islandora_bagit.module

<?php

/**
 * @file
 * Module to create BagIt Bags from Islandora objects. Requires the library at
 * https://github.com/scholarslab/BagItPHP. Consult the README.txt for
 * instructions.
 */

/**
 * Implements hook_init().
 */
function islandora_bagit_init() {
  if (isset($_SESSION['islandora_object'])) {
    unset($_SESSION['islandora_object']);
  }
  drupal_register_shutdown_function('islandora_bagit_create_bag_on_shutdown');
}

/**
 * Implements hook_permission().
 */
function islandora_bagit_permission() {
  return array(
    'create Islandora Bags' => array(
      'title' => t('Create Islandora Bags'),
      'description' => t('Create Islandora Bags'),
    ),
    'administer Islandora BagIt' => array(
      'title' => t('Administer Islandora BagIt'),
      'description' => t('Administer Islandora BagIt'),
    ),
  );
}

/**
 * Custom menu access callback.
 *
 * @param object $islandora_object
 *   The Islandora object to create a Bag for.
 *
 * @return bool
 *   TRUE if user has access, FALSE if not.
 */
function islandora_bagit_access_check($islandora_object) {
  return islandora_object_access('create Islandora Bags', $islandora_object);

}

/**
 * Implements hook_menu().
 */
function islandora_bagit_menu() {
  $items = array();
  $items['admin/islandora/tools/bagit'] = array(
    'title' => 'Islandora BagIt',
    'description' => 'Configure Islandora BagIt.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('islandora_bagit_admin_settings'),
    'access arguments' => array('administer Islandora BagIt'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['islandora/object/%islandora_object/manage/bagit'] = array(
    'title' => 'Create Bag(s)',
    'page callback' => 'islandora_bagit_create_bag',
    'page arguments' => array(2),
    'access callback' => 'islandora_bagit_access_check',
    'access arguments' => array(2),
    'type' => MENU_LOCAL_TASK,
  );
  return $items;
}

/**
 * Admin settings form builder.
 */
function islandora_bagit_admin_settings() {
  $form['islandora_bagit_library_dir'] = array(
    '#title' => t('Location of the BagIt library'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_library_dir', 'BagItPHP'),
    '#description' => t("Directory where the Scholars' Lab BagIt for PHP library
      is installed, relative to sites/all/libraries. Do not use a leading or
      trailing slash."),
    '#maxlength' => 255,
    '#required' => TRUE,
  );
  $form['islandora_bagit_bag_tmp_dir'] = array(
    '#title' => t('Temporary directory for unserialized Bags'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_bag_tmp_dir', file_directory_temp()),
    '#description' => t("Filesystem directory where the unserialized Bag
      directories are written, named by PID. Needs to exist and to be
      writable by the web server. Do not include the trailing slash."),
    '#maxlength' => 255,
    '#required' => TRUE,
  );

  $form['islandora_bagit_delete_unserialized_bag'] = array(
    '#type' => 'checkbox',
    '#title' => t('Delete unserialized (zipped) Bags.'),
    '#default_value' => variable_get('islandora_bagit_delete_unserialized_bag', 1),
    '#description' => t('Check this option if you want to automatically delete
      unserialized Bags.'),
  );

  $bagit_output_dir = variable_get('file_public_path', conf_path() . '/files');
  $form['islandora_bagit_bag_output_dir'] = array(
    '#title' => t('Output directory for serialized Bags'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_bag_output_dir', $bagit_output_dir),
    '#description' => t("Filesystem directory where serialized (.tgz or .zip) Bags
      are written, named by PID. Needs to exist and to be writable by the web
      server. Do not include the trailing slash. If you want users to be able to
      download the Bags, this directory needs to be below your site's file system directory."),
    '#maxlength' => 255,
    '#required' => TRUE,
  );

  $form['islandora_bagit_create_on_ingest'] = array(
    '#type' => 'checkbox',
    '#title' => t('Create Bag on object ingest'),
    '#default_value' => variable_get('islandora_bagit_create_on_ingest', 0),
    '#description' => t('Check this option if you want to create a Bag for
      an object when it is ingested.'),
  );

  $form['islandora_bagit_create_on_modify'] = array(
    '#type' => 'checkbox',
    '#title' => t('Create Bag on object modification'),
    '#default_value' => variable_get('islandora_bagit_create_on_modify', 0),
    '#description' => t('Check this option if you want to create a Bag for
      an object when it or any of its datasteams are modified.'),
  );

