ActorStore.php

Go to the documentation of this file.

<?php
namespace MediaWiki\User;
 
use CannotCreateActorException;
use DBAccessObjectUtils;
use ExternalUserNames;
use InvalidArgumentException;
use MediaWiki\DAO\WikiAwareEntity;
use Psr\Log\LoggerInterface;
use stdClass;
use User;
use Wikimedia\Assert\Assert;
use Wikimedia\IPUtils;
use Wikimedia\Rdbms\DBQueryError;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\ILoadBalancer;
 
class ActorStore implements UserIdentityLookup, ActorNormalization {
 
    public const UNKNOWN_USER_NAME = 'Unknown user';
 
    private const LOCAL_CACHE_SIZE = 100;
 
    private $loadBalancer;
 
    private $userNameUtils;
 
    private $logger;
 
    private $wikiId;
 
    private $cache;
 
    public function __construct(
        ILoadBalancer $loadBalancer,
        UserNameUtils $userNameUtils,
        LoggerInterface $logger,
        $wikiId = WikiAwareEntity::LOCAL
    ) {
        Assert::parameterType( [ 'string', 'false' ], $wikiId, '$wikiId' );
 
        $this->loadBalancer = $loadBalancer;
        $this->userNameUtils = $userNameUtils;
        $this->logger = $logger;
        $this->wikiId = $wikiId;
 
        $this->cache = new ActorCache( self::LOCAL_CACHE_SIZE );
    }
 
    public function newActorFromRow( stdClass $row ): UserIdentity {
        $actorId = (int)$row->actor_id;
        $userId = isset( $row->actor_user ) ? (int)$row->actor_user : 0;
        if ( $actorId === 0 ) {
            throw new InvalidArgumentException( "Actor ID is 0 for {$row->actor_name} and {$userId}" );
        }
 
        $normalizedName = $this->normalizeUserName( $row->actor_name );
        if ( $normalizedName === null ) {
            $this->logger->warning( 'Encountered invalid actor name in database', [
                'user_id' => $userId,
                'actor_id' => $actorId,
                'actor_name' => $row->actor_name,
                'wiki_id' => $this->wikiId ?: 'local'
            ] );
            // TODO: once we have guaranteed db only contains valid actor names,
            // we can skip normalization here - T273933
            if ( $row->actor_name === '' ) {
                throw new InvalidArgumentException( "Actor name can not be empty for {$userId} and {$actorId}" );
            }
        }
 
        $actor = new UserIdentityValue( $userId, $row->actor_name, $this->wikiId );
        $this->cache->add( $actorId, $actor );
        return $actor;
    }
 
    public function newActorFromRowFields( $userId, $name, $actorId ): UserIdentity {
        // For backwards compatibility we are quite relaxed about what to accept,
        // but try not to create entirely incorrect objects. As we move more code
        // from ActorMigration aliases to proper join with the actor table,
        // we should use ::newActorFromRow more, and eventually deprecate this method.
        $userId = $userId === null ? 0 : (int)$userId;
        $name ??= '';
        if ( $actorId === null ) {
            throw new InvalidArgumentException( "Actor ID is null for {$name} and {$userId}" );
        }
        if ( (int)$actorId === 0 ) {
            throw new InvalidArgumentException( "Actor ID is 0 for {$name} and {$userId}" );
        }
 
        $normalizedName = $this->normalizeUserName( $name );
        if ( $normalizedName === null ) {
            $this->logger->warning( 'Encountered invalid actor name in database', [
                'user_id' => $userId,
                'actor_id' => $actorId,
                'actor_name' => $name,
                'wiki_id' => $this->wikiId ?: 'local'
            ] );
            // TODO: once we have guaranteed the DB entries only exist for normalized names,
            // we can skip normalization here - T273933
            if ( $name === '' ) {
                throw new InvalidArgumentException( "Actor name can not be empty for {$userId} and {$actorId}" );
            }
        }
 
        $actorId = (int)$actorId;
        $actor = new UserIdentityValue(
            $userId,
            $name,
            $this->wikiId
        );
 
