PageHistoryCountHandler.php

Go to the documentation of this file.

<?php
 
namespace MediaWiki\Rest\Handler;
 
use ActorMigration;
use ChangeTags;
use MediaWiki\Page\ExistingPageRecord;
use MediaWiki\Page\PageLookup;
use MediaWiki\Permissions\GroupPermissionsLookup;
use MediaWiki\Rest\LocalizedHttpException;
use MediaWiki\Rest\Response;
use MediaWiki\Rest\SimpleHandler;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\RevisionStore;
use MediaWiki\Storage\NameTableAccessException;
use MediaWiki\Storage\NameTableStore;
use MediaWiki\Storage\NameTableStoreFactory;
use WANObjectCache;
use Wikimedia\Message\MessageValue;
use Wikimedia\Message\ParamType;
use Wikimedia\Message\ScalarParam;
use Wikimedia\ParamValidator\ParamValidator;
use Wikimedia\Rdbms\ILoadBalancer;
 
class PageHistoryCountHandler extends SimpleHandler {
    private const COUNT_LIMITS = [
        'anonymous' => 10000,
        'bot' => 10000,
        'editors' => 25000,
        'edits' => 30000,
        'minor' => 1000,
        'reverted' => 30000
    ];
 
    private const DEPRECATED_COUNT_TYPES = [
        'anonedits' => 'anonymous',
        'botedits' => 'bot',
        'revertededits' => 'reverted'
    ];
 
    private const MAX_AGE_200 = 60;
 
    private $revisionStore;
 
    private $changeTagDefStore;
 
    private $groupPermissionsLookup;
 
    private $loadBalancer;
 
    private $pageLookup;
 
    private $cache;
 
    private $actorMigration;
 
    private $revision = false;
 
    private $lastModifiedTimes;
 
    private $page = false;
 
    public function __construct(
        RevisionStore $revisionStore,
        NameTableStoreFactory $nameTableStoreFactory,
        GroupPermissionsLookup $groupPermissionsLookup,
        ILoadBalancer $loadBalancer,
        WANObjectCache $cache,
        PageLookup $pageLookup,
        ActorMigration $actorMigration
    ) {
        $this->revisionStore = $revisionStore;
        $this->changeTagDefStore = $nameTableStoreFactory->getChangeTagDef();
        $this->groupPermissionsLookup = $groupPermissionsLookup;
        $this->loadBalancer = $loadBalancer;
        $this->cache = $cache;
        $this->pageLookup = $pageLookup;
        $this->actorMigration = $actorMigration;
    }
 
    private function normalizeType( $type ) {
        return self::DEPRECATED_COUNT_TYPES[$type] ?? $type;
    }
 
    private function validateParameterCombination( $type ) {
        $params = $this->getValidatedParams();
        if ( !$params ) {
            return;
        }
 
        if ( $params['from'] || $params['to'] ) {
            if ( $type === 'edits' || $type === 'editors' ) {
                if ( !$params['from'] || !$params['to'] ) {
                    throw new LocalizedHttpException(
                        new MessageValue( 'rest-pagehistorycount-parameters-invalid' ),
                        400
                    );
                }
            } else {
                throw new LocalizedHttpException(
                    new MessageValue( 'rest-pagehistorycount-parameters-invalid' ),
                    400
                );
            }
        }
    }
 
    public function run( $title, $type ) {
        $normalizedType = $this->normalizeType( $type );
        $this->validateParameterCombination( $normalizedType );
        $page = $this->getPage();
        if ( !$page ) {
            throw new LocalizedHttpException(
                new MessageValue( 'rest-nonexistent-title',
                    [ new ScalarParam( ParamType::PLAINTEXT, $title ) ]
                ),
                404
            );
        }
 
        if ( !$this->getAuthority()->authorizeRead( 'read', $page ) ) {
            throw new LocalizedHttpException(
                new MessageValue( 'rest-permission-denied-title',
                    [ new ScalarParam( ParamType::PLAINTEXT, $title ) ]
                ),
                403
            );
        }
 