  $form['islandora_bagit_provide_download_link'] = array(
    '#type' => 'checkbox',
    '#title' => t('Provide link to download the Bag'),
    '#default_value' => variable_get('islandora_bagit_provide_download_link', 1),
    '#description' => t("This link will work only if 'Output directory for
      serialized Bags' is accessible via the web."),
  );

  $form['islandora_bagit_multiple_bag_type'] = array(
    '#type' => 'select',
    '#title' => t('Collection batch type'),
    '#default_value' => variable_get('islandora_bagit_multiple_bag_type', 'object'),
    '#options' => array(
      'object' => t('one Bag per object'),
      'collection' => t('one Bag per collection'),
      'no_children' => t('the Collection object only; children will not be bagged'),
    ),
    '#description' => t('Choose whether collection-level batches create one Bag per object or one Bag per collection.'),
  );

  $form['islandora_bagit_serialize_collection_bags'] = array(
    '#type' => 'checkbox',
    '#title' => t('Serialize collection Bags.'),
    '#default_value' => variable_get('islandora_bagit_serialize_collection_bags', 0),
    '#description' => t('Check this option if you want to serialize (zip) one-Bag-per-collection Bags.'),
  );

  $description = t("Base name for the Bags. The object's PID is appended to this value.");
  $default_bag_name = variable_get('islandora_bagit_bag_name', 'Bag-');
  $form['islandora_bagit_bag_name'] = array(
    '#title' => t('Bag name'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => $default_bag_name,
    '#description' => $description,
    '#maxlength' => 255,
    '#required' => TRUE,
  );

  $form['islandora_bagit_show_messages'] = array(
    '#type' => 'checkbox',
    '#title' => t('Display messages'),
    '#default_value' => variable_get('islandora_bagit_show_messages', 1),
    '#description' => t('Check this option if you want to display messages
      associated with Bag creation to the end user.'),
  );

  $form['islandora_bagit_log_bag_creation'] = array(
    '#type' => 'checkbox',
    '#title' => t('Log the creation of Bags'),
    '#default_value' => variable_get('islandora_bagit_log_bag_creation', 1),
    '#description' => t('Check this option if you want to log the
      creation of Bags to the Drupal watchdog.'),
  );
  $form['islandora_bagit_compression_type'] = array(
    '#type' => 'select',
    '#title' => t('Compression type'),
    '#default_value' => variable_get('islandora_bagit_compression_type', 'tgz'),
    '#options' => array(
      'tgz' => t('tgz'),
      'zip' => t('zip'),
    ),
    '#description' => t('The compression used on serialized Bags.'),
  );

  $object_plugins = islandora_bagit_get_plugins('object');
  foreach ($object_plugins as $plugin) {
    $object_plugin_options[$plugin] = $plugin;
  }
  $form['islandora_bagit_object_plugins'] = array(
    '#title' => t('Object plugins'),
    '#type' => 'checkboxes',
    '#options' => $object_plugin_options,
    '#default_value' => variable_get('islandora_bagit_object_plugins', array(
      'plugin_object_ds_basic' => 'plugin_object_ds_basic',
    )),
    '#description' => t("Object-level plugins. Plugins are fired in their order in this list and all add files to the same Bag. You should choose at least one plugin."),
  );
  $collection_plugins = islandora_bagit_get_plugins('collection');
  foreach ($collection_plugins as $plugin) {
    $collection_plugin_options[$plugin] = $plugin;
  }
  $form['islandora_bagit_collection_plugin'] = array(
    '#title' => t('Collection plugins'),
    '#type' => 'select',
    '#options' => $collection_plugin_options,
    '#default_value' => variable_get('islandora_bagit_collection_plugin', ''),
  );

