Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 122 |
|
0.00% |
0 / 11 |
CRAP | |
0.00% |
0 / 1 |
| RevalidateLinkRecommendations | |
0.00% |
0 / 116 |
|
0.00% |
0 / 11 |
1406 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
2 | |||
| checkRequiredExtensions | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| execute | |
0.00% |
0 / 31 |
|
0.00% |
0 / 1 |
132 | |||
| initGrowthConfig | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| initServices | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| validateRecommendation | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
72 | |||
| regenerateRecommendation | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| getTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| verboseLog | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getAllowedChecksums | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
20 | |||
| getOlderThanTimestamp | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\Maintenance; |
| 4 | |
| 5 | use GrowthExperiments\GrowthExperimentsServices; |
| 6 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendation; |
| 7 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationHelper; |
| 8 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationLink; |
| 9 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationStore; |
| 10 | use GrowthExperiments\NewcomerTasks\AddLink\LinkRecommendationUpdater; |
| 11 | use GrowthExperiments\WikiConfigException; |
| 12 | use LogicException; |
| 13 | use Maintenance; |
| 14 | use MediaWiki\Config\Config; |
| 15 | use MediaWiki\MediaWikiServices; |
| 16 | use MediaWiki\Status\Status; |
| 17 | use MediaWiki\Title\Title; |
| 18 | use MediaWiki\Title\TitleFactory; |
| 19 | use UnexpectedValueException; |
| 20 | use Wikimedia\Rdbms\DBReadOnlyError; |
| 21 | |
| 22 | $path = dirname( dirname( dirname( __DIR__ ) ) ); |
| 23 | |
| 24 | if ( getenv( 'MW_INSTALL_PATH' ) !== false ) { |
| 25 | $path = getenv( 'MW_INSTALL_PATH' ); |
| 26 | } |
| 27 | |
| 28 | require_once $path . '/maintenance/Maintenance.php'; |
| 29 | |
| 30 | /** |
| 31 | * Iterate through the growthexperiments_link_recommendations table and regenerate the ones which |
| 32 | * do not match the specified criteria. If a valid task cannot be generated, the existing task will |
| 33 | * be discarded. |
| 34 | * This is mainly meant for updating tasks after the recommendation algorithm changes. |
| 35 | */ |
| 36 | class RevalidateLinkRecommendations extends Maintenance { |
| 37 | |
| 38 | /** @var TitleFactory */ |
| 39 | private $titleFactory; |
| 40 | |
| 41 | /** @var LinkRecommendationStore */ |
| 42 | private $linkRecommendationStore; |
| 43 | |
| 44 | /** @var LinkRecommendationHelper */ |
| 45 | private $linkRecommendationHelper; |
| 46 | |
| 47 | /** @var LinkRecommendationUpdater */ |
| 48 | private $linkRecommendationUpdater; |
| 49 | |
| 50 | /** @var Config */ |
| 51 | private $growthConfig; |
| 52 | |
| 53 | /** @var string[] */ |
| 54 | private $allowedChecksums; |
| 55 | |
| 56 | private ?int $olderThanTimestamp = null; |
| 57 | |
| 58 | public function __construct() { |
| 59 | parent::__construct(); |
| 60 | $this->requireExtension( 'GrowthExperiments' ); |
| 61 | |
| 62 | $this->addDescription( 'Iterate through the growthexperiments_link_recommendations table and ' |
| 63 | . 'regenerate the ones which do not match the specified criteria. If a valid task cannot be ' |
| 64 | . 'generated, the existing task will be discarded.' ); |
| 65 | $this->addOption( 'fromPageId', 'Start iterating upwards from this page ID.', false, true ); |
| 66 | $this->addOption( 'all', 'Regenerate all tasks.' ); |
| 67 | $this->addOption( 'exceptDatasetChecksums', 'Regenerate a task unless its ' |
| 68 | . 'model checksum appears in the given file (one checksum per line)', false, true ); |
| 69 | $this->addOption( 'olderThan', 'Regenerate a task which was generated ' |
| 70 | . 'before this date', false, true ); |
| 71 | $this->addOption( 'scoreLessThan', 'Regenerate a task when any suggested link has ' |
| 72 | . 'a lower score than this one.', false, true ); |
| 73 | $this->addOption( 'limit', 'Limit the number of changes.', false, true ); |
| 74 | $this->addOption( 'force', 'Store the new recommendation even if it fails quality criteria.' ); |
| 75 | $this->addOption( 'dry-run', 'Do not actually make any changes.' ); |
