ContribsPager.php

Go to the documentation of this file.

<?php
use MediaWiki\Cache\LinkBatchFactory;
use MediaWiki\HookContainer\HookContainer;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\Linker\LinkRenderer;
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\RevisionStore;
use Wikimedia\IPUtils;
use Wikimedia\Rdbms\FakeResultWrapper;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\ILoadBalancer;
use Wikimedia\Rdbms\IResultWrapper;
 
class ContribsPager extends RangeChronologicalPager {
 
    private $messages;
 
    private $target;
 
    private $namespace;
 
    private $tagFilter;
 
    private $nsInvert;
 
    private $associated;
 
    private $deletedOnly;
 
    private $topOnly;
 
    private $newOnly;
 
    private $hideMinor;
 
    private $revisionsOnly;
 
    private $preventClickjacking = false;
 
    private $mParentLens;
 
    private $templateParser;
 
    private $linkBatchFactory;
 
    private $hookRunner;
 
    private $actorMigration;
 
    private $revisionStore;
 
    private $namespaceInfo;
 
    public function __construct(
        IContextSource $context,
        array $options,
        LinkRenderer $linkRenderer = null,
        LinkBatchFactory $linkBatchFactory = null,
        HookContainer $hookContainer = null,
        ILoadBalancer $loadBalancer = null,
        ActorMigration $actorMigration = null,
        RevisionStore $revisionStore = null,
        NamespaceInfo $namespaceInfo = null
    ) {
        // Class is used directly in extensions - T266484
        $services = MediaWikiServices::getInstance();
        $loadBalancer = $loadBalancer ?? $services->getDBLoadBalancer();
        // Set ->target before calling parent::__construct() so
        // parent can call $this->getIndexField() and get the right result. Set
        // the rest too just to keep things simple.
        $this->target = $options['target'] ?? '';
        $this->namespace = $options['namespace'] ?? '';
        $this->tagFilter = $options['tagfilter'] ?? false;
        $this->nsInvert = $options['nsInvert'] ?? false;
        $this->associated = $options['associated'] ?? false;
 
        $this->deletedOnly = !empty( $options['deletedOnly'] );
        $this->topOnly = !empty( $options['topOnly'] );
        $this->newOnly = !empty( $options['newOnly'] );
        $this->hideMinor = !empty( $options['hideMinor'] );
        $this->revisionsOnly = !empty( $options['revisionsOnly'] );
 
        // Most of this code will use the 'contributions' group DB, which can map to replica DBs
        // with extra user based indexes or partioning by user.
        // Set database before parent constructor to avoid setting it there with wfGetDB
        $this->mDb = $loadBalancer->getConnectionRef( ILoadBalancer::DB_REPLICA, 'contributions' );
        // Needed by call to getIndexField -> getTargetTable from parent constructor
        $this->actorMigration = $actorMigration ?? $services->getActorMigration();
        parent::__construct( $context, $linkRenderer ?? $services->getLinkRenderer() );
 
        $msgs = [
            'diff',
            'hist',
            'pipe-separator',
            'uctop'
        ];
 
        foreach ( $msgs as $msg ) {
            $this->messages[$msg] = $this->msg( $msg )->escaped();
        }
 
        // Date filtering: use timestamp if available
        $startTimestamp = '';
        $endTimestamp = '';
        if ( isset( $options['start'] ) && $options['start'] ) {
            $startTimestamp = $options['start'] . ' 00:00:00';
        }
        if ( isset( $options['end'] ) && $options['end'] ) {
            $endTimestamp = $options['end'] . ' 23:59:59';
        }
        $this->getDateRangeCond( $startTimestamp, $endTimestamp );
 
        $this->templateParser = new TemplateParser();
        $this->linkBatchFactory = $linkBatchFactory ?? $services->getLinkBatchFactory();
        $this->hookRunner = new HookRunner( $hookContainer ?? $services->getHookContainer() );
        $this->revisionStore = $revisionStore ?? $services->getRevisionStore();
        $this->namespaceInfo = $namespaceInfo ?? $services->getNamespaceInfo();
    }
 
    public function getDefaultQuery() {
        $query = parent::getDefaultQuery();
        $query['target'] = $this->target;
 
        return $query;
    }
 
    public function getNavigationBar() {
        return Html::rawElement( 'p', [ 'class' => 'mw-pager-navigation-bar' ],
            parent::getNavigationBar()
        );
    }
 
    public function reallyDoQuery( $offset, $limit, $order ) {
        list( $tables, $fields, $conds, $fname, $options, $join_conds ) = $this->buildQueryInfo(
            $offset,
            $limit,
            $order
        );
 