        $this->cache->add( $actorId, $actor );
        return $actor;
    }
 
    public function deleteUserIdentityFromCache( UserIdentity $actor ) {
        $this->cache->remove( $actor );
    }
 
    public function getActorById( int $actorId, IDatabase $db ): ?UserIdentity {
        $this->checkDatabaseDomain( $db );
 
        if ( !$actorId ) {
            return null;
        }
 
        return $this->cache->getActor( ActorCache::KEY_ACTOR_ID, $actorId ) ??
            $this->newSelectQueryBuilder( $db )
                ->caller( __METHOD__ )
                ->conds( [ 'actor_id' => $actorId ] )
                ->fetchUserIdentity() ??
            // The actor ID mostly comes from DB, so if we can't find an actor by ID,
            // it's most likely due to lagged replica and not cause it doesn't actually exist.
            // Probably we just inserted it? Try primary database.
            $this->newSelectQueryBuilder( self::READ_LATEST )
                ->caller( __METHOD__ )
                ->conds( [ 'actor_id' => $actorId ] )
                ->fetchUserIdentity();
    }
 
    public function getUserIdentityByName( string $name, int $queryFlags = self::READ_NORMAL ): ?UserIdentity {
        $normalizedName = $this->normalizeUserName( $name );
        if ( $normalizedName === null ) {
            return null;
        }
 
        return $this->cache->getActor( ActorCache::KEY_USER_NAME, $normalizedName ) ??
            $this->newSelectQueryBuilder( $queryFlags )
                ->caller( __METHOD__ )
                ->whereUserNames( $normalizedName )
                ->fetchUserIdentity();
    }
 
    public function getUserIdentityByUserId( int $userId, int $queryFlags = self::READ_NORMAL ): ?UserIdentity {
        if ( !$userId ) {
            return null;
        }
 
        return $this->cache->getActor( ActorCache::KEY_USER_ID, $userId ) ??
            $this->newSelectQueryBuilder( $queryFlags )
                ->caller( __METHOD__ )
                ->whereUserIds( $userId )
                ->fetchUserIdentity();
    }
 
    private function attachActorId( UserIdentity $user, int $id, bool $assigned ) {
        if ( $user instanceof User ) {
            $user->setActorId( $id );
            if ( $assigned ) {
                $user->invalidateCache();
            }
        }
    }
 
    private function detachActorId( UserIdentity $user ) {
        if ( $user instanceof User ) {
            $user->setActorId( 0 );
        }
    }
 
    public function findActorId( UserIdentity $user, IDatabase $db ): ?int {
        // TODO: we want to assert this user belongs to the correct wiki,
        // but User objects are always local and we used to use them
        // on a non-local DB connection. We need to first deprecate this
        // possibility and then throw on mismatching User object - T273972
        // $user->assertWiki( $this->wikiId );
        $this->deprecateInvalidCrossWikiParam( $user );
 
        // TODO: In the future we would be able to assume UserIdentity name is ok
        // and will be able to skip normalization here - T273933
        $name = $this->normalizeUserName( $user->getName() );
        if ( $name === null ) {
            $this->logger->warning( 'Encountered a UserIdentity with invalid name', [
                'user_name' => $user->getName()
            ] );
            return null;
        }
 
        $id = $this->findActorIdInternal( $name, $db );
 
        // Set the actor ID in the User object. To be removed, see T274148.
        if ( $id && $user instanceof User ) {
            $user->setActorId( $id );
        }
 
        return $id;
    }
 
    public function findActorIdByName( $name, IDatabase $db ): ?int {
        $name = $this->normalizeUserName( $name );
        if ( $name === null ) {
            return null;
        }
 
        return $this->findActorIdInternal( $name, $db );
    }
 
    private function findActorIdInternal(
        string $name,
        IDatabase $db,
        array $queryOptions = []
    ): ?int {
        // Note: UserIdentity::getActorId will be deprecated and removed,
        // and this is the replacement for it. Can't call User::getActorId, cause
        // User always thinks it's local, so we could end up fetching the ID
        // from the wrong database.
 
        $cachedValue = $this->cache->getActorId( ActorCache::KEY_USER_NAME, $name );
        if ( $cachedValue ) {
            return $cachedValue;
        }
 
        $row = $db->selectRow(
            'actor',
            [ 'actor_user', 'actor_name', 'actor_id' ],
            [ 'actor_name' => $name ],
            __METHOD__,
            $queryOptions
        );
 
        if ( !$row || !$row->actor_id ) {
            return null;
        }
        // to cache row
        $this->newActorFromRow( $row );
 
        return (int)$row->actor_id;
    }
 
    public function acquireActorId( UserIdentity $user, IDatabase $dbw ): int {
        $this->checkDatabaseDomain( $dbw );
        [ $userId, $userName ] = $this->validateActorForInsertion( $user );
 
        // allow cache to be used, because if it is in the cache, it already has an actor ID
        $existingActorId = $this->findActorIdInternal( $userName, $dbw );
        if ( $existingActorId ) {
            $this->attachActorId( $user, $existingActorId, false );
            return $existingActorId;
        }
 