        $count = $this->getCount( $normalizedType );
        $countLimit = self::COUNT_LIMITS[$normalizedType];
        $response = $this->getResponseFactory()->createJson( [
                'count' => $count > $countLimit ? $countLimit : $count,
                'limit' => $count > $countLimit
        ] );
        $response->setHeader( 'Cache-Control', 'max-age=' . self::MAX_AGE_200 );
 
        // Inform clients who use a deprecated "type" value, so they can adjust
        if ( isset( self::DEPRECATED_COUNT_TYPES[$type] ) ) {
            $docs = '<https://www.mediawiki.org/wiki/API:REST/History_API' .
                '#Get_page_history_counts>; rel="deprecation"';
            $response->setHeader( 'Deprecation', 'version="v1"' );
            $response->setHeader( 'Link', $docs );
        }
 
        return $response;
    }
 
    private function getCount( $type ) {
        $pageId = $this->getPage()->getId();
        switch ( $type ) {
            case 'anonymous':
                return $this->getCachedCount( $type,
                    function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                        return $this->getAnonCount( $pageId, $fromRev );
                    }
                );
 
            case 'bot':
                return $this->getCachedCount( $type,
                    function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                        return $this->getBotCount( $pageId, $fromRev );
                    }
                );
 
            case 'editors':
                $from = $this->getValidatedParams()['from'] ?? null;
                $to = $this->getValidatedParams()['to'] ?? null;
                if ( $from || $to ) {
                    return $this->getEditorsCount(
                        $pageId,
                        $from ? $this->getRevisionOrThrow( $from ) : null,
                        $to ? $this->getRevisionOrThrow( $to ) : null
                    );
                } else {
                    return $this->getCachedCount( $type,
                        function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                            return $this->getEditorsCount( $pageId, $fromRev );
                        } );
                }
 
            case 'edits':
                $from = $this->getValidatedParams()['from'] ?? null;
                $to = $this->getValidatedParams()['to'] ?? null;
                if ( $from || $to ) {
                    return $this->getEditsCount(
                        $pageId,
                        $from ? $this->getRevisionOrThrow( $from ) : null,
                        $to ? $this->getRevisionOrThrow( $to ) : null
                    );
                } else {
                    return $this->getCachedCount( $type,
                        function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                            return $this->getEditsCount( $pageId, $fromRev );
                        }
                    );
                }
 
            case 'reverted':
                return $this->getCachedCount( $type,
                    function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                        return $this->getRevertedCount( $pageId, $fromRev );
                    }
                );
 
            case 'minor':
                // The query for minor counts is inefficient for the database for pages with many revisions.
                // If the specified title contains more revisions than allowed, we will return an error.
                $editsCount = $this->getCachedCount( 'edits',
                    function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                        return $this->getEditsCount( $pageId, $fromRev );
                    }
                );
                if ( $editsCount > self::COUNT_LIMITS[$type] * 2 ) {
                    throw new LocalizedHttpException(
                        new MessageValue( 'rest-pagehistorycount-too-many-revisions' ),
                        500
                    );
                }
                return $this->getCachedCount( $type,
                    function ( RevisionRecord $fromRev = null ) use ( $pageId ) {
                        return $this->getMinorCount( $pageId, $fromRev );
                    }
                );
 
            default:
                throw new LocalizedHttpException(
                    new MessageValue( 'rest-pagehistorycount-type-unrecognized',
                        [ new ScalarParam( ParamType::PLAINTEXT, $type ) ]
                    ),
                    500
                );
        }
    }
 
    private function getCurrentRevision(): ?RevisionRecord {
        if ( $this->revision === false ) {
            $page = $this->getPage();
            if ( $page ) {
                $this->revision = $this->revisionStore->getKnownCurrentRevision( $page ) ?: null;
            } else {
                $this->revision = null;
            }
        }
        return $this->revision;
    }
 
