[BUGFIX] Fix indexing of access protected pages

Fixes indexing of access protected pages and/or page contents.

Resolves: TYPO3-Solr#3735

Build/Test/phpstan.neon

@@ -20,6 +20,5 @@ parameters:
     - '#^Variable \$_EXTKEY might not be defined\.#'
     - '#^Method .*\\QueryBuilder::buildSuggestQuery\(\) should return .*\\SuggestQuery but returns .*\\Search\\Query\\Query\|null.#'
     - '#^Method .*\\QueryBuilder::use.*FromTypoScript\(\) should return .*\\QueryBuilder but returns .*\\AbstractQueryBuilder.#'
-    - '#^Access to protected property .*\\PageRepository::\$where_groupAccess.#'
     - '#^Unreachable statement - code above always terminates.#'
     - '#^Parameter \#1 \$queryAlternative of method .*\\DisMax::setQueryAlternative\(\) expects string, null given.#'

Classes/EventListener/PageIndexer/FrontendGroupsModifier.php

@@ -40,9 +40,13 @@ class FrontendGroupsModifier
     public function __invoke(ModifyResolvedFrontendGroupsEvent $event): void
     {
         $pageIndexerRequest = $event->getRequest()->getAttribute('solr.pageIndexingInstructions');
-        if (!$pageIndexerRequest instanceof PageIndexerRequest) {
+        if (!$pageIndexerRequest instanceof PageIndexerRequest
+            || ((int)$pageIndexerRequest->getParameter('userGroup') === 0
+                && (int)$pageIndexerRequest->getParameter('pageUserGroup') < 1)
+        ) {
             return;
         }
+
         if (!$pageIndexerRequest->isAuthenticated()) {
             $logger = GeneralUtility::makeInstance(SolrLogManager::class, self::class);
             $logger->error(
@@ -58,7 +62,7 @@ public function __invoke(ModifyResolvedFrontendGroupsEvent $event): void
                         'error' => [
                             'code' => 403,
                             'message' => 'Invalid Index Queue Request.',
-                            ],
+                        ],
                     ],
                     403
                 ),
@@ -67,6 +71,9 @@ public function __invoke(ModifyResolvedFrontendGroupsEvent $event): void
         }
 
         $groups = $this->resolveFrontendUserGroups($pageIndexerRequest);
+        if ((int)$pageIndexerRequest->getParameter('pageUserGroup') > 0) {
+            $groups[] = (int)$pageIndexerRequest->getParameter('pageUserGroup');
+        }
         $groupData = [];
         foreach ($groups as $groupUid) {
             if (in_array($groupUid, [-2, -1])) {

Classes/IndexQueue/FrontendHelper/PageIndexer.php

@@ -54,6 +54,8 @@
  */
 class PageIndexer implements FrontendHelper, SingletonInterface
 {
+    protected bool $activated = false;
+
     /**
      * This frontend helper's executed action.
      */
@@ -92,6 +94,7 @@ class PageIndexer implements FrontendHelper, SingletonInterface
     public function activate(): void
     {
         $this->logger = GeneralUtility::makeInstance(SolrLogManager::class, __CLASS__);
+        $this->activated = true;
     }
 
     /**
@@ -141,7 +144,8 @@ protected function generatePageUrl(TypoScriptFrontendController $controller): st
     public function __invoke(AfterCacheableContentIsGeneratedEvent $event): void
     {
         $this->request = $event->getRequest()->getAttribute('solr.pageIndexingInstructions');
-        if (!$this->request) {
+
+        if (!$this->request || $this->activated === false) {
             return;
         }
         $this->setupConfiguration();
@@ -381,6 +385,7 @@ protected function getEventDispatcher(): EventDispatcherInterface
      */
     public function deactivate(PageIndexerResponse $response): void
     {
+        $this->activated = false;
         $response->addActionResult($this->action, $this->responseData);
     }
 }

Classes/IndexQueue/FrontendHelper/UserGroupDetector.php

@@ -124,17 +124,13 @@ public function deactivateTcaFrontendGroupEnableFields(
      * @param int $uid The page ID
      * @param bool $disableGroupAccessCheck If set, the check for group access is disabled. VERY rarely used
      * @param PageRepository $parentObject parent \TYPO3\CMS\Core\Domain\Repository\PageRepository object
-     *
-     * @noinspection PhpMissingReturnTypeInspection
      */
     public function getPage_preProcess(
         &$uid,
         &$disableGroupAccessCheck,
         PageRepository $parentObject
-    ) {
+    ): void {
         $disableGroupAccessCheck = true;
-        // @todo: Check if reset "where_groupAccess" really wanted. Most probably the core aspect 'frontend.user' must be used instead.
-        $parentObject->where_groupAccess = ''; // just to be on the safe side
     }
 