        /*
         * This hook will allow extensions to add in additional queries, so they can get their data
         * in My Contributions as well. Extensions should append their results to the $data array.
         *
         * Extension queries have to implement the navbar requirement as well. They should
         * - have a column aliased as $pager->getIndexField()
         * - have LIMIT set
         * - have a WHERE-clause that compares the $pager->getIndexField()-equivalent column to the offset
         * - have the ORDER BY specified based upon the details provided by the navbar
         *
         * See includes/Pager.php buildQueryInfo() method on how to build LIMIT, WHERE & ORDER BY
         *
         * &$data: an array of results of all contribs queries
         * $pager: the ContribsPager object hooked into
         * $offset: see phpdoc above
         * $limit: see phpdoc above
         * $descending: see phpdoc above
         */
        $dbr = $this->getDatabase();
        $data = [ $dbr->select(
            $tables, $fields, $conds, $fname, $options, $join_conds
        ) ];
        if ( !$this->revisionsOnly ) {
            $this->hookRunner->onContribsPager__reallyDoQuery(
                $data, $this, $offset, $limit, $order );
        }
 
        $result = [];
 
        // loop all results and collect them in an array
        foreach ( $data as $query ) {
            foreach ( $query as $i => $row ) {
                // If the query results are in descending order, the indexes must also be in descending order
                $index = $order === self::QUERY_ASCENDING ? $i : $limit - 1 - $i;
                // Left-pad with zeroes, because these values will be sorted as strings
                $index = str_pad( $index, strlen( $limit ), '0', STR_PAD_LEFT );
                // use index column as key, allowing us to easily sort in PHP
                $result[$row->{$this->getIndexField()} . "-$index"] = $row;
            }
        }
 
        // sort results
        if ( $order === self::QUERY_ASCENDING ) {
            ksort( $result );
        } else {
            krsort( $result );
        }
 
        // enforce limit
        $result = array_slice( $result, 0, $limit );
 
        // get rid of array keys
        $result = array_values( $result );
 
        return new FakeResultWrapper( $result );
    }
 
    private function getTargetTable() {
        $dbr = $this->getDatabase();
        $user = User::newFromName( $this->target, false );
        $ipRangeConds = $user->isAnon() ? $this->getIpRangeConds( $dbr, $this->target ) : null;
        if ( $ipRangeConds ) {
            return 'ip_changes';
        } else {
            $conds = $this->actorMigration->getWhere( $dbr, 'rev_user', $user );
            if ( isset( $conds['orconds']['actor'] ) ) {
                return 'revision_actor_temp';
            }
        }
 
        return 'revision';
    }
 
    public function getQueryInfo() {
        $revQuery = $this->revisionStore->getQueryInfo( [ 'page', 'user' ] );
        $queryInfo = [
            'tables' => $revQuery['tables'],
            'fields' => array_merge( $revQuery['fields'], [ 'page_is_new' ] ),
            'conds' => [],
            'options' => [],
            'join_conds' => $revQuery['joins'],
        ];
 
        // WARNING: Keep this in sync with getTargetTable()!
        $user = User::newFromName( $this->target, false );
        $dbr = $this->getDatabase();
        $ipRangeConds = $user->isAnon() ? $this->getIpRangeConds( $dbr, $this->target ) : null;
        if ( $ipRangeConds ) {
            // Put ip_changes first (T284419)
            array_unshift( $queryInfo['tables'], 'ip_changes' );
            $queryInfo['join_conds']['revision'] = [
                'JOIN', [ 'rev_id = ipc_rev_id' ]
            ];
            $queryInfo['conds'][] = $ipRangeConds;
        } else {
            // tables and joins are already handled by RevisionStore::getQueryInfo()
            $conds = $this->actorMigration->getWhere( $dbr, 'rev_user', $user );
            $queryInfo['conds'][] = $conds['conds'];
            // Force the appropriate index to avoid bad query plans (T189026)
            if ( isset( $conds['orconds']['actor'] ) ) {
                $queryInfo['options']['USE INDEX']['temp_rev_user'] = 'actor_timestamp';
            }
        }
 