| 76 | $this->addOption( 'verbose', 'Show debug output.' ); |
| 77 | $this->setBatchSize( 500 ); |
| 78 | } |
| 79 | |
| 80 | public function checkRequiredExtensions() { |
| 81 | // Hack: must be early enough for requireExtension to work but late enough for config |
| 82 | // to be available. |
| 83 | $growthServices = GrowthExperimentsServices::wrap( MediaWikiServices::getInstance() ); |
| 84 | if ( $growthServices->getGrowthConfig()->get( 'GELinkRecommendationsUseEventGate' ) ) { |
| 85 | $this->requireExtension( 'EventBus' ); |
| 86 | } else { |
| 87 | $this->requireExtension( 'CirrusSearch' ); |
| 88 | } |
| 89 | parent::checkRequiredExtensions(); |
| 90 | } |
| 91 | |
| 92 | public function execute() { |
| 93 | $this->initGrowthConfig(); |
| 94 | if ( !$this->growthConfig->get( 'GENewcomerTasksLinkRecommendationsEnabled' ) ) { |
| 95 | $this->output( "Disabled\n" ); |
| 96 | return; |
| 97 | } elseif ( $this->growthConfig->get( 'GENewcomerTasksRemoteApiUrl' ) ) { |
| 98 | $this->output( "Local tasks disabled\n" ); |
| 99 | return; |
| 100 | } |
| 101 | $this->initServices(); |
| 102 | |
| 103 | $this->output( "Revalidating link recommendations:\n" ); |
| 104 | |
| 105 | $replaced = $discarded = 0; |
| 106 | $fromPageId = (int)$this->getOption( 'fromPageId', 0 ); |
| 107 | while ( $fromPageId !== false ) { |
| 108 | $this->output( " fetching task batch starting with page $fromPageId\n" ); |
| 109 | $linkRecommendations = $this->linkRecommendationStore->getAllRecommendations( |
| 110 | $this->getBatchSize(), $fromPageId ); |
| 111 | foreach ( $linkRecommendations as $linkRecommendation ) { |
| 112 | if ( !$this->validateRecommendation( $linkRecommendation ) ) { |
| 113 | $this->verboseLog( ' ' . $this->getTitle( $linkRecommendation )->getPrefixedText() |
| 114 | . ' is outdated, regenerating... ' ); |
| 115 | if ( $this->getOption( 'dry-run' ) ) { |
| 116 | $replaced++; |
| 117 | $this->verboseLog( "(dry-run)\n" ); |
| 118 | } else { |
| 119 | $status = $this->regenerateRecommendation( $linkRecommendation ); |
| 120 | $this->verboseLog( $status->isOK() ? "success\n" |
| 121 | : $status->getWikiText( false, false, 'en' ) . "\n" ); |
| 122 | $replaced += $status->isOK() ? 1 : 0; |
| 123 | $discarded += $status->isOK() ? 0 : 1; |
| 124 | } |
| 125 | if ( $replaced + $discarded == $this->getOption( 'limit', -1 ) ) { |
| 126 | $this->verboseLog( "Limit reached, aborting.\n" ); |
| 127 | break 2; |
| 128 | } |
| 129 | } |
| 130 | } |
| 131 | } |
| 132 | $this->output( "Done; replaced $replaced, discarded $discarded\n" ); |
| 133 | } |
| 134 | |
| 135 | protected function initGrowthConfig(): void { |
| 136 | // Needs to be separate from initServices/initConfig as checking whether the script |
| 137 | // should run on a given wiki relies on this, but initServices/initConfig will break |
| 138 | // on some wikis where the script is not supposed to run and the task configuration |
| 139 | // is missing. |
| 140 | $services = MediaWikiServices::getInstance(); |
| 141 | $growthServices = GrowthExperimentsServices::wrap( $services ); |
| 142 | $this->growthConfig = $growthServices->getGrowthConfig(); |
| 143 | } |
| 144 | |
| 145 | protected function initServices(): void { |
| 146 | $services = MediaWikiServices::getInstance(); |
| 147 | $growthServices = GrowthExperimentsServices::wrap( $services ); |
| 148 | $this->titleFactory = $services->getTitleFactory(); |
| 149 | $this->linkRecommendationStore = $growthServices->getLinkRecommendationStore(); |
| 150 | $this->linkRecommendationHelper = $growthServices->getLinkRecommendationHelper(); |
| 151 | $this->linkRecommendationUpdater = $growthServices->getLinkRecommendationUpdater(); |
| 152 | } |
| 153 | |
| 154 | /** |
| 155 | * Check whether the recommendation still meets our standards. |
| 156 | * @param LinkRecommendation $linkRecommendation |
| 157 | * @return bool |
| 158 | */ |
| 159 | private function validateRecommendation( LinkRecommendation $linkRecommendation ): bool { |
| 160 | if ( $this->hasOption( 'all' ) ) { |
| 161 | return false; |
| 162 | } |
| 163 | if ( $this->hasOption( 'exceptDatasetChecksums' ) ) { |
| 164 | $allowedChecksums = $this->getAllowedChecksums(); |
| 165 | $actualChecksum = $linkRecommendation->getMetadata()->getDatasetChecksums()['model'] ?? 'wrong'; |
| 166 | |
| 167 | // Abort if the recommendation is invalid and give chance to other checks |