    private function getPage(): ?ExistingPageRecord {
        if ( $this->page === false ) {
            $this->page = $this->pageLookup->getExistingPageByText(
                $this->getValidatedParams()['title']
            );
        }
        return $this->page;
    }
 
    protected function getLastModified() {
        $lastModifiedTimes = $this->getLastModifiedTimes();
        if ( $lastModifiedTimes ) {
            return max( array_values( $lastModifiedTimes ) );
        }
        return null;
    }
 
    protected function getLastModifiedTimes() {
        $currentRev = $this->getCurrentRevision();
        if ( !$currentRev ) {
            return null;
        }
        if ( $this->lastModifiedTimes === null ) {
            $currentRevTime = (int)wfTimestampOrNull( TS_UNIX, $currentRev->getTimestamp() );
            $loggingTableTime = $this->loggingTableTime( $currentRev->getPageId() );
            $this->lastModifiedTimes = [
                'currentRevTS' => $currentRevTime,
                'dependencyModTS' => $loggingTableTime
            ];
        }
        return $this->lastModifiedTimes;
    }
 
    private function loggingTableTime( $pageId ) {
        $res = $this->loadBalancer->getConnectionRef( DB_REPLICA )->selectField(
            'logging',
            'MAX(log_timestamp)',
            [ 'log_page' => $pageId ],
            __METHOD__
        );
        return $res ? (int)wfTimestamp( TS_UNIX, $res ) : null;
    }
 
    protected function getEtag() {
        return null;
    }
 
    private function getCachedCount( $type,
        callable $fetchCount
    ) {
        $pageId = $this->getPage()->getId();
        return $this->cache->getWithSetCallback(
            $this->cache->makeKey( 'rest', 'pagehistorycount', $pageId, $type ),
            WANObjectCache::TTL_WEEK,
            function ( $oldValue ) use ( $fetchCount ) {
                $currentRev = $this->getCurrentRevision();
                if ( $oldValue ) {
                    // Last modified timestamp was NOT a dependency change (e.g. revdel)
                    $doIncrementalUpdate = (
                        // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
                        $this->getLastModified() != $this->getLastModifiedTimes()['dependencyModTS']
                    );
                    if ( $doIncrementalUpdate ) {
                        $rev = $this->revisionStore->getRevisionById( $oldValue['revision'] );
                        if ( $rev ) {
                            $additionalCount = $fetchCount( $rev );
                            return [
                                'revision' => $currentRev->getId(),
                                'count' => $oldValue['count'] + $additionalCount,
                                // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
                                'dependencyModTS' => $this->getLastModifiedTimes()['dependencyModTS']
                            ];
                        }
                    }
                }
                // Nothing was previously stored, or incremental update was done for too long,
                // recalculate from scratch.
                return [
                    'revision' => $currentRev->getId(),
                    'count' => $fetchCount(),
                    // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
                    'dependencyModTS' => $this->getLastModifiedTimes()['dependencyModTS']
                ];
            },
            [
                'touchedCallback' => function (){
                    return $this->getLastModified();
                },
                'version' => 2,
                'lockTSE' => WANObjectCache::TTL_MINUTE * 5
            ]
        )['count'];
    }
 
    protected function getAnonCount( $pageId, RevisionRecord $fromRev = null ) {
        $dbr = $this->loadBalancer->getConnectionRef( DB_REPLICA );
 
        $revQuery = $this->actorMigration->getJoin( 'rev_user' );
 
        $cond = [
            'rev_page' => $pageId,
            'actor_user IS NULL',
            $dbr->bitAnd( 'rev_deleted',
                RevisionRecord::DELETED_TEXT | RevisionRecord::DELETED_USER ) . " = 0"
        ];
 
        if ( $fromRev ) {
            $oldTs = $dbr->addQuotes( $dbr->timestamp( $fromRev->getTimestamp() ) );
            $cond[] = "(rev_timestamp = {$oldTs} AND rev_id > {$fromRev->getId()}) " .
                "OR rev_timestamp > {$oldTs}";
        }
 