     /**
@@ -190,7 +186,7 @@ protected function findFrontendGroups(array $record, string $table): void
         if (isset($this->originalTca[$table]['ctrl']['enablecolumns']['fe_group'])) {
             $frontendGroups = $record[$this->originalTca[$table]['ctrl']['enablecolumns']['fe_group']] ?? null;
 
-            if (empty($frontendGroups)) {
+            if (empty($frontendGroups) || $frontendGroups === '-1') {
                 // default = public access
                 $frontendGroups = 0;
             } elseif ($this->request->getParameter('loggingEnabled')) {
@@ -222,7 +218,10 @@ protected function getFrontendGroups(): array
 
         // clean up: filter double groups
         $frontendGroups = array_unique($frontendGroups);
-        $frontendGroups = array_values($frontendGroups);
+        $frontendGroups = array_filter(
+            array_values($frontendGroups),
+            static fn (int $val): bool => ($val !== -1)
+        );
 
         if (empty($frontendGroups)) {
             // most likely an empty page with no content elements => public

Classes/IndexQueue/PageIndexer.php

@@ -59,16 +59,9 @@ public function index(Item $item): bool
             return false;
         }
 
-        $solrConnections = $this->getSolrConnectionsByItem($item);
-        foreach ($solrConnections as $systemLanguageUid => $solrConnection) {
+        $systemLanguageUids = array_keys($this->getSolrConnectionsByItem($item));
+        foreach ($systemLanguageUids as $systemLanguageUid) {
             $contentAccessGroups = $this->getAccessGroupsFromContent($item, $systemLanguageUid);
-
-            if (empty($contentAccessGroups)) {
-                // might be an empty page w/no content elements or some TYPO3 error / bug
-                // FIXME logging needed
-                continue;
-            }
-
             foreach ($contentAccessGroups as $userGroup) {
                 $this->indexPage($item, $systemLanguageUid, $userGroup);
             }
@@ -300,6 +293,12 @@ protected function indexPage(Item $item, ?int $language = 0, ?int $userGroup = 0
         $request->setIndexQueueItem($item);
         $request->addAction('indexPage');
         $request->setParameter('accessRootline', $accessRootline);
+        $request->setParameter('userGroup', $userGroup);
+
+        $feGroupColumn = $GLOBALS['TCA']['pages']['ctrl']['enablecolumns']['fe_group'] ?? '';
+        if (!empty($feGroupColumn)) {
+            $request->setParameter('pageUserGroup', $item->getRecord()[$feGroupColumn]);
+        }
 
         $indexRequestUrl = $this->getDataUrl($item, $language);
         $response = $request->send($indexRequestUrl);
@@ -330,7 +329,6 @@ protected function indexPage(Item $item, ?int $language = 0, ?int $userGroup = 0
 
         if (empty($indexActionResult['pageIndexed'])) {
             $message = 'Failed indexing page Index Queue item: ' . $item->getIndexQueueUid() . ' url: ' . $indexRequestUrl;
-
             throw new \RuntimeException($message, 1331837081);
         }
 

Configuration/RequestMiddlewares.php

@@ -5,7 +5,7 @@
     'frontend' => [
         'apache-solr-for-typo3/page-indexer-initialization' => [
             'target' => \ApacheSolrForTypo3\Solr\Middleware\PageIndexerInitialization::class,
-            'before' => ['typo3/cms-frontend/tsfe'],
+            'before' => ['typo3/cms-frontend/tsfe', 'typo3/cms-frontend/authentication'],
             'after' => ['typo3/cms-core/normalized-params-attribute'],
         ],
         'apache-solr-for-typo3/solr-route-enhancer' => [

Tests/Integration/IndexQueue/Fixtures/can_index_access_protected_page.csv

@@ -0,0 +1,12 @@
+"fe_groups",
+,"uid","pid","title"
+,1,1,"group 1"
+"pages",
+,"uid","pid","is_siteroot","doktype","slug","title","subtitle","crdate","tstamp","fe_group"
+,2,1,0,1,"/protected_page","protected page","",1449151778,1449151778,1
+"tt_content",
+,"uid","pid","CType","header","fe_group"
+,11,"2","header","public content of protected page",0
+"tx_solr_indexqueue_item",
+,"uid","root","item_type","item_uid","indexing_configuration","changed","indexed","has_indexing_properties","indexing_priority","indexed","errors"
+,4711,1,"pages",2,"pages",1449151778,0,0,0,0,0