        $dbw->insert(
            'actor',
            [
                'actor_user' => $userId,
                'actor_name' => $userName,
            ],
            __METHOD__,
            [ 'IGNORE' ]
        );
 
        if ( $dbw->affectedRows() ) {
            $actorId = $dbw->insertId();
        } else {
            // Outdated cache?
            // Use LOCK IN SHARE MODE to bypass any MySQL REPEATABLE-READ snapshot.
            $actorId = $this->findActorIdInternal(
                $userName,
                $dbw,
                [ 'LOCK IN SHARE MODE' ]
            );
            if ( !$actorId ) {
                throw new CannotCreateActorException(
                    "Failed to create actor ID for " .
                    "user_id={$userId} user_name=\"{$userName}\""
                );
            }
        }
 
        $this->attachActorId( $user, $actorId, true );
        // Cache row we've just created
        $cachedUserIdentity = $this->newActorFromRowFields( $userId, $userName, $actorId );
        $this->setUpRollbackHandler( $dbw, $cachedUserIdentity, $user );
        return $actorId;
    }
 
    public function createNewActor( UserIdentity $user, IDatabase $dbw ): int {
        $this->checkDatabaseDomain( $dbw );
        [ $userId, $userName ] = $this->validateActorForInsertion( $user );
 
        try {
            $dbw->insert(
                'actor',
                [
                    'actor_user' => $userId,
                    'actor_name' => $userName,
                ],
                __METHOD__
            );
        } catch ( DBQueryError $e ) {
            // We rely on the database to crash on unique actor_name constraint.
            throw new CannotCreateActorException( $e->getMessage() );
        }
        $actorId = $dbw->insertId();
 
        $this->attachActorId( $user, $actorId, true );
        // Cache row we've just created
        $cachedUserIdentity = $this->newActorFromRowFields( $userId, $userName, $actorId );
        $this->setUpRollbackHandler( $dbw, $cachedUserIdentity, $user );
 
        return $actorId;
    }
 
    public function acquireSystemActorId( UserIdentity $user, IDatabase $dbw ): int {
        $this->checkDatabaseDomain( $dbw );
        [ $userId, $userName ] = $this->validateActorForInsertion( $user );
 
        $existingActorId = $this->findActorIdInternal( $userName, $dbw );
        if ( $existingActorId ) {
            // It certainly will be cached if we just found it.
            $existingActor = $this->cache->getActor( ActorCache::KEY_ACTOR_ID, $existingActorId );
 
            // If we already have an existing actor with a matching user ID
            // just return it, nothing to do here.
            if ( $existingActor->getId( $this->wikiId ) === $user->getId( $this->wikiId ) ) {
                return $existingActorId;
            }
 
            // Allow overwriting user ID for already existing actor with reserved user name, see T236444
            if ( $this->userNameUtils->isUsable( $userName ) || $existingActor->isRegistered() ) {
                throw new CannotCreateActorException(
                    'Cannot replace user for existing actor: ' .
                    "actor_id=$existingActorId, new user_id=$userId"
                );
            }
        }
 
        $dbw->upsert(
            'actor',
            [ 'actor_name' => $userName, 'actor_user' => $userId ],
            [ [ 'actor_name' ] ],
            [ 'actor_user' => $userId ],
            __METHOD__
        );
        if ( !$dbw->affectedRows() ) {
            throw new CannotCreateActorException(
                'Failed to replace user for actor: ' .
                "actor_id=$existingActorId, new user_id=$userId"
            );
        }
        $actorId = $dbw->insertId() ?: $existingActorId;
 
        $this->cache->remove( $user );
        $this->attachActorId( $user, $actorId, true );
        // Cache row we've just created
        $cachedUserIdentity = $this->newActorFromRowFields( $userId, $userName, $actorId );
        $this->setUpRollbackHandler( $dbw, $cachedUserIdentity, $user );
        return $actorId;
    }
 
    public function deleteActor( UserIdentity $actor, IDatabase $dbw ): bool {
        $this->checkDatabaseDomain( $dbw );
        $this->deprecateInvalidCrossWikiParam( $actor );
 