  $form['islandora_bagit_baginfo_settings'] = array(
    '#title' => t('Bag metadata'),
    '#type' => 'fieldset',
    '#collapsible' => TRUE,
    '#collapsed' => TRUE,
    '#description' => t("Optional metadata written to the Bag's bag-info.txt file."),
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_transferring_organization'] = array(
    '#title' => t('Organization transferring the content'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_transferring_organization', ''),
    '#description' => t("Leave blank to omit from bag-info.txt."),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_transferring_organization_address'] = array(
    '#title' => t('Mailing address of the organization'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_transferring_organization_address', ''),
    '#description' => t("Leave blank to omit from bag-info.txt."),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_contact_name'] = array(
    '#title' => t('Contact name'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_contact_name', ''),
    '#description' => t("Person at the source organization who is responsible
      for the content transfer. Leave blank to omit from bag-info.txt."),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_contact_phone'] = array(
    '#title' => t('Contact phone'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_contact_phone', ''),
    '#description' => t("International format telephone number of person or
      position responsible. Leave blank to omit from bag-info.txt."),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_contact_email'] = array(
    '#title' => t('Contact email'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_bagit_contact_email', ''),
    '#description' => t("Fully qualified email address of person or
      position responsible. Leave blank to omit from bag-info.txt."),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_profile_uri'] = array(
    '#type' => 'textfield',
    '#title' => t('BagIt profile URI'),
    '#default_value' => variable_get('islandora_bagit_profile_uri', ''),
    '#description' => t('See https://github.com/ruebot/bagit-profiles for more information. Note that Islandora BagIt does not validate Bags against profiles.'),
    '#maxlength' => 255,
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_bagging_date'] = array(
    '#type' => 'checkbox',
    '#title' => t('Bagging date'),
    '#default_value' => variable_get('islandora_bagit_bagging_date', 0),
    '#description' => t("Date (YYYY-MM-DD) that the content was prepared for
      delivery."),
  );
  $form['islandora_bagit_baginfo_settings']['islandora_bagit_payload_octetstream_sum'] = array(
    '#type' => 'checkbox',
    '#title' => t('Octetstream sum'),
    '#default_value' => variable_get('islandora_bagit_payload_octetstream_sum', FALSE),
    '#description' => t('The "octetstream sum" of the payload.'),
  );

  return system_settings_form($form);
}

/**
 * Creates the object-level Bag.
 *
 * Also acts as the 'controller' to pass program flow off to the appropriate
 * batch functions.
 *
 * @param object $islandora_object
 *   The Islandora object to create a Bag for.
 *
 * @return string|array
 *   Either an empty array, a blank string, or a string containing
 *   a link to 'Download the Bag.'
 */
function islandora_bagit_create_bag($islandora_object) {
  // First, check to see if the object is a collection, and if it is,
  // reroute to the relevant batch function.
  foreach ($islandora_object as $ds) {
    if ($ds->id == 'COLLECTION_POLICY') {
      if (variable_get('islandora_bagit_multiple_bag_type', 'object') == 'object') {
        islandora_bagit_create_object_bag_batch($islandora_object);
      }
      elseif (variable_get('islandora_bagit_multiple_bag_type', 'object') == 'collection') {
        islandora_bagit_create_collection_bag_batch($islandora_object);
      }
      elseif (variable_get('islandora_bagit_multiple_bag_type', 'object') == 'no_children') {
        break;
      }
      return array();
    }
  }

  // Sanitize the PID so it is usable in file paths.
  $pid = str_replace(array(':', '-'), '_', $islandora_object->id);

  // Save all the datastreams to a randomly named temporary directory so
  // they can be added to the Bag. We delete these files after creating the Bag.
  $random_string = substr(md5(rand()), 0, 7);
  $tmp_ds_directory = variable_get('islandora_bagit_bag_tmp_dir', file_directory_temp()) .
    '/islandora_bagit_tmp/' . $random_string;
  if (!file_exists($tmp_ds_directory)) {
    mkdir($tmp_ds_directory, 0777, TRUE);
  }

  // Load the BagItPHP library.
  $bagit_library_dir = variable_get('islandora_bagit_library_dir', 'BagItPHP');
  if ($bagit_library_path = libraries_get_path($bagit_library_dir)) {
    require_once $bagit_library_path . '/lib/bagit.php';
  }

  $bag_file_name = variable_get('islandora_bagit_bag_name', 'Bag-') . $pid;
  $bag_output_path = variable_get('islandora_bagit_bag_output_dir', '/tmp') .
    DIRECTORY_SEPARATOR . $bag_file_name;

  // Because the BagItPHP library does some things by default if the bag output
  // directory already exists (like read the fetch.txt file), we always need to
  // delete the directory if it exists.
  if (file_exists($bag_output_path)) {
    rrmdir($bag_output_path);
  }

  // A list of all the files added to the bag, to show the user and add to
  // the watchdog entries.
  $all_added_files = array();
  // Needed for islandora_bagit_get_octetstream_sum().
  $all_added_files_sources = array();