        if ( $this->deletedOnly ) {
            $queryInfo['conds'][] = 'rev_deleted != 0';
        }
 
        if ( $this->topOnly ) {
            $queryInfo['conds'][] = 'rev_id = page_latest';
        }
 
        if ( $this->newOnly ) {
            $queryInfo['conds'][] = 'rev_parent_id = 0';
        }
 
        if ( $this->hideMinor ) {
            $queryInfo['conds'][] = 'rev_minor_edit = 0';
        }
 
        $queryInfo['conds'] = array_merge( $queryInfo['conds'], $this->getNamespaceCond() );
 
        // Paranoia: avoid brute force searches (T19342)
        if ( !$this->getAuthority()->isAllowed( 'deletedhistory' ) ) {
            $queryInfo['conds'][] = $dbr->bitAnd(
                'rev_deleted', RevisionRecord::DELETED_USER
                ) . ' = 0';
        } elseif ( !$this->getAuthority()->isAllowedAny( 'suppressrevision', 'viewsuppressed' ) ) {
            $queryInfo['conds'][] = $dbr->bitAnd(
                'rev_deleted', RevisionRecord::SUPPRESSED_USER
                ) . ' != ' . RevisionRecord::SUPPRESSED_USER;
        }
 
        // $this->getIndexField() must be in the result rows, as reallyDoQuery() tries to access it.
        $indexField = $this->getIndexField();
        if ( $indexField !== 'rev_timestamp' ) {
            $queryInfo['fields'][] = $indexField;
        }
 
        ChangeTags::modifyDisplayQuery(
            $queryInfo['tables'],
            $queryInfo['fields'],
            $queryInfo['conds'],
            $queryInfo['join_conds'],
            $queryInfo['options'],
            $this->tagFilter
        );
 
        $this->hookRunner->onContribsPager__getQueryInfo( $this, $queryInfo );
 
        return $queryInfo;
    }
 
    protected function getNamespaceCond() {
        if ( $this->namespace !== '' ) {
            $dbr = $this->getDatabase();
            $selectedNS = $dbr->addQuotes( $this->namespace );
            $eq_op = $this->nsInvert ? '!=' : '=';
            $bool_op = $this->nsInvert ? 'AND' : 'OR';
 
            if ( !$this->associated ) {
                return [ "page_namespace $eq_op $selectedNS" ];
            }
 
            $associatedNS = $dbr->addQuotes( $this->namespaceInfo->getAssociated( $this->namespace ) );
 
            return [
                "page_namespace $eq_op $selectedNS " .
                $bool_op .
                " page_namespace $eq_op $associatedNS"
            ];
        }
 
        return [];
    }
 
    private function getIpRangeConds( $db, $ip ) {
        // First make sure it is a valid range and they are not outside the CIDR limit
        if ( !$this->isQueryableRange( $ip ) ) {
            return false;
        }
 
        list( $start, $end ) = IPUtils::parseRange( $ip );
 
        return 'ipc_hex BETWEEN ' . $db->addQuotes( $start ) . ' AND ' . $db->addQuotes( $end );
    }
 
    public function isQueryableRange( $ipRange ) {
        $limits = $this->getConfig()->get( 'RangeContributionsCIDRLimit' );
 