| 168 | if ( !in_array( $actualChecksum, $allowedChecksums, true ) ) { |
| 169 | return false; |
| 170 | } |
| 171 | } |
| 172 | if ( $this->hasOption( 'olderThan' ) ) { |
| 173 | // Abort if the recommendation is invalid and give chance to other checks |
| 174 | if ( |
| 175 | $linkRecommendation->getMetadata()->getTaskTimestamp() < |
| 176 | $this->getOlderThanTimestamp() |
| 177 | ) { |
| 178 | return false; |
| 179 | } |
| 180 | } |
| 181 | if ( $this->hasOption( 'scoreLessThan' ) ) { |
| 182 | $recommendationScore = min( array_map( static function ( LinkRecommendationLink $link ) { |
| 183 | return $link->getScore(); |
| 184 | }, $linkRecommendation->getLinks() ) ); |
| 185 | |
| 186 | // Abort if the recommendation is invalid and give chance to other checks |
| 187 | if ( $recommendationScore < (float)$this->getOption( 'scoreLessThan' ) ) { |
| 188 | return false; |
| 189 | } |
| 190 | } |
| 191 | return true; |
| 192 | } |
| 193 | |
| 194 | /** |
| 195 | * Discard the existing recommendation and try to fetch a new one. |
| 196 | * @param LinkRecommendation $linkRecommendation |
| 197 | * @return Status |
| 198 | */ |
| 199 | private function regenerateRecommendation( LinkRecommendation $linkRecommendation ): Status { |
| 200 | $title = $this->titleFactory->newFromLinkTarget( $linkRecommendation->getTitle() ); |
| 201 | // Deleting from the search index is instantaneous, adding to the search index takes a few |
| 202 | // hours, so in theory the script could deplete the existing pool temporarily. In practice |
| 203 | // fetching a new candidate takes about a second, so the script shouldn't progess fast |
| 204 | // enough for that to be a problem. |
| 205 | $this->linkRecommendationHelper->deleteLinkRecommendation( $title->toPageIdentity(), true ); |
| 206 | try { |
| 207 | $force = $this->hasOption( 'force' ); |
| 208 | return Status::wrap( $this->linkRecommendationUpdater->processCandidate( $title, $force ) ); |
| 209 | } catch ( DBReadOnlyError $e ) { |
| 210 | $this->fatalError( 'DB is readonly, aborting' ); |
| 211 | } catch ( WikiConfigException $e ) { |
| 212 | $this->fatalError( $e->getMessage() ); |
| 213 | } |
| 214 | throw new LogicException( 'Cannot reach here' ); |
| 215 | } |
| 216 | |
| 217 | private function getTitle( LinkRecommendation $linkRecommendation ): Title { |
| 218 | // The title is already cached by this point so no need for a LinkBatch. |
| 219 | return $this->titleFactory->newFromLinkTarget( $linkRecommendation->getTitle() ); |
| 220 | } |
| 221 | |
| 222 | private function verboseLog( string $message ): void { |
| 223 | if ( $this->hasOption( 'verbose' ) ) { |
| 224 | $this->output( $message ); |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | /** |
| 229 | * Helper method to handle caching of the checksum file. |
| 230 | * @return string[] |
| 231 | */ |
| 232 | private function getAllowedChecksums(): array { |
| 233 | if ( !$this->allowedChecksums ) { |
| 234 | $filename = $this->getOption( 'exceptDatasetChecksums' ); |
| 235 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 236 | $content = @file_get_contents( $filename ); |
| 237 | if ( $content === false ) { |
| 238 | throw new UnexpectedValueException( "File $filename could not be opened" ); |
| 239 | } |
| 240 | $this->allowedChecksums = array_filter( |
| 241 | array_map( |
| 242 | 'trim', |
| 243 | file( $filename, FILE_IGNORE_NEW_LINES ) |
| 244 | ) |
| 245 | ); |
| 246 | if ( !$this->allowedChecksums ) { |
| 247 | throw new UnexpectedValueException( "File $filename did not contain checksums" ); |
| 248 | } |
| 249 | } |
| 250 | return $this->allowedChecksums; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * Helper method to handle caching/fetching of the older than timestamp |
| 255 | * @return int |
| 256 | */ |
| 257 | private function getOlderThanTimestamp(): int { |
| 258 | if ( !$this->olderThanTimestamp ) { |
| 259 | $rawTS = wfTimestamp( |
| 260 | TS_UNIX, |
| 261 | $this->getOption( 'olderThan' ) |
| 262 | ); |
| 263 | if ( !$rawTS ) { |
| 264 | throw new UnexpectedValueException( "Parameter olderThan does not contain a valid timestamp" ); |
| 265 | } |
| 266 | $this->olderThanTimestamp = (int)$rawTS; |
| 267 | } |
| 268 | return $this->olderThanTimestamp; |
| 269 | } |
| 270 | |
| 271 | } |
| 272 | |
| 273 | $maintClass = RevalidateLinkRecommendations::class; |
| 274 | require_once RUN_MAINTENANCE_IF_MAIN; |