Tests/Integration/IndexQueue/Fixtures/can_index_access_protected_page_with_protected_contents.csv

@@ -0,0 +1,17 @@
+"fe_groups",
+,"uid","pid","title"
+,1,1,"group 1"
+,2,1,"group 2"
+"pages",
+,"uid","pid","sys_language_uid","l10n_parent","is_siteroot","doktype","slug","title","subtitle","crdate","tstamp","fe_group",
+,2,1,0,0,0,1,"/protected_page","protected page","",1449151778,1449151778,1
+,3,1,1,2,0,1,"/protected_page_de","protected page de","",1449151778,1449151778,1
+"tt_content",
+,"uid","pid","sys_language_uid","l18n_parent","CType","header","fe_group","sorting"
+,11,"2",0,0,"header","public content of protected page",0,10
+,12,"2",1,11,"header","public content of protected page de",0,10
+,13,"2",0,0,"header","protected content of protected page",2,20
+,14,"2",1,13,"header","protected content of protected page de",2,20
+"tx_solr_indexqueue_item",
+,"uid","root","item_type","item_uid","indexing_configuration","changed","indexed","has_indexing_properties","indexing_priority","indexed","errors"
+,4711,1,"pages",2,"pages",1449151778,0,0,0,0,0

Tests/Integration/IndexQueue/Fixtures/can_index_page_with_protected_and_hideatlogin_content.csv

@@ -0,0 +1,13 @@
+"fe_groups",
+,"uid","pid","title"
+,1,1,"group 1"
+"pages",
+,"uid","pid","is_siteroot","doktype","slug","title","subtitle","crdate","tstamp","fe_group"
+,2,1,0,1,"/protected_page","protected page","",1449151778,1449151778,0
+"tt_content",
+,"uid","pid","CType","header","fe_group","sorting"
+,10,"2","header","protected ce",1,10
+,11,"2","header","hide at login content",-1,20
+"tx_solr_indexqueue_item",
+,"uid","root","item_type","item_uid","indexing_configuration","changed","indexed","has_indexing_properties","indexing_priority","indexed","errors"
+,4711,1,"pages",2,"pages",1449151778,0,0,0,0,0

Tests/Integration/IndexQueue/Fixtures/can_index_page_with_protected_content.csv

@@ -0,0 +1,13 @@
+"fe_groups",
+,"uid","pid","title"
+,1,1,"group 1"
+"pages",
+,"uid","pid","is_siteroot","doktype","slug","title","subtitle","crdate","tstamp","fe_group"
+,2,1,0,1,"/protected_page","protected page","",1449151778,1449151778,0
+"tt_content",
+,"uid","pid","CType","header","fe_group","sorting"
+,10,"2","header","protected ce",1,10
+,11,"2","header","public ce",0,20
+"tx_solr_indexqueue_item",
+,"uid","root","item_type","item_uid","indexing_configuration","changed","indexed","has_indexing_properties","indexing_priority","indexed","errors"
+,4711,1,"pages",2,"pages",1449151778,0,0,0,0,0

Tests/Integration/IndexQueue/FrontendHelper/PageIndexerTest.php

@@ -17,10 +17,8 @@
 
 namespace ApacheSolrForTypo3\Solr\Tests\Integration\IndexQueue\FrontendHelper;
 
-use ApacheSolrForTypo3\Solr\IndexQueue\Item;
 use ApacheSolrForTypo3\Solr\Tests\Integration\IntegrationTest;
 use Psr\Http\Message\ResponseInterface;
-use TYPO3\CMS\Core\Database\ConnectionPool;
 use TYPO3\CMS\Core\Exception\SiteNotFoundException;
 use TYPO3\CMS\Core\Utility\GeneralUtility;
 use TYPO3\CMS\Frontend\Page\CacheHashCalculator;
@@ -352,8 +350,7 @@ protected function indexQueuedPage(int $pageId = 1, string $siteLanguageBase = '
         $url = rtrim($domain, '/') . '/' . ltrim($siteLanguageBase, '/') . '?' . $queryString;
 
         // Now add the headers for item 4711 to the request
-        $connection = GeneralUtility::makeInstance(ConnectionPool::class)->getConnectionForTable('tx_solr_indexqueue_item');
-        $item = $connection->select(['*'], 'tx_solr_indexqueue_item', ['uid' => 4711])->fetchAssociative();
-        return $this->executePageIndexer($url, new Item($item));
+        $item = $this->getIndexQueueItem(4711);
+        return $this->executePageIndexer($url, $item);
     }
 }