        $bits = IPUtils::parseCIDR( $ipRange )[1];
        if (
            ( $bits === false ) ||
            ( IPUtils::isIPv4( $ipRange ) && $bits < $limits['IPv4'] ) ||
            ( IPUtils::isIPv6( $ipRange ) && $bits < $limits['IPv6'] )
        ) {
            return false;
        }
 
        return true;
    }
 
    public function getIndexField() {
        // The returned column is used for sorting and continuation, so we need to
        // make sure to use the right denormalized column depending on which table is
        // being targeted by the query to avoid bad query plans.
        // See T200259, T204669, T220991, and T221380.
        $target = $this->getTargetTable();
        switch ( $target ) {
            case 'revision':
                return 'rev_timestamp';
            case 'ip_changes':
                return 'ipc_rev_timestamp';
            case 'revision_actor_temp':
                return 'revactor_timestamp';
            default:
                wfWarn(
                    __METHOD__ . ": Unknown value '$target' from " . static::class . '::getTargetTable()', 0
                );
                return 'rev_timestamp';
        }
    }
 
    public function getTagFilter() {
        return $this->tagFilter;
    }
 
    public function getTarget() {
        return $this->target;
    }
 
    public function isNewOnly() {
        return $this->newOnly;
    }
 
    public function getNamespace() {
        return $this->namespace;
    }
 
    protected function getExtraSortFields() {
        // The returned columns are used for sorting, so we need to make sure
        // to use the right denormalized column depending on which table is
        // being targeted by the query to avoid bad query plans.
        // See T200259, T204669, T220991, and T221380.
        $target = $this->getTargetTable();
        switch ( $target ) {
            case 'revision':
                return [ 'rev_id' ];
            case 'ip_changes':
                return [ 'ipc_rev_id' ];
            case 'revision_actor_temp':
                return [ 'revactor_rev' ];
            default:
                wfWarn(
                    __METHOD__ . ": Unknown value '$target' from " . static::class . '::getTargetTable()', 0
                );
                return [ 'rev_id' ];
        }
    }
 
    protected function doBatchLookups() {
        # Do a link batch query
        $this->mResult->seek( 0 );
        $parentRevIds = [];
        $this->mParentLens = [];
        $batch = $this->linkBatchFactory->newLinkBatch();
        $isIpRange = $this->isQueryableRange( $this->target );
        # Give some pointers to make (last) links
        foreach ( $this->mResult as $row ) {
            if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
                $parentRevIds[] = $row->rev_parent_id;
            }
            if ( isset( $row->rev_id ) ) {
                $this->mParentLens[$row->rev_id] = $row->rev_len;
                if ( $isIpRange ) {
                    // If this is an IP range, batch the IP's talk page
                    $batch->add( NS_USER_TALK, $row->rev_user_text );
                }
                $batch->add( $row->page_namespace, $row->page_title );
            }
        }
        # Fetch rev_len for revisions not already scanned above
        $this->mParentLens += $this->revisionStore->getRevisionSizes(
            array_diff( $parentRevIds, array_keys( $this->mParentLens ) )
        );
        $batch->execute();
        $this->mResult->seek( 0 );
    }
 
    protected function getStartBody() {
        return "<ul class=\"mw-contributions-list\">\n";
    }
 
    protected function getEndBody() {
        return "</ul>\n";
    }
 
    public function tryCreatingRevisionRecord( $row, $title = null ) {
        /*
         * There may be more than just revision rows. To make sure that we'll only be processing
         * revisions here, let's _try_ to build a revision out of our row (without displaying
         * notices though) and then trying to grab data from the built object. If we succeed,
         * we're definitely dealing with revision data and we may proceed, if not, we'll leave it
         * to extensions to subscribe to the hook to parse the row.
         */
        Wikimedia\suppressWarnings();
        try {
            $revRecord = $this->revisionStore->newRevisionFromRow( $row, 0, $title );
            return $revRecord->getId() ? $revRecord : null;
        } catch ( Exception $e ) {
            return null;
        } finally {
            Wikimedia\restoreWarnings();
        }
    }
 
    public function formatRow( $row ) {
        $ret = '';
        $classes = [];
        $attribs = [];
 
        $linkRenderer = $this->getLinkRenderer();
 
        $page = null;
        // Create a title for the revision if possible
        // Rows from the hook may not include title information
        if ( isset( $row->page_namespace ) && isset( $row->page_title ) ) {
            $page = Title::newFromRow( $row );
        }
        $revRecord = $this->tryCreatingRevisionRecord( $row, $page );
        if ( $revRecord ) {
            $attribs['data-mw-revid'] = $revRecord->getId();
 
