initial commit

islandora_collection_harvest.info.yml

@@ -0,0 +1,7 @@
+name: Islandora Collection Harvest
+type: module
+description: Provides a form to select an Islandora collection and media use, build a ZIP, and download it via streams.
+core_version_requirement: ^10
+package: Islandora
+dependencies:
+  - drupal:islandora

islandora_collection_harvest.permissions.yml

@@ -0,0 +1,3 @@
+access collection harvest:
+  title: 'Access Collection Harvest form'
+  description: 'Allows a user to access the collection harvest form to download media as ZIP'

islandora_collection_harvest.routing.yml

@@ -0,0 +1,14 @@
+islandora_collection_harvest.harvest_form:
+  path: '/admin/islandora/collection-harvest'
+  defaults:
+    _form: '\Drupal\islandora_collection_harvest\Form\CollectionHarvestForm'
+    _title: 'Collection Harvest'
+  requirements:
+    _permission: 'access collection harvest'
+
+islandora_collection_harvest.download_zip:
+  path: '/islandora/collection-harvest/download/{filename}'
+  defaults:
+    _controller: '\Drupal\islandora_collection_harvest\Controller\HarvestDownloadController::download'
+  requirements:
+    _permission: 'access collection harvest'

src/Controller/HarvestDownloadController.php

@@ -0,0 +1,117 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Drupal\islandora_collection_harvest\Controller;
+
+use Drupal\Core\StreamWrapper\StreamWrapperManagerInterface;
+use Drupal\flysystem\FlysystemFactory;
+use Drupal\Core\File\FileSystemInterface;
+use Drupal\Core\TempStore\PrivateTempStoreFactory;
+use Symfony\Component\HttpFoundation\StreamedResponse;
+use Drupal\Core\Controller\ControllerBase;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+
+final class HarvestDownloadController extends ControllerBase {
+
+  protected FileSystemInterface $fileSystem;
+
+  protected PrivateTempStoreFactory $tempStoreFactory;
+
+  protected FlysystemFactory $flysystemFactory;
+
+  protected StreamWrapperManagerInterface $streamWrapperManager;
+
+  public function __construct(
+    FileSystemInterface $file_system,
+    PrivateTempStoreFactory $tempStoreFactory,
+    FlysystemFactory $flysystemFactory,
+    StreamWrapperManagerInterface $streamWrapperManager
+  ) {
+    $this->fileSystem = $file_system;
+    $this->tempStoreFactory = $tempStoreFactory;
+    $this->flysystemFactory = $flysystemFactory;
+    $this->streamWrapperManager = $streamWrapperManager;
+  }
+
+  public static function create(ContainerInterface $container): self {
+    return new self(
+      $container->get('file_system'),
+      $container->get('tempstore.private'),
+      $container->get('flysystem_factory'),
+      $container->get('stream_wrapper_manager')
+    );
+  }
+
+  /**
+   * Builds and downloads archive.
+   *
+   * @param string $filename
+   *
+   * @return \Symfony\Component\HttpFoundation\StreamedResponse
+   */
+  public function download(string $filename): StreamedResponse {
+    $temp_store = $this->tempStoreFactory->get('islandora_collection_harvest');
+    $media_entities = $temp_store->get($filename);
+    if (empty($media_entities)) {
+      throw new \Symfony\Component\HttpKernel\Exception\NotFoundHttpException('No media found.');
+    }
+
+    return new StreamedResponse(function() use ($media_entities) {
+      $zip = new \ZipArchive();
+      $tmpfile = tempnam(sys_get_temp_dir(), 'collection_zip_');
+      $zip->open($tmpfile, \ZipArchive::CREATE | \ZipArchive::OVERWRITE);
+
+      foreach ($media_entities as $media) {
+        $source_field = $media->getSource()
+          ->getConfiguration()['source_field'] ?? NULL;
+        if (!$source_field || !$media->hasField($source_field)) {
+          continue;
+        }
+        $file = $media->get($source_field)->entity;
+        if (!$file) {
+          continue;
+        }
+
+        $uri = $file->getFileUri();
+        $filename_in_zip = basename($uri);
+        $real_path = $this->fileSystem->realpath($uri);
+
+        if ($real_path && file_exists($real_path)) {
+          $zip->addFile($real_path, $filename_in_zip);
+        }
+        else {
+          // Flysystem v1 fallback
+          try {
+            $scheme = $this->streamWrapperManager->getScheme($uri);
+            $path = substr($uri, strlen($scheme) + 3);
+            $filesystem = $this->flysystemFactory->getFilesystem($scheme);
+
+            if ($filesystem && $filesystem->has($path)) {
+              $contents = $filesystem->read($path);
+              if ($contents !== FALSE) {
+                $tmp_fly = tempnam(sys_get_temp_dir(), 'zip_');
+                file_put_contents($tmp_fly, $contents);
+                $zip->addFile($tmp_fly, $filename_in_zip);
+              }
+            }
+          }
+          catch (\Exception $e) {
+            $this->logger('islandora_collection_harvest')->error(
+              'Error reading @uri from Flysystem: @msg',
+              ['@uri' => $uri, '@msg' => $e->getMessage()]
+            );
+          }
+        }
+      }
+
+      $zip->close();
+      readfile($tmpfile);
+      unlink($tmpfile);
+    }, 200, [
+      'Content-Type' => 'application/zip',
+      'Content-Disposition' => 'attachment; filename="' . $filename . '"',
+    ]);
+  }
+
+}