        $edits = $dbr->selectRowCount(
            [
                'revision',
            ] + $revQuery['tables'],
            '1',
            $cond,
            __METHOD__,
            [ 'LIMIT' => self::COUNT_LIMITS['anonymous'] + 1 ], // extra to detect truncation
            $revQuery['joins']
        );
        return $edits;
    }
 
    protected function getBotCount( $pageId, RevisionRecord $fromRev = null ) {
        $dbr = $this->loadBalancer->getConnectionRef( DB_REPLICA );
 
        $revQuery = $this->actorMigration->getJoin( 'rev_user' );
 
        // This redundant join condition tells MySQL that rev_page and revactor_page are the
        // same, so it can propagate the condition
        if ( isset( $revQuery['tables']['temp_rev_user'] ) /* SCHEMA_COMPAT_READ_TEMP */ ) {
            $revQuery['joins']['temp_rev_user'][1] =
                "temp_rev_user.revactor_rev = rev_id AND revactor_page = rev_page";
        }
 
        $cond = [
            'rev_page=' . intval( $pageId ),
            $dbr->bitAnd( 'rev_deleted',
                RevisionRecord::DELETED_TEXT | RevisionRecord::DELETED_USER ) . " = 0",
            'EXISTS(' .
                $dbr->selectSQLText(
                    'user_groups',
                    '1',
                    [
                        $revQuery['fields']['rev_user'] . ' = ug_user',
                        'ug_group' => $this->groupPermissionsLookup->getGroupsWithPermission( 'bot' ),
                        'ug_expiry IS NULL OR ug_expiry >= ' . $dbr->addQuotes( $dbr->timestamp() )
                    ],
                    __METHOD__
                ) .
            ')'
        ];
        if ( $fromRev ) {
            $oldTs = $dbr->addQuotes( $dbr->timestamp( $fromRev->getTimestamp() ) );
            $cond[] = "(rev_timestamp = {$oldTs} AND rev_id > {$fromRev->getId()}) " .
                "OR rev_timestamp > {$oldTs}";
        }
 
        $edits = $dbr->selectRowCount(
            [
                'revision',
                // @phan-suppress-next-line PhanTypePossiblyInvalidDimOffset False positive
            ] + $revQuery['tables'],
            '1',
            $cond,
            __METHOD__,
            [ 'LIMIT' => self::COUNT_LIMITS['bot'] + 1 ], // extra to detect truncation
            $revQuery['joins']
        );
        return $edits;
    }
 
    protected function getEditorsCount( $pageId,
        RevisionRecord $fromRev = null,
        RevisionRecord $toRev = null
    ) {
        list( $fromRev, $toRev ) = $this->orderRevisions( $fromRev, $toRev );
        return $this->revisionStore->countAuthorsBetween( $pageId, $fromRev,
            $toRev, $this->getAuthority(), self::COUNT_LIMITS['editors'] );
    }
 
    protected function getRevertedCount( $pageId, RevisionRecord $fromRev = null ) {
        $tagIds = [];
 
        foreach ( ChangeTags::REVERT_TAGS as $tagName ) {
            try {
                $tagIds[] = $this->changeTagDefStore->getId( $tagName );
            } catch ( NameTableAccessException $e ) {
                // If no revisions are tagged with a name, no tag id will be present
            }
        }
        if ( !$tagIds ) {
            return 0;
        }
 
        $dbr = $this->loadBalancer->getConnectionRef( DB_REPLICA );
 