  // Get bag-info.txt metadata.
  $bag_info = islandora_bagit_create_baginfo();

  // Create a new bag.
  $bag = new BagIt($bag_output_path, TRUE, TRUE, FALSE, $bag_info);

  // Iterate through all the object plugins. Each plugin must return the
  // parameters required for addFile() (i.e., a list of file source and
  // destination paths) or FALSE if it doesn't want to create a file.
  $plugins = variable_get('islandora_bagit_object_plugins', array(''));
  $files_to_add = array();
  if (count($plugins)) {
    foreach ($plugins as $plugin => $enabled) {
      if ($enabled) {
        module_load_include('inc', 'islandora_bagit', 'plugins/' . $plugin);
        $plugin_init_function = 'islandora_bagit_' . $plugin . '_init';
        // Process the plugins.
        if ($files_to_add = $plugin_init_function($islandora_object, $tmp_ds_directory)) {
          foreach ($files_to_add as $file) {
            $bag->addFile($file['source'], $file['dest']);
            $all_added_files[] = $file['dest'];
            $all_added_files_sources[] = $file['source'];
          }
        }
      }
    }
    // Generate octetstream sum.
    if (variable_get('islandora_bagit_payload_octetstream_sum', 0)) {
      $sum = islandora_bagit_get_octetstream_sum($all_added_files_sources);
      $bag->setBagInfoData('Payload-Oxum', $sum);
    }
    $bag->update();
  }
  else {
    drupal_set_message(t('There are no files to add to the Bag.'), 'warning');
    watchdog('bagit', 'BagIt Bag not created for !object: plugins found no files.',
      array('!object' => $islandora_object->id));
    return '';
  }

  // Allow other modules to modify the Bag using
  // mymodule_islandora_bagit_alter($bag, $islandora_object).
  drupal_alter('islandora_bagit', $bag, $islandora_object);

  // Write out the serialized (i.e., compressed) Bag.
  $serialized_bag_path = variable_get('islandora_bagit_bag_output_dir', '/tmp') .
    DIRECTORY_SEPARATOR . $bag_file_name;
  $compression_type = variable_get('islandora_bagit_compression_type', 'tgz');

  if (file_exists($serialized_bag_path . '.' . $compression_type)) {
    unlink($serialized_bag_path . '.' . $compression_type);
  }

  $bag->package($serialized_bag_path, $compression_type);

  if (variable_get('islandora_bagit_delete_unserialized_bag', 1)) {
    rrmdir($bag_output_path);
  }

  // Delete the temp directory created by file create plugins, if it exists.
  $bag_tmp_dir = variable_get('islandora_bagit_bag_tmp_dir', file_directory_temp()) .
    DIRECTORY_SEPARATOR . $pid;
  if (file_exists($bag_tmp_dir)) {
    rrmdir($bag_tmp_dir);
  }

  // Clean up the temp directory where we downloaded the datastreams.
  if (file_exists($tmp_ds_directory)) {
    rrmdir($tmp_ds_directory);
  }

  $all_added_files = array_unique($all_added_files);
  $serialized_all_added_files = implode(', ', $all_added_files);
  if (variable_get('islandora_bagit_log_bag_creation', 1)) {
    watchdog('islandora_bagit', 'Bag created for PID !pid (!files).',
      array('!pid' => $islandora_object->id, '!files' => $serialized_all_added_files));
  }

  $serialized_bag_path .= '.' . $compression_type;

  if (variable_get('islandora_bagit_show_messages', 1)) {
    drupal_set_message(t("Bag created and saved at %path", array(
      '%path' => $serialized_bag_path,
      )));
    drupal_set_message(t("Files added: %files",
      array('%files' => $serialized_all_added_files)));
  }

  // Allow other modules to fire the post-Bag creation hook.
  $post_create_data = module_invoke_all('islandora_bagit_post_create', $islandora_object->id, $serialized_bag_path);

  if (variable_get('islandora_bagit_provide_download_link', 1)) {
    // file_build_uri() needs a relative path.
    if (variable_get('islandora_file_default_scheme') == 'private') {
      $drupal_files_path = variable_get('islandora_file_private_path');
    }
    else {
      $drupal_files_path = variable_get('islandora_file_public_path', conf_path() . '/files');
    }
    $relative_bag_path = preg_replace("#$drupal_files_path#", '', $serialized_bag_path);
    $download_path = file_create_url(file_build_uri($relative_bag_path));
    return l(t('Download the Bag'), $download_path);
  }
  else {
    return array();
  }
}

