Bulk Import Script with pagination support. Allows massive import of entities from Content Hub.

ach-bulk-import.php

<?php
/**
 * @file
 * Add entities from Content Hub to the Import Queue.
 *
 * Please locate this field in the 'scripts' directory as a sibling of docroot:
 * <DOCROOT>/../scripts/ach-bulk-import.php
 *
 * To run the script, execute the drush command:
 * $drush scr ../scripts/ach-bulk-import.php
 *
 * Make sure to enable the Import Queue before executing this script.
 *
 * Notes:
 *
 * 1) If you want to explicitly avoid importing a particular entity type, please
 *    add it to the list of $global_excluded_types.
 * 2) By default importing includes all dependencies. To change this behavior
 *    change the variable $include_dependencies to FALSE.
 * 3) You can decided whether to publish entities after importing them. To
 *    publish entities after importing, set variable $publishing_status to 1.
 *    Setting $publishing_status to 0 imports them as unpublished.
 * 4) You can decide to use FIFO (first exported entities are imported first),
 *    or LIFO (last exported entities are imported first), according to the
 *    $fifo variable: $fifo = 1 uses FIFO, $fifo = 0 uses LIFO.
 * 5) You can set the author of the nodes to be imported locally. Example: If
 *    you set the $uid = 1, it will import all nodes as administrator (author
 *    is administrator). Change it to specific UID to use as author.
 */

use Drupal\acquia_contenthub\ContentHubEntityDependency;
use Drupal\Component\Serialization\Json;

// Global exclusion of entity types.
$global_excluded_types = [
  //   'redirect' => 'redirect',
];

// Include importing dependencies. By default it is "TRUE".
$include_dependencies = TRUE;

// Determine if we want to publish imported entities or not.
// 1: Publish entities, 0: Do not publish.
$publishing_status = 1;

// If TRUE, it will import from the last page to the first (FIFO: first entities
// exported will be the first to import), otherwise will use LIFO (Last exported
// entities will be imported first).
$fifo = TRUE;

// Determine the author UUID for the nodes to be created.
$uid = 1; // administrator.
$user = \Drupal\user\Entity\User::load($uid);
$author = $user->uuid();

/** @var \Drupal\acquia_contenthub\ContentHubEntitiesTracking $entities_tracking */
$entities_tracking = \Drupal::service('acquia_contenthub.acquia_contenthub_entities_tracking');

// Loading ClientManager to be able to execute requests to Content Hub and
// to check connection.
/** @var \Drupal\acquia_contenthub\Client\ClientManager $client_manager */
$client_manager = \Drupal::service('acquia_contenthub.client_manager');
$client = $client_manager->getConnection();

// The ImportEntityManager Service allows to import entities.
/** @var \Drupal\acquia_contenthub\ImportEntityManager $import_manager */
$import_manager = \Drupal::service("acquia_contenthub.import_entity_manager");

// List all the 'dependent' entities type IDs.
$dependent_entity_type_ids = ContentHubEntityDependency::getPostDependencyEntityTypes();

$excluded_types = array_merge($global_excluded_types, $dependent_entity_type_ids);

// Checks whether the import queue has been enabled.
$import_with_queue = \Drupal::config('acquia_contenthub.entity_config')->get('import_with_queue');
if (!$import_with_queue) {
  drush_user_abort('Please enable the Import Queue.');
}

// Check if the site is connected to Content Hub.
if (!$client_manager->isConnected()) {
  return;
}

$list = $client_manager->createRequest('listEntities', [[]]);
$total = floor($list['total'] / 1000) * 1000;

// Starting page.
$start = $fifo ? $total : 0;
// Step
$step = $fifo ? -1000 : 1000;

// Counter of queued entities.
$i = 0;
do {
  // List all entities you want to import by modifying the $options array.
  /*
  * Example of how to structure the $options parameter:
  *
  * $options = [
  *     'type'   => 'node',
  *     'origin' => '11111111-1111-1111-1111-111111111111',
  *     'filters' => [
  *         'status' => 1,
  *         'title' => 'New*',
  *         'body' => '/Boston/',
  *     ],
  * ];
  *
  */
  $options = [
    'start' => $start,
  ];
  $list = $client_manager->createRequest('listEntities', [$options]);
  foreach ($list['data'] as $entity) {
    $i++;
    // We do not want to import "dependent" entities.
    // These 3 lines are not needed in this example, but if we are listing all
    // entities, make sure to exclude dependent entities to be sent directly to
    // the importRemoteEntity() method because you would not be sure if their
    // host (parent) entity exist in the system yet.
    if (in_array($entity['type'], $excluded_types)) {
      drush_print("{$i}) Skipped entity type = {$entity['type']} , UUID = {$entity['uuid']} (Dependent or excluded entity type)");
      continue;
    }

    // Do not import the entity if it has been previously imported and has the
    // same "modified" flag, which means there are no new updates on the entity.
    if ($imported_entity = $entities_tracking->loadImportedByUuid($entity['uuid'])) {
      if ($imported_entity->getModified() === $entity['modified']) {
        drush_print("{$i}) Skipped entity type = {$entity['type']} , UUID = {$entity['uuid']} (Entity already imported)");
        continue;
      }
    }

    // Add entity to import queue.
    try {
      $response = $import_manager->addEntityToImportQueue($entity['uuid'], $include_dependencies, $author, $publishing_status);
      $status = Json::decode($response->getContent());
      if (!empty($status['status']) && $status['status'] == 200) {
        drush_print("{$i}) Entity added to import queue: type = {$entity['type']} , UUID = {$entity['uuid']}");
      }
      else {
        drush_print("{$i}) ERROR: Cannot add entity to import queue: type = {$entity['type']} , UUID = {$entity['uuid']}");
      }
    } catch (\Drupal\Core\Entity\EntityStorageException $ex) {
      drush_print("{$i}) ERROR: Failed to add entity to import queue: type = {$entity['type']} , UUID = {$entity['uuid']} [{$ex->getMessage()}]");
    }
  }
  $start = $start + $step;
  $exit_condition = $fifo ? $start >= 0 : $start <= $total;
} while ($exit_condition);