Filter by title on the dataset import status dashboard

modules/datastore/src/Form/DashboardForm.php

@@ -2,14 +2,16 @@
 
 namespace Drupal\datastore\Form;
 
-use Drupal\common\DataResource;
-use Drupal\Core\Pager\PagerManagerInterface;
-use Drupal\Core\StringTranslation\StringTranslationTrait;
+use Drupal\Core\Datetime\DateFormatter;
+use Drupal\Core\Entity\EntityStorageInterface;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
 use Drupal\Core\Form\FormBase;
 use Drupal\Core\Form\FormStateInterface;
-use Drupal\common\DatasetInfo;
-use Drupal\Core\Datetime\DateFormatter;
+use Drupal\Core\Pager\PagerManagerInterface;
+use Drupal\Core\StringTranslation\StringTranslationTrait;
 use Drupal\Core\Url;
+use Drupal\common\DataResource;
+use Drupal\common\DatasetInfo;
 use Drupal\common\UrlHostTokenResolver;
 use Drupal\harvest\HarvestService;
 use Drupal\metastore\MetastoreService;
@@ -71,6 +73,13 @@ class DashboardForm extends FormBase {
    */
   protected PostImportResultFactory $postImportResultFactory;
 
+  /**
+   * Node storage service.
+   *
+   * @var \Drupal\Core\Entity\EntityStorageInterface
+   */
+  protected EntityStorageInterface $nodeStorage;
+
   /**
    * DashboardController constructor.
    *
@@ -86,20 +95,24 @@ class DashboardForm extends FormBase {
    *   Date formatter service.
    * @param \Drupal\datastore\PostImportResultFactory $postImportResultFactory
    *   The PostImportResultFactory service..
+   * @param \Drupal\Core\Entity\EntityTypeManagerInterface $entityTypeManager
+   *   Entity type manager service.
    */
   public function __construct(
     HarvestService $harvestService,
     DatasetInfo $datasetInfo,
     MetastoreService $metastoreService,
     PagerManagerInterface $pagerManager,
     DateFormatter $dateFormatter,
-    PostImportResultFactory $postImportResultFactory
+    PostImportResultFactory $postImportResultFactory,
+    EntityTypeManagerInterface $entityTypeManager,
   ) {
     $this->harvest = $harvestService;
     $this->datasetInfo = $datasetInfo;
     $this->metastore = $metastoreService;
     $this->pagerManager = $pagerManager;
     $this->dateFormatter = $dateFormatter;
+    $this->nodeStorage = $entityTypeManager->getStorage('node');
     $this->itemsPerPage = 10;
     $this->postImportResultFactory = $postImportResultFactory;
   }
@@ -115,6 +128,7 @@ public static function create(ContainerInterface $container): self {
       $container->get('pager.manager'),
       $container->get('date.formatter'),
       $container->get('dkan.datastore.post_import_result_factory'),
+      $container->get('entity_type.manager'),
     );
   }
 