/**************************
 * Object batch functions *
 *************************/

/**
 * Creates a set of Bags, one per child object, via the Drupal Batch API.
 *
 * @param object $islandora_object
 *   The Islandora collection object to create the Bags for.
 */
function islandora_bagit_create_object_bag_batch($islandora_object) {
  $objects_to_include = islandora_bagit_get_batch_members($islandora_object);
  $objects_to_include[] = $islandora_object->id;
  $num_objects = count($objects_to_include);
  $operations = array();

  // Add earch object in the collection to the $operations array.
  foreach ($objects_to_include as $pid) {
    $operations[] = array(
      'islandora_bagit_process_object_batch',
      array($pid, $num_objects),
    );
  }

  $batch = array(
    'operations' => $operations,
    'title' => t('Creating Bags'),
    'init_message' => t('Initializing'),
    'error_message' => t('An error occurred'),
    'finished' => 'islandora_bagit_object_batch_finished',
  );

  batch_set($batch);
  // batch_process() is required if the batch is not initiated within a form
  // handler. Parameter is the URL to redirect the user to, in this case, do
  // not redirect.
  batch_process('');
}

/**
 * Process one object in the batch.
 *
 * @param string $pid
 *   The current object's PID.
 *
 * @param string $num_objects
 *   The number of objects in this batch.
 *
 * @param array $context
 *   The Batch API $context array.
 */
function islandora_bagit_process_object_batch($pid, $num_objects, &$context) {
  if (empty($context['sandbox'])) {
    $context['sandbox'] = array();
    $context['sandbox']['progress'] = 0;
    $context['sandbox']['current_node'] = 0;
    $context['sandbox']['max'] = $num_objects;
    $context['results'][] = $pid;
    $context['message'] = t('Creating Bag for object @pid', array('@pid' => $pid));
  }

  $islandora_object = islandora_object_load($pid);
  islandora_bagit_create_bag($islandora_object);
}

/**
 * Batch API 'finished' callback.
 *
 * @param bool $success
 *   TRUE if no fatal PHP errors were detected.
 *
 * @param array $results
 *   List of all the object PIDs processed.
 *
 * @param array $operations
 *   List of all the operations processed.
 */
function islandora_bagit_object_batch_finished($success, $results, $operations) {
  if ($success) {
    drupal_set_message(t('@count Bags processed.', array('@count' => count($results))));
  }
  else {
    $error_operation = reset($operations);
    drupal_set_message(t('An error occurred while processing @operation with arguments : @args',
      array(
        '@operation' => $error_operation[0],
        '@args' => print_r($error_operation[0], TRUE),
      )));
  }
}

/******************************
 * Collection batch functions *
 * ***************************/

/**
 * Creates a single Bag containing all objects in a collection.
 *
 * Uses the Drupal Batch API.
 *
 * @param object $islandora_object
 *   The Islandora collection object to create the Bag for.
 */
function islandora_bagit_create_collection_bag_batch($islandora_object) {
  // Sanitize the PID so it is usable in file paths.
  $pid = str_replace(array(':', '-'), '_', $islandora_object->id);

  // Save all the datastreams to a randomly named temporary directory so
  // they can be added to the Bag. We delete these files after creating the Bag.
  $random_string = substr(md5(rand()), 0, 7);
  $tmp_ds_directory = variable_get('islandora_bagit_bag_tmp_dir', file_directory_temp()) .
    '/islandora_bagit_tmp/' . $random_string;
  if (!file_exists($tmp_ds_directory)) {
    mkdir($tmp_ds_directory, 0777, TRUE);
  }

  // Load the BagItPHP library.
  $bagit_library_dir = variable_get('islandora_bagit_library_dir', 'BagItPHP');
  if ($bagit_library_path = libraries_get_path($bagit_library_dir)) {
    require_once $bagit_library_path . '/lib/bagit.php';
  }

  $bag_file_name = variable_get('islandora_bagit_bag_name', 'Bag-') . $pid;
  $bag_output_path = variable_get('islandora_bagit_bag_output_dir', '/tmp') .
    DIRECTORY_SEPARATOR . $bag_file_name;

