintClass = RefreshLinkRecommendations::class; |
| 313 | require_once RUN_MAINTENANCE_IF_MAIN; |