        $normalizedName = $this->normalizeUserName( $actor->getName() );
        if ( $normalizedName === null ) {
            throw new InvalidArgumentException(
                "Unable to normalize the provided actor name {$actor->getName()}"
            );
        }
        $dbw->delete(
            'actor',
            [ 'actor_name' => $normalizedName ],
            __METHOD__
        );
        if ( $dbw->affectedRows() !== 0 ) {
            $this->cache->remove( $actor );
            return true;
        }
        return false;
    }
 
    public function normalizeUserName( string $name ): ?string {
        if ( $this->userNameUtils->isIP( $name ) ) {
            return IPUtils::sanitizeIP( $name );
        } elseif ( ExternalUserNames::isExternal( $name ) ) {
            // TODO: ideally, we should probably canonicalize external usernames,
            // but it was not done before, so we can not start doing it unless we
            // fix existing DB rows - T273933
            return $name;
        } else {
            $normalized = $this->userNameUtils->getCanonical( $name );
            return $normalized === false ? null : $normalized;
        }
    }
 
    private function validateActorForInsertion( UserIdentity $user ): array {
        // TODO: we want to assert this user belongs to the correct wiki,
        // but User objects are always local and we used to use them
        // on a non-local DB connection. We need to first deprecate this
        // possibility and then throw on mismatching User object - T273972
        // $user->assertWiki( $this->wikiId );
        $this->deprecateInvalidCrossWikiParam( $user );
 
        $userName = $this->normalizeUserName( $user->getName() );
        if ( $userName === null || $userName === '' ) {
            $userIdForErrorMessage = $user->getId( $this->wikiId );
            throw new CannotCreateActorException(
                'Cannot create an actor for a user with no name: ' .
                "user_id={$userIdForErrorMessage} user_name=\"{$user->getName()}\""
            );
        }
 
        $userId = $user->getId( $this->wikiId ) ?: null;
        if ( $userId === null && $this->userNameUtils->isUsable( $user->getName() ) ) {
            throw new CannotCreateActorException(
                'Cannot create an actor for a usable name that is not an existing user: ' .
                "user_name=\"{$user->getName()}\""
            );
        }
        return [ $userId, $userName ];
    }
 
    private function setUpRollbackHandler(
        IDatabase $dbw,
        UserIdentity $cachedActor,
        UserIdentity $originalActor
    ) {
        if ( $dbw->trxLevel() ) {
            // If called within a transaction and it was rolled back, the cached actor ID
            // becomes invalid, so cache needs to be invalidated as well. See T277795.
            $dbw->onTransactionResolution(
                function ( int $trigger ) use ( $cachedActor, $originalActor ) {
                    if ( $trigger === IDatabase::TRIGGER_ROLLBACK ) {
                        $this->cache->remove( $cachedActor );
                        $this->detachActorId( $originalActor );
                    }
                },
                __METHOD__
            );
        }
    }
 
    private function getDBConnectionRefForQueryFlags( int $queryFlags ): array {
        [ $mode, $options ] = DBAccessObjectUtils::getDBOptions( $queryFlags );
        return [ $this->loadBalancer->getConnectionRef( $mode, [], $this->wikiId ), $options ];
    }
 
    private function checkDatabaseDomain( IDatabase $db ) {
        $dbDomain = $db->getDomainID();
        $storeDomain = $this->loadBalancer->resolveDomainID( $this->wikiId );
        if ( $dbDomain !== $storeDomain ) {
            throw new InvalidArgumentException(
                "DB connection domain '$dbDomain' does not match '$storeDomain'"
            );
        }
    }
 
    public function getUnknownActor(): UserIdentity {
        $actor = $this->getUserIdentityByName( self::UNKNOWN_USER_NAME );
        if ( $actor ) {
            return $actor;
        }
        $actor = new UserIdentityValue( 0, self::UNKNOWN_USER_NAME, $this->wikiId );
 
        [ $db, ] = $this->getDBConnectionRefForQueryFlags( self::READ_LATEST );
        $this->acquireActorId( $actor, $db );
        return $actor;
    }
 
    public function newSelectQueryBuilder( $dbOrQueryFlags = self::READ_NORMAL ): UserSelectQueryBuilder {
        if ( $dbOrQueryFlags instanceof IDatabase ) {
            [ $db, $options ] = [ $dbOrQueryFlags, [] ];
            $this->checkDatabaseDomain( $db );
        } else {
            [ $db, $options ] = $this->getDBConnectionRefForQueryFlags( $dbOrQueryFlags );
        }
 
        return ( new UserSelectQueryBuilder( $db, $this ) )->options( $options );
    }
 
    private function deprecateInvalidCrossWikiParam( UserIdentity $user ) {
        if ( $user->getWikiId() !== $this->wikiId ) {
            $expected = $this->wikiIdToString( $user->getWikiId() );
            $actual = $this->wikiIdToString( $this->wikiId );
            wfDeprecatedMsg(
                'Deprecated passing invalid cross-wiki user. ' .
                "Expected: {$expected}, Actual: {$actual}.",
                '1.37'
            );
        }
    }
 
    private function wikiIdToString( $wikiId ): string {
        return $wikiId === WikiAwareEntity::LOCAL ? 'the local wiki' : "'{$wikiId}'";
    }
}