src/Form/CollectionHarvestForm.php

@@ -0,0 +1,173 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Drupal\islandora_collection_harvest\Form;
+
+use Drupal\Core\Form\FormBase;
+use Drupal\Core\Form\FormStateInterface;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\Core\File\FileSystemInterface;
+use Drupal\flysystem\FlysystemFactory;
+use Drupal\islandora\IslandoraUtils;
+use Drupal\Core\TempStore\PrivateTempStoreFactory;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+
+/**
+ * Form for selecting Collection and Media Use and preparing ZIP download.
+ */
+final class CollectionHarvestForm extends FormBase {
+
+  protected EntityTypeManagerInterface $entityTypeManager;
+
+  protected IslandoraUtils $utils;
+
+  protected FileSystemInterface $fileSystem;
+
+  protected FlysystemFactory $flysystemFactory;
+
+  protected PrivateTempStoreFactory $tempStoreFactory;
+
+  /**
+   * {@inheritdoc}
+   */
+  public function __construct(
+    EntityTypeManagerInterface $entity_type_manager,
+    IslandoraUtils $utils,
+    FileSystemInterface $file_system,
+    FlysystemFactory $flysystem_factory,
+    PrivateTempStoreFactory $temp_store_factory
+  ) {
+    $this->entityTypeManager = $entity_type_manager;
+    $this->utils = $utils;
+    $this->fileSystem = $file_system;
+    $this->flysystemFactory = $flysystem_factory;
+    $this->tempStoreFactory = $temp_store_factory;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container): self {
+    return new self(
+      $container->get('entity_type.manager'),
+      $container->get('islandora.utils'),
+      $container->get('file_system'),
+      $container->get('flysystem_factory'),
+      $container->get('tempstore.private')
+    );
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function getFormId(): string {
+    return 'islandora_collection_harvest_form';
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function buildForm(array $form, FormStateInterface $form_state): array {
+    $term = $this->utils->getTermForUri('http://purl.org/dc/dcmitype/Collection');
+    $collections = [];
+    if ($term) {
+      $query = $this->entityTypeManager->getStorage('node')->getQuery();
+      $query->condition('type', 'islandora_object');
+      $query->condition('field_model', $term->id(), 'IN');
+      $query->accessCheck(FALSE);
+      $nids = $query->execute();
+      $nodes = $this->entityTypeManager->getStorage('node')
+        ->loadMultiple($nids);
+      foreach ($nodes as $node) {
+        $collections[$node->id()] = $node->label();
+      }
+    }
+
+    $vid = 'islandora_media_use';
+    $terms = $this->entityTypeManager->getStorage('taxonomy_term')
+      ->loadTree($vid);
+    $media_use_options = [];
+    foreach ($terms as $term) {
+      $media_use_options[$term->tid] = $term->name;
+    }
+
+    $form['#attributes']['id'] = 'collection-harvest-form';
+    $form['collection'] = [
+      '#type' => 'select',
+      '#title' => $this->t('Collection'),
+      '#options' => $collections,
+      '#required' => TRUE,
+    ];
+
+    $form['media_use'] = [
+      '#type' => 'select',
+      '#title' => $this->t('Media Use'),
+      '#options' => $media_use_options,
+      '#required' => TRUE,
+    ];
+
+    $form['actions'] = [
+      '#type' => 'actions',
+      'submit' => [
+        '#type' => 'submit',
+        '#value' => $this->t('Build Download'),
+        '#button_type' => 'primary',
+      ],
+    ];
+    return $form;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function submitForm(array &$form, FormStateInterface $form_state): void {
+    $collection_id = $form_state->getValue('collection');
+    $media_use_tid = $form_state->getValue('media_use');
+
+    $term = $this->entityTypeManager->getStorage('taxonomy_term')
+      ->load($media_use_tid);
+    if (!$term) {
+      $this->messenger()->addError($this->t('Media use term not found.'));
+      return;
+    }
+
+    // Load nodes in collection
+    $query = $this->entityTypeManager->getStorage('node')->getQuery();
+    $query->condition('type', 'islandora_object');
+    $query->condition('field_member_of', $collection_id);
+    $query->accessCheck(FALSE);
+    $nids = $query->execute();
+
+    $media_entities = [];
+    foreach ($nids as $nid) {
+      $node = $this->entityTypeManager->getStorage('node')->load($nid);
+      $media = $this->utils->getMediaWithTerm($node, $term);
+      if ($media) {
+        $source_field = $media->getSource()
+          ->getConfiguration()['source_field'] ?? NULL;
+        if ($source_field && $media->hasField($source_field) && !$media->get($source_field)
+            ->isEmpty()) {
+          $media_entities[] = $media;
+        }
+      }
+    }
+
+    if (empty($media_entities)) {
+      $this->messenger()
+        ->addError($this->t('No media found for this collection.'));
+      return;
+    }
+
+    // Store in tempstore
+    $temp_store = $this->tempStoreFactory->get('islandora_collection_harvest');
+    $zip_filename = 'collection_' . time() . '.zip';
+    $temp_store->set($zip_filename, $media_entities);
+    $this->messenger()
+      ->addStatus($this->t('Your ZIP download has been initiated. Please wait for the download to start.'));
+
+    // Redirect to download route
+    $form_state->setRedirect('islandora_collection_harvest.download_zip', ['filename' => $zip_filename]);
+  }
+
+}