  // Because the BagItPHP library does some things by default if the bag output
  // directory already exists (like read the fetch.txt file), we always need to
  // delete the directory if it exists.
  if (file_exists($bag_output_path)) {
    rrmdir($bag_output_path);
  }

  // A list of all the files added to the bag, to show the user and add to
  // the watchdog entries. This is populated in
  // islandora_bagit_process_collection_batch().
  $_SESSION['islandora_bagit_all_added_files'] = array();

  if (variable_get('islandora_bagit_payload_octetstream_sum', FALSE)) {
    // Variables to hold the number of files in the batch bag and their
    // summed size.
    $_SESSION['islandora_bagit_batch_octetstream_sum_count'] = '0';
    $_SESSION['islandora_bagit_batch_octetstream_sum_size'] = '0';
  }

  // Get list of all objects to include in this Bag.
  $objects_to_include = islandora_bagit_get_batch_members($islandora_object);
  $objects_to_include[] = $islandora_object->id;
  $num_objects = count($objects_to_include);
  $operations = array();

  $collection_plugins = variable_get('islandora_bagit_collection_plugins', array(''));
  $object_plugins = variable_get('islandora_bagit_object_plugins', array(''));

  // Get bag-info.txt metadata.
  $bag_info = islandora_bagit_create_baginfo();

  // Create a new bag.
  $bag = new BagIt($bag_output_path, TRUE, TRUE, FALSE, $bag_info);
  islandora_bagit_serialize_bag_object($bag, $pid, $tmp_ds_directory);
  // We need the $pid in islandora_bagit_collection_batch_finish_bag().
  $_SESSION['islandora_bagit_collection_object_pid'] = $pid;

  // Iterate through all the objects and create an entry in $operations
  // for each one.
  foreach ($objects_to_include as $object_to_include_pid) {
    if (count($collection_plugins) && count($object_plugins)) {
      $operations[] = array('islandora_bagit_process_collection_batch',
        array($object_to_include_pid, $num_objects, $tmp_ds_directory));
    }
    else {
      drupal_set_message(t('There are no files to add to the Bag.'), 'warning');
      watchdog('bagit', 'BagIt Bag not created for !object: plugins found no files.',
        array('!object' => $islandora_object->id));
      return '';
    }
  }

  if (count($operations)) {
    $batch = array(
      'operations' => $operations,
      'title' => t('Creating Bag'),
      'init_message' => t('Initializing'),
      'error_message' => t('An error occurred'),
      'finished' => 'islandora_bagit_collection_batch_finished',
    );

    // It would be awesome if we could use $batch['progressive'] = FALSE
    // but that is pointless (see https://drupal.org/node/638712).
    batch_set($batch);

    // batch_process() is required if the batch is not initiated within a form
    // handler. Paramter is the URL to redirect the user to, in this case, do
    // not redirect.
    batch_process('');
  }
  else {
    return '';
  }
}

/**
 * Process one object in the collection batch.
 *
 * @param string $pid
 *   The current object's PID.
 *
 * @param string $num_objects
 *   The number of objects in this batch.
 *
 * @param array $context
 *   The Batch API $context array.
 */
function islandora_bagit_process_collection_batch($pid, $num_objects, $tmp_ds_directory, &$context) {
  // Load the BagItPHP library to avoid 'incomplete object' errors.
  $bagit_library_dir = variable_get('islandora_bagit_library_dir', 'BagItPHP');
  if ($bagit_library_path = libraries_get_path($bagit_library_dir)) {
    require_once $bagit_library_path . '/lib/bagit.php';
  }

  $bag = islandora_bagit_unserialize_bag_object();

  if (empty($context['sandbox'])) {
    $context['sandbox'] = array();
    $context['sandbox']['progress'] = 0;
    $context['sandbox']['current_node'] = 0;
    $context['sandbox']['max'] = $num_objects;
    $context['results'][] = $pid;
    $context['message'] = t('Adding object @pid to Bag', array('@pid' => $pid));
  }

  $islandora_object = islandora_object_load($pid);