@@ -190,6 +204,12 @@ protected function buildFilters(array $filters): array {
           '#title' => $this->t('Dataset ID'),
           '#default_value' => $filters['uuid'] ?? '',
         ],
+        'dataset_title' => [
+          '#type' => 'textfield',
+          '#weight' => 1,
+          '#title' => $this->t('Dataset Title'),
+          '#default_value' => $filters['dataset_title'] ?? '',
+        ],
         'harvest_id' => [
           '#type' => 'select',
           '#weight' => 1,
@@ -241,11 +261,20 @@ public function buildTable(array $datasets): array {
   /**
    * Retrieve list of UUIDs for datasets matching the given filters.
    *
+   * Filters over-ride each other, in this order of priority:
+   * - UUID
+   * - Title search
+   * - Harvest plan ID.
+   *
    * @param string[] $filters
-   *   Datasets filters.
+   *   Datasets filters. Keys determine the filter. Recognized keys:
+   *   - uuid - Dataset UUID.
+   *   - dataset_title - A CONTAINS search within the dataset title field.
+   *   - harvest_id - A harvest plan ID.
    *
    * @return string[]
-   *   Filtered list of dataset UUIDs.
+   *   Paged, filtered list of dataset UUIDs. If no filter was supplied, all
+   *   dataset UUIDs will be returned, paged.
    */
   protected function getDatasets(array $filters): array {
     $datasets = [];
@@ -255,16 +284,17 @@ protected function getDatasets(array $filters): array {
     if (isset($filters['uuid'])) {
       $datasets = [$filters['uuid']];
     }
+    // Is the user searching for a dataset title?
+    elseif (isset($filters['dataset_title'])) {
+      $results = $this->getDatasetsByTitle($filters);
+      $datasets = $this->pagedFilteredList($results);
+    }
     // If a value was supplied for the harvest ID filter, retrieve dataset UUIDs
-    // belonging to the specfied harvest.
+    // belonging to the specified harvest.
     elseif (isset($filters['harvest_id'])) {
       $harvestLoad = iterator_to_array($this->getHarvestLoadStatus($filters['harvest_id']));
       $datasets = array_keys($harvestLoad);
-      $total = count($datasets);
-      $currentPage = $this->pagerManager->createPager($total, $this->itemsPerPage)->getCurrentPage();
-
-      $chunks = array_chunk($datasets, $this->itemsPerPage) ?: [[]];
-      $datasets = $chunks[$currentPage];
+      $datasets = $this->pagedFilteredList($datasets);
     }
     // If no filter values were supplied, fetch from the list of all dataset
     // UUIDs.
@@ -282,6 +312,50 @@ protected function getDatasets(array $filters): array {
     return $datasets;
   }
 
+  /**
+   * Entity query for nodes containing the dataset title.
+   *
+   * @param string[] $filters
+   *   Datasets filters.
+   *
+   * @return string[]
+   *   Dataset UUIDs .
+   */
+  protected function getDatasetsByTitle(array $filters): array {
+    // Get the ids using an entity query, because our dataset title is in the
+    // node title field.
+    // @todo Unify different queries against Data nodes using a repository or
+    // the NodeData wrapper.
+    $results = $this->nodeStorage->getQuery()
+      ->accessCheck(FALSE)
+      ->condition('type', 'data')
+      ->condition('field_data_type', 'dataset')
+      ->condition('title', $filters['dataset_title'], 'CONTAINS')
+      ->execute();
+    foreach ($this->nodeStorage->loadMultiple($results) as $node) {
+      $datasets[] = $node->uuid();
+    }
+
+    return $datasets;
+  }
+
+  /**
+   * Paged, filtered list of dataset UUIDs.
+   *
+   * @param string[] $datasets
+   *   Dataset UUIDs.
+   *
+   * @return string[]
+   *   Paged, filtered list of dataset UUIDs.
+   */
+  protected function pagedFilteredList(array $datasets): array {
+    $total = count($datasets);
+    $currentPage = $this->pagerManager->createPager($total, $this->itemsPerPage)->getCurrentPage();
+    $chunks = array_chunk($datasets, $this->itemsPerPage) ?: [[]];
+
+    return $chunks[$currentPage];
+  }
+
   /**
    * Builds dataset rows array.
    *

modules/datastore/tests/src/Unit/Form/DashboardFormTest.php

@@ -28,6 +28,10 @@
 use Drupal\datastore\PostImportResult;
 use Drupal\datastore\PostImportResultFactory;
 use Drupal\common\DataResource;
+use Drupal\Core\Entity\EntityStorageInterface;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\Core\Entity\Query\QueryInterface;
+use Drupal\node\NodeInterface;
 
 /**
  * @group dkan
@@ -151,6 +155,72 @@ public function testBuildTableRowsWithHarvestIdFilter() {
     $this->assertEquals(NULL, $form['table']['#rows'][0][6]['data']['#error']);
   }
 
