[BUGFIX] Ensure that DataHandlerHook will not add pages to queue that…

… does not exist (#635) Resolves #634
tomasnorre · Oct 17, 2020 · 7a7d347 · 7a7d347
1 parent 11798cb
commit 7a7d347
Show file tree

Hide file tree

Showing 3 changed files with 64 additions and 1 deletion.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -23,6 +23,7 @@
 * Frontend User initialization with UserGroups for crawling protected pages
 * Making sure PageUid added with ExcludeString is kept as integers
 * Instatiation of ProcessRepository and QueueRepository change to GeneralUtility::makeInstance
+* Ensure that DataHandlerHook will not add pages to queue that does not exist
 
 ## Crawler 9.1.0
 Crawler 9.1.0 was released on August 2nd, 2020

diff --git a/Classes/Hooks/DataHandlerHook.php b/Classes/Hooks/DataHandlerHook.php
@@ -23,6 +23,7 @@
 use AOE\Crawler\Domain\Repository\QueueRepository;
 use TYPO3\CMS\Core\Utility\GeneralUtility;
 use TYPO3\CMS\Extbase\Object\ObjectManager;
+use TYPO3\CMS\Frontend\Page\PageRepository;
 
 class DataHandlerHook
 {
@@ -34,7 +35,7 @@ public function addFlushedPagesToCrawlerQueue(array $parameters, \TYPO3\CMS\Core
         }
         foreach ($pageIdsToBeFlushedFromCache as $pageId) {
             $pageId = (int) $pageId;
-            if ($pageId < 1) {
+            if ($pageId < 1 || empty($this->getPageRepository()->getPage($pageId))) {
                 continue;
             }
             if ($this->getQueueRepository()->isPageInQueue($pageId)) {
@@ -53,4 +54,10 @@ private function getCrawlerApi(): CrawlerApi
     {
         return GeneralUtility::makeInstance(ObjectManager::class)->get(CrawlerApi::class);
     }
+
+    private function getPageRepository(): PageRepository
+    {
+        // Todo: Switch to TYPO3\CMS\Core\Repository\PageRepository when dropping support for TYPO3 9LTS
+        return GeneralUtility::makeInstance(ObjectManager::class)->get(PageRepository::class);
+    }
 }
diff --git a/Tests/Unit/Hooks/DataHandlerHookTest.php b/Tests/Unit/Hooks/DataHandlerHookTest.php
@@ -29,6 +29,7 @@
 use TYPO3\CMS\Core\DataHandling\DataHandler;
 use TYPO3\CMS\Core\Utility\GeneralUtility;
 use TYPO3\CMS\Extbase\Object\ObjectManager;
+use TYPO3\CMS\Frontend\Page\PageRepository;
 
 class DataHandlerHookTest extends UnitTestCase
 {
@@ -49,9 +50,14 @@ public function itShouldAddPageToQueue(): void
         $queueRepository->isPageInQueue(1)->willReturn(false);
         $queueRepository->isPageInQueue(2)->willReturn(true);
 
+        $pageRepository = $this->prophesize(PageRepository::class);
+        $pageRepository->getPage(1)->willReturn(['Faking that page exists as not empty array']);
+        $pageRepository->getPage(2)->willReturn(['Faking that page exists as not empty array']);
+
         $objectManager = $this->prophesize(ObjectManager::class);
         $objectManager->get(QueueRepository::class)->willReturn($queueRepository->reveal());
         $objectManager->get(CrawlerApi::class)->willReturn($crawlerApi->reveal());
+        $objectManager->get(PageRepository::class)->willReturn($pageRepository->reveal());
 
         $cacheManager = $this->prophesize(CacheManager::class);
         $cacheManager->getCache(Argument::any())->willReturn($this->prophesize(FrontendInterface::class)->reveal());
@@ -89,9 +95,15 @@ public function itShouldAddPageToQueueWithMorePages(): void
         $queueRepository->isPageInQueue(2)->willReturn(true);
         $queueRepository->isPageInQueue(3)->willReturn(false);
 
+        $pageRepository = $this->prophesize(PageRepository::class);
+        $pageRepository->getPage(1)->willReturn(['Faking that page exists as not empty array']);
+        $pageRepository->getPage(2)->willReturn(['Faking that page exists as not empty array']);
+        $pageRepository->getPage(3)->willReturn(['Faking that page exists as not empty array']);
+
         $objectManager = $this->prophesize(ObjectManager::class);
         $objectManager->get(QueueRepository::class)->willReturn($queueRepository->reveal());
         $objectManager->get(CrawlerApi::class)->willReturn($crawlerApi->reveal());
+        $objectManager->get(PageRepository::class)->willReturn($pageRepository->reveal());
 
         $cacheManager = $this->prophesize(CacheManager::class);
         $cacheManager->getCache(Argument::any())->willReturn($this->prophesize(FrontendInterface::class)->reveal());
@@ -109,4 +121,47 @@ public function itShouldAddPageToQueueWithMorePages(): void
             $dataHandler
         );
     }
+
+    /**
+     * Page with ID 1 is not in queue, should be added
+     * Page with ID 2 is already in queue. Should NOT be added.
+     * Page with ID 3 is not in queue, should be added
+     *
+     * @test
+     */
+    public function nothingToBeAddedAsPageDoNotExists(): void
+    {
+        $dataHandlerHook = new DataHandlerHook();
+        $crawlerApi = $this->prophesize(CrawlerApi::class);
+        $crawlerApi->addPageToQueue(1)->shouldBeCalled();
+
+        $queueRepository = $this->prophesize(QueueRepository::class);
+        $queueRepository->isPageInQueue(1)->willReturn(false);
+
+        $pageRepository = $this->prophesize(PageRepository::class);
+        $pageRepository->getPage(1)->willReturn(['Faking that page exists as not empty array']);
+        // Empty array to act like pages doesn't exist
+        $pageRepository->getPage(3000)->willReturn([]);
+
+        $objectManager = $this->prophesize(ObjectManager::class);
+        $objectManager->get(QueueRepository::class)->willReturn($queueRepository->reveal());
+        $objectManager->get(CrawlerApi::class)->willReturn($crawlerApi->reveal());
+        $objectManager->get(PageRepository::class)->willReturn($pageRepository->reveal());
+
+        $cacheManager = $this->prophesize(CacheManager::class);
+        $cacheManager->getCache(Argument::any())->willReturn($this->prophesize(FrontendInterface::class)->reveal());
+
+        GeneralUtility::setSingletonInstance(CacheManager::class, $cacheManager->reveal());
+        GeneralUtility::setSingletonInstance(ObjectManager::class, $objectManager->reveal());
+
+        $dataHandler = new DataHandler();
+
+        $dataHandlerHook->addFlushedPagesToCrawlerQueue(
+            [
+                'table' => 'tt_content',
+                'pageIdArray' => [0, 1, 3000],
+            ],
+            $dataHandler
+        );
+    }
 }