            $link = $linkRenderer->makeLink(
                $page,
                $page->getPrefixedText(),
                [ 'class' => 'mw-contributions-title' ],
                $page->isRedirect() ? [ 'redirect' => 'no' ] : []
            );
            # Mark current revisions
            $topmarktext = '';
            $user = $this->getUser();
 
            if ( $row->rev_id === $row->page_latest ) {
                $topmarktext .= '<span class="mw-uctop">' . $this->messages['uctop'] . '</span>';
                $classes[] = 'mw-contributions-current';
                # Add rollback link
                if ( !$row->page_is_new &&
                    $this->getAuthority()->probablyCan( 'rollback', $page ) &&
                    $this->getAuthority()->probablyCan( 'edit', $page )
                ) {
                    $this->preventClickjacking();
                    $topmarktext .= ' ' . Linker::generateRollback(
                        $revRecord,
                        $this->getContext(),
                        [ 'noBrackets' ]
                    );
                }
            }
            # Is there a visible previous revision?
            if ( $revRecord->getParentId() !== 0 &&
                $revRecord->userCan( RevisionRecord::DELETED_TEXT, $this->getAuthority() )
            ) {
                $difftext = $linkRenderer->makeKnownLink(
                    $page,
                    new HtmlArmor( $this->messages['diff'] ),
                    [ 'class' => 'mw-changeslist-diff' ],
                    [
                        'diff' => 'prev',
                        'oldid' => $row->rev_id
                    ]
                );
            } else {
                $difftext = $this->messages['diff'];
            }
            $histlink = $linkRenderer->makeKnownLink(
                $page,
                new HtmlArmor( $this->messages['hist'] ),
                [ 'class' => 'mw-changeslist-history' ],
                [ 'action' => 'history' ]
            );
 
            if ( $row->rev_parent_id === null ) {
                // For some reason rev_parent_id isn't populated for this row.
                // Its rumoured this is true on wikipedia for some revisions (T36922).
                // Next best thing is to have the total number of bytes.
                $chardiff = ' <span class="mw-changeslist-separator"></span> ';
                $chardiff .= Linker::formatRevisionSize( $row->rev_len );
                $chardiff .= ' <span class="mw-changeslist-separator"></span> ';
            } else {
                $parentLen = 0;
                if ( isset( $this->mParentLens[$row->rev_parent_id] ) ) {
                    $parentLen = $this->mParentLens[$row->rev_parent_id];
                }
 
                $chardiff = ' <span class="mw-changeslist-separator"></span> ';
                $chardiff .= ChangesList::showCharacterDifference(
                    $parentLen,
                    $row->rev_len,
                    $this->getContext()
                );
                $chardiff .= ' <span class="mw-changeslist-separator"></span> ';
            }
 
            $lang = $this->getLanguage();
            $comment = $lang->getDirMark() . Linker::revComment( $revRecord, false, true, false );
            $d = ChangesList::revDateLink( $revRecord, $user, $lang, $page );
 
            # When querying for an IP range, we want to always show user and user talk links.
            $userlink = '';
            $revUser = $revRecord->getUser();
            $revUserId = $revUser ? $revUser->getId() : 0;
            $revUserText = $revUser ? $revUser->getName() : '';
            if ( $this->isQueryableRange( $this->target ) ) {
                $userlink = ' <span class="mw-changeslist-separator"></span> '
                    . $lang->getDirMark()
                    . Linker::userLink( $revUserId, $revUserText );
                $userlink .= ' ' . $this->msg( 'parentheses' )->rawParams(
                    Linker::userTalkLink( $revUserId, $revUserText ) )->escaped() . ' ';
            }
 
            $flags = [];
            if ( $revRecord->getParentId() === 0 ) {
                $flags[] = ChangesList::flag( 'newpage' );
            }
 
            if ( $revRecord->isMinor() ) {
                $flags[] = ChangesList::flag( 'minor' );
            }
 
            $del = Linker::getRevDeleteLink( $user, $revRecord, $page );
            if ( $del !== '' ) {
                $del .= ' ';
            }
 
