Indexing pages below a backend user section leads to errors

[mschwemer]

If a standard page (or any other indexable page) is below a page of type backend user section this leads to indexing error. The error message in the TYPO3 log is

Uncaught TYPO3 Exception: #1517584045: Solr Document can not be prepared. The Reason: ID was not an accessible page

This is correct so far, as these contents may not be indexed, but also they should not occur as errors.

[wazum]

I wrote this event listener to solve this issue for our projects. Seems to do the trick.

<?php

declare(strict_types=1);

namespace Vendor\CustomSolr\Solr\EventListener;

use ApacheSolrForTypo3\Solr\Event\IndexQueue\AfterIndexQueueHasBeenInitializedEvent;
use Doctrine\DBAL\ArrayParameterType;
use Doctrine\DBAL\Exception;
use TYPO3\CMS\Core\Database\ConnectionPool;
use TYPO3\CMS\Core\Database\Query\QueryBuilder;
use TYPO3\CMS\Core\Utility\GeneralUtility;

final readonly class IndexQueueInitializationEventListener
{
    /**
     * @throws Exception
     */
    public function __invoke(AfterIndexQueueHasBeenInitializedEvent $event): void
    {
        if ($event->getIndexingConfigurationName() !== 'pages') {
            return;
        }

        $site = $event->getSite();
        $rootPageId = $site->getRootPageId();

        $this->removeBackendUserSectionSubpages($rootPageId);
    }

    /**
     * @throws Exception
     */
    private function removeBackendUserSectionSubpages(int $rootPageId): void
    {
        $backendUserSectionPageIds = $this->findBackendUserSectionPages($rootPageId);
        if (empty($backendUserSectionPageIds)) {
            return;
        }

        $subpageIds = [];
        foreach ($backendUserSectionPageIds as $pageId) {
            $subpageIds = array_merge($subpageIds, $this->findSubpagesOfPage($pageId));
        }

        if (empty($subpageIds)) {
            return;
        }

        $this->removeItemsFromIndexQueue($subpageIds);
    }

    /**
     * @throws Exception
     */
    private function findBackendUserSectionPages(int $rootPageId): array
    {
        /** @var QueryBuilder $queryBuilder */
        $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)
            ->getQueryBuilderForTable('pages');

        $result = $queryBuilder
            ->select('uid')
            ->from('pages')
            ->where(
                $queryBuilder->expr()->eq('doktype', $queryBuilder->createNamedParameter(6, \PDO::PARAM_INT)),
                $queryBuilder->expr()->eq('deleted', $queryBuilder->createNamedParameter(0, \PDO::PARAM_INT))
            )
            ->executeQuery()
            ->fetchAllAssociative();

        $potentialBackendUserSectionPageIds = array_map(static function(array $page) {
            return (int)$page['uid'];
        }, $result);

        $backendUserSectionPageIds = [];
        foreach ($potentialBackendUserSectionPageIds as $pageId) {
            $isInSite = $this->isPageInSiteRootline($pageId, $rootPageId);
            if ($isInSite) {
                $backendUserSectionPageIds[] = $pageId;
            }
        }

        return $backendUserSectionPageIds;
    }

    /**
     * @throws Exception
     */
    private function isPageInSiteRootline(int $pageId, int $rootPageId): bool
    {
        if ($pageId === $rootPageId) {
            return true;
        }

        $connection = GeneralUtility::makeInstance(ConnectionPool::class)
            ->getConnectionForTable('pages');

        $currentPageId = $pageId;
        $iterations = 0;
        while ($currentPageId > 0 && $iterations < 100) {
            $record = $connection->select(
                ['pid'],
                'pages',
                ['uid' => $currentPageId, 'deleted' => 0]
            )->fetchAssociative();

            if ($record === false) {
                return false;
            }

            $currentPageId = (int)$record['pid'];

            if ($currentPageId === $rootPageId) {
                return true;
            }

            $iterations++;
        }

        return false;
    }

    /**
     * @throws Exception
     */
    private function findSubpagesOfPage(int $pageId): array
    {
        $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)
            ->getQueryBuilderForTable('pages');

        $allSubpages = [];
        $pagesToProcess = [$pageId];
        $processedPages = [];

        while (!empty($pagesToProcess)) {
            $currentPageId = array_shift($pagesToProcess);

            if (in_array($currentPageId, $processedPages, true)) {
                continue;
            }

            $processedPages[] = $currentPageId;

            $result = $queryBuilder->select('uid')
                ->from('pages')
                ->where(
                    $queryBuilder->expr()->eq('pid', $queryBuilder->createNamedParameter($currentPageId, \PDO::PARAM_INT)),
                    $queryBuilder->expr()->eq('deleted', $queryBuilder->createNamedParameter(0, \PDO::PARAM_INT))
                )
                ->executeQuery()
                ->fetchAllAssociative();

            foreach ($result as $page) {
                $childPageId = (int)$page['uid'];
                $allSubpages[] = $childPageId;
                $pagesToProcess[] = $childPageId;
            }
        }

        return $allSubpages;
    }

    private function removeItemsFromIndexQueue(array $pageIds): void
    {
        if (empty($pageIds)) {
            return;
        }

        $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)
            ->getQueryBuilderForTable('tx_solr_indexqueue_item');

        $queryBuilder
            ->delete('tx_solr_indexqueue_item')
            ->where(
                $queryBuilder->expr()->eq('item_type', $queryBuilder->createNamedParameter('pages')),
                $queryBuilder->expr()->in(
                    'item_uid',
                    $queryBuilder->createNamedParameter($pageIds, ArrayParameterType::INTEGER)
                )
            )
            ->executeStatement();
    }
}

Services.yaml

  Vendor\CustomSolr\Solr\EventListener\IndexQueueInitializationEventListener:
    tags:
      - name: event.listener
        identifier: 'removeBackendUserSectionSubpages'
        event: ApacheSolrForTypo3\Solr\Event\IndexQueue\AfterIndexQueueHasBeenInitializedEvent