+  /**
+   * Test building the dashboard table with a Dataset Title filter.
+   */
+  public function testBuildTableRowsWithDatasetTitleFilter() {
+    $info = [
+      'uuid' => 'dataset-1',
+      'title' => 'Dataset 1',
+      'revision_id' => '2',
+      'moderation_state' => 'published',
+      'modified_date_metadata' => '2020-01-15',
+      'modified_date_dkan' => '2021-02-11',
+    ];
+    $distribution = [
+      'distribution_uuid' => 'dist-1',
+      'resource_id' => '9ad17d45894f823c6a8e4f6d32b9535f',
+      'resource_version' => '1679508886',
+      'fetcher_status' => 'done',
+      'fetcher_percent_done' => 100,
+      'importer_status' => 'done',
+      'importer_percent_done' => 100,
+      'importer_error' => '',
+      'source_path' => 'http://example.com/file.csv',
+    ];
+
+    $postImportInfo = [
+      'resource_version' => '1679508886',
+      'post_import_status' => 'done',
+      'post_import_error' => NULL,
+    ];
+
+    $connectionMock = $this->createMock(Connection::class);
+    $resourceMappermock = $this->createMock(ResourceMapper::class);
+    $dataResourceMock = $this->createMock(DataResource::class);
+    $postImportResultMock = $this->getMockBuilder(PostImportResult::class)
+      ->setConstructorArgs(['', '', NULL, $dataResourceMock, $connectionMock, $resourceMappermock])
+      ->onlyMethods(['retrieveJobStatus'])
+      ->getMock();
+
+    $postImportResultMock->method('retrieveJobStatus')->willReturn($postImportInfo);
+
+    $nodeMock = (new Chain($this))
+      ->add(NodeInterface::class, 'uuid', 'dataset-1')
+      ->getMock();
+
+    $container = $this->buildContainerChain()
+      ->add(EntityTypeManagerInterface::class, 'getStorage', EntityStorageInterface::class)
+      ->add(RequestStack::class, 'getCurrentRequest', new Request(['dataset_title' => 'Dataset 1']))
+      ->add(EntityStorageInterface::class, 'getQuery', QueryInterface::class)
+      ->add(EntityStorageInterface::class, 'loadMultiple', [$nodeMock])
+      ->add(QueryInterface::class, 'accessCheck', QueryInterface::class)
+      ->add(QueryInterface::class, 'condition', QueryInterface::class)
+      ->add(QueryInterface::class, 'execute', [$nodeMock])
+      ->add(DatasetInfo::class, 'gather', ['latest_revision' => $info + ['distributions' => [$distribution]]])
+      ->add(PostImportResultFactory::class, 'initializeFromDistribution', $postImportResultMock)
+      ->getMock();
+    \Drupal::setContainer($container);
+    $form = DashboardForm::create($container)->buildForm([], new FormState());
+
+    $this->assertEquals(1, count($form['table']['#rows']));
+    $this->assertEquals('dataset-1', $form['table']['#rows'][0][0]['data']['#uuid']);
+    $this->assertEquals('Dataset 1', $form['table']['#rows'][0][0]['data']['#title']);
+    $this->assertEquals('NEW', $form['table']['#rows'][0][2]['data']);
+    $this->assertEquals('done', $form['table']['#rows'][0][6]['data']['#status']);
+    $this->assertEquals(NULL, $form['table']['#rows'][0][6]['data']['#error']);
+  }
+
   /**
    * Test building the dashboard table with a UUID filter.
    */
@@ -429,6 +499,8 @@ private function buildContainerChain(): Chain {
       ->add('dkan.harvest.storage.harvest_run_repository', HarvestRunRepository::class)
       ->add('database', Connection::class)
       ->add('dkan.datastore.post_import_result_factory', PostImportResultFactory::class)
+      ->add('entity.storage.interface', EntityStorageInterface::class)
+      ->add('entity_type.manager', EntityTypeManagerInterface::class)
       ->index(0);
 
     $runStatus = [
@@ -459,6 +531,7 @@ private function buildContainerChain(): Chain {
       ->add(PathValidator::class, 'getUrlIfValidWithoutAccessCheck', NULL)
       ->add(StreamWrapperManager::class, 'getViaUri', PublicStream::class)
       ->add(PublicStream::class, 'getExternalUrl', 'http://example.com')
-      ->add(Pager::class, 'getCurrentPage', 0);
+      ->add(Pager::class, 'getCurrentPage', 0)
+      ->add(EntityTypeManagerInterface::class, 'getStorage', EntityStorageInterface::class);
   }
 }