            // While it might be tempting to use a list here
            // this would result in clutter and slows down navigating the content
            // in assistive technology.
            // See https://phabricator.wikimedia.org/T205581#4734812
            $diffHistLinks = Html::rawElement( 'span',
                [ 'class' => 'mw-changeslist-links' ],
                // The spans are needed to ensure the dividing '|' elements are not
                // themselves styled as links.
                Html::rawElement( 'span', [], $difftext ) .
                ' ' . // Space needed for separating two words.
                Html::rawElement( 'span', [], $histlink )
            );
 
            # Tags, if any.
            list( $tagSummary, $newClasses ) = ChangeTags::formatSummaryRow(
                $row->ts_tags,
                'contributions',
                $this->getContext()
            );
            $classes = array_merge( $classes, $newClasses );
 
            $this->hookRunner->onSpecialContributions__formatRow__flags(
                $this->getContext(), $row, $flags );
 
            $templateParams = [
                'del' => $del,
                'timestamp' => $d,
                'diffHistLinks' => $diffHistLinks,
                'charDifference' => $chardiff,
                'flags' => $flags,
                'articleLink' => $link,
                'userlink' => $userlink,
                'logText' => $comment,
                'topmarktext' => $topmarktext,
                'tagSummary' => $tagSummary,
            ];
 
            # Denote if username is redacted for this edit
            if ( $revRecord->isDeleted( RevisionRecord::DELETED_USER ) ) {
                $templateParams['rev-deleted-user-contribs'] =
                    $this->msg( 'rev-deleted-user-contribs' )->escaped();
            }
 
            $ret = $this->templateParser->processTemplate(
                'SpecialContributionsLine',
                $templateParams
            );
        }
 
        // Let extensions add data
        $this->hookRunner->onContributionsLineEnding( $this, $ret, $row, $classes, $attribs );
        $attribs = array_filter( $attribs,
            [ Sanitizer::class, 'isReservedDataAttribute' ],
            ARRAY_FILTER_USE_KEY
        );
 
        // TODO: Handle exceptions in the catch block above.  Do any extensions rely on
        // receiving empty rows?
 
        if ( $classes === [] && $attribs === [] && $ret === '' ) {
            wfDebug( "Dropping Special:Contribution row that could not be formatted" );
            return "<!-- Could not format Special:Contribution row. -->\n";
        }
        $attribs['class'] = $classes;
 
        // FIXME: The signature of the ContributionsLineEnding hook makes it
        // very awkward to move this LI wrapper into the template.
        return Html::rawElement( 'li', $attribs, $ret ) . "\n";
    }
 
    protected function getSqlComment() {
        if ( $this->namespace || $this->deletedOnly ) {
            // potentially slow, see CR r58153
            return 'contributions page filtered for namespace or RevisionDeleted edits';
        } else {
            return 'contributions page unfiltered';
        }
    }
 
    protected function preventClickjacking() {
        $this->preventClickjacking = true;
    }
 
    public function getPreventClickjacking() {
        return $this->preventClickjacking;
    }
 
    public static function processDateFilter( array $opts ) {
        $start = $opts['start'] ?? '';
        $end = $opts['end'] ?? '';
        $year = $opts['year'] ?? '';
        $month = $opts['month'] ?? '';
 
        if ( $start !== '' && $end !== '' && $start > $end ) {
            $temp = $start;
            $start = $end;
            $end = $temp;
        }
 
        // If year/month legacy filtering options are set, convert them to display the new stamp
        if ( $year !== '' || $month !== '' ) {
            // Reuse getDateCond logic, but subtract a day because
            // the endpoints of our date range appear inclusive
            // but the internal end offsets are always exclusive
            $legacyTimestamp = ReverseChronologicalPager::getOffsetDate( $year, $month );
            $legacyDateTime = new DateTime( $legacyTimestamp->getTimestamp( TS_ISO_8601 ) );
            $legacyDateTime = $legacyDateTime->modify( '-1 day' );
 
            // Clear the new timestamp range options if used and
            // replace with the converted legacy timestamp
            $start = '';
            $end = $legacyDateTime->format( 'Y-m-d' );
        }
 
        $opts['start'] = $start;
        $opts['end'] = $end;
 
        return $opts;
    }
}