        $cond = [
            'rev_page' => $pageId,
            $dbr->bitAnd( 'rev_deleted', RevisionRecord::DELETED_TEXT ) . " = 0"
        ];
        if ( $fromRev ) {
            $oldTs = $dbr->addQuotes( $dbr->timestamp( $fromRev->getTimestamp() ) );
            $cond[] = "(rev_timestamp = {$oldTs} AND rev_id > {$fromRev->getId()}) " .
                "OR rev_timestamp > {$oldTs}";
        }
        $edits = $dbr->selectRowCount(
            [
                'revision',
                'change_tag'
            ],
            '1',
            [ 'rev_page' => $pageId ],
            __METHOD__,
            [
                'LIMIT' => self::COUNT_LIMITS['reverted'] + 1, // extra to detect truncation
                'GROUP BY' => 'rev_id'
            ],
            [
                'change_tag' => [
                    'JOIN',
                    [
                        'ct_rev_id = rev_id',
                        'ct_tag_id' => $tagIds,
                    ]
                ],
            ]
        );
        return $edits;
    }
 
    protected function getMinorCount( $pageId, RevisionRecord $fromRev = null ) {
        $dbr = $this->loadBalancer->getConnectionRef( DB_REPLICA );
        $cond = [
            'rev_page' => $pageId,
            'rev_minor_edit != 0',
            $dbr->bitAnd( 'rev_deleted', RevisionRecord::DELETED_TEXT ) . " = 0"
        ];
        if ( $fromRev ) {
            $oldTs = $dbr->addQuotes( $dbr->timestamp( $fromRev->getTimestamp() ) );
            $cond[] = "(rev_timestamp = {$oldTs} AND rev_id > {$fromRev->getId()}) " .
                "OR rev_timestamp > {$oldTs}";
        }
        $edits = $dbr->selectRowCount( 'revision', '1',
            $cond,
            __METHOD__,
            [ 'LIMIT' => self::COUNT_LIMITS['minor'] + 1 ] // extra to detect truncation
        );
 
        return $edits;
    }
 
    protected function getEditsCount(
        $pageId,
        RevisionRecord $fromRev = null,
        RevisionRecord $toRev = null
    ) {
        list( $fromRev, $toRev ) = $this->orderRevisions( $fromRev, $toRev );
        return $this->revisionStore->countRevisionsBetween(
            $pageId,
            $fromRev,
            $toRev,
            self::COUNT_LIMITS['edits'] // Will be increased by 1 to detect truncation
        );
    }
 
    private function getRevisionOrThrow( $revId ) {
        $rev = $this->revisionStore->getRevisionById( $revId );
        if ( !$rev ) {
            throw new LocalizedHttpException(
                new MessageValue( 'rest-nonexistent-revision', [ $revId ] ),
                404
            );
        }
        return $rev;
    }
 
    private function orderRevisions(
        RevisionRecord $fromRev = null,
        RevisionRecord $toRev = null
    ) {
        if ( $fromRev && $toRev && ( $fromRev->getTimestamp() > $toRev->getTimestamp() ||
                ( $fromRev->getTimestamp() === $toRev->getTimestamp()
                    && $fromRev->getId() > $toRev->getId() ) )
        ) {
            return [ $toRev, $fromRev ];
        }
        return [ $fromRev, $toRev ];
    }
 
    public function needsWriteAccess() {
        return false;
    }
 
    public function getParamSettings() {
        return [
            'title' => [
                self::PARAM_SOURCE => 'path',
                ParamValidator::PARAM_TYPE => 'string',
                ParamValidator::PARAM_REQUIRED => true,
            ],
            'type' => [
                self::PARAM_SOURCE => 'path',
                ParamValidator::PARAM_TYPE => array_merge(
                    array_keys( self::COUNT_LIMITS ),
                    array_keys( self::DEPRECATED_COUNT_TYPES )
                ),
                ParamValidator::PARAM_REQUIRED => true,
            ],
            'from' => [
                self::PARAM_SOURCE => 'query',
                ParamValidator::PARAM_TYPE => 'integer',
                ParamValidator::PARAM_REQUIRED => false
            ],
            'to' => [
                self::PARAM_SOURCE => 'query',
                ParamValidator::PARAM_TYPE => 'integer',
                ParamValidator::PARAM_REQUIRED => false
            ]
        ];
    }
}