  // Iterate through all the object plugins. Each plugin must return the
  // parameters required for addFile() (i.e., a list of file source and
  // destination paths) or FALSE if it doesn't want to create a file.
  $collection_plugin = variable_get('islandora_bagit_collection_plugin', '');
  $object_plugins = variable_get('islandora_bagit_object_plugins', array(''));
  // Since $bag->addFile() is being called after we loop through all the object
  // plugins, we need an array where we can accumulate the returned arrays from
  // each plugin.
  $all_files_to_add = array();
  foreach ($object_plugins as $object_plugin => $enabled) {
    if ($enabled) {
      module_load_include('inc', 'islandora_bagit', 'plugins/' . $object_plugin);
      $object_plugin_init_function = 'islandora_bagit_' . $object_plugin . '_init';
      // Process the object plugins.
      $files_to_add = $object_plugin_init_function($islandora_object, $tmp_ds_directory);
    }
    if (count($files_to_add)) {
      foreach ($files_to_add as $file_to_add) {
        if (!in_array($file_to_add, $all_files_to_add)) {
          $all_files_to_add[] = $file_to_add;
        }
      }
    }
  }

  if (count($all_files_to_add)) {
    // Pass $all_files_to_add from the object plugins to the collection plugin.
    module_load_include('inc', 'islandora_bagit', 'plugins/' . $collection_plugin);
    $collection_plugin_init_function = 'islandora_bagit_' . $collection_plugin . '_init';
    $all_files_to_add = $collection_plugin_init_function($islandora_object, $all_files_to_add, $tmp_ds_directory);

    // Now that the collection plugin has been fired, update the Bag.
    $all_added_files = $_SESSION['islandora_bagit_all_added_files'];
    foreach ($all_files_to_add as $file) {
      $bag->addFile($file['source'], $file['dest']);
      $all_added_files[] = $file['dest'];

      // Update the octetstream values for this batch Bag.
      if (variable_get('islandora_bagit_payload_octetstream_sum', FALSE)) {
        $_SESSION['islandora_bagit_batch_octetstream_sum_count']++;
        $_SESSION['islandora_bagit_batch_octetstream_sum_size'] = $_SESSION['islandora_bagit_batch_octetstream_sum_size'] +
            filesize($file['source']);
      }
    }
    $bag->update();
    $_SESSION['islandora_bagit_all_added_files'] = array_unique($all_added_files);

    // Serialize the Bag object for later use.
    islandora_bagit_serialize_bag_object($bag, $pid, $tmp_ds_directory);
  }
}

/**
 * Batch API 'finished' callback.
 *
 * @param bool $success
 *   TRUE if no fatal PHP errors were detected.
 *
 * @param array $results
 *   List of all the object PIDs processed.
 *
 * @param array $operations
 *   List of all the operations processed.
 */
function islandora_bagit_collection_batch_finished($success, $results, $operations) {
  if ($success) {
    islandora_bagit_collection_batch_finish_bag();
    // Clean up session variables added by the collection-level batch.
    unset($_SESSION['islandora_bagit_tmp_ds_directory']);
    unset($_SESSION['islandora_bagit_collection_object_pid']);
    unset($_SESSION['islandora_bagit_all_added_files']);
    drupal_set_message(t('@count objects added to Bag.', array('@count' => count($results))));
  }
  else {
    $error_operation = reset($operations);
    drupal_set_message(t('An error occurred while processing @operation with arguments : @args',
      array(
        '@operation' => $error_operation[0],
        '@args' => print_r($error_operation[0], TRUE),
      )));
  }
}

/**
 * Placeholder line to meet Coder style checks.
 *
 * Performs the tasks that follow batch_process() in
 * islandora_bagit_create_collection_bag_batch().
 *
 * Since batch_process() performs a redirect, any code that follows it
 * is not run. Setting $batch['progressive'] to FALSE is supposed to
 * avoid the redirect but is broken; see issue at
 * https://drupal.org/node/638712. The  only way to get the code
 * in this function to fire is to call it in the batch's'finished'
 * callback.
 *
 * return @array
 */
function islandora_bagit_collection_batch_finish_bag() {
  $pid = $_SESSION['islandora_bagit_collection_object_pid'];
  $tmp_ds_directory = $_SESSION['islandora_bagit_tmp_ds_directory'];
  $all_added_files = $_SESSION['islandora_bagit_all_added_files'];

  $bag_file_name = variable_get('islandora_bagit_bag_name', 'Bag-') . $pid;
  $bag_output_path = variable_get('islandora_bagit_bag_output_dir', '/tmp') .
    DIRECTORY_SEPARATOR . $bag_file_name;

  // Load the BagItPHP library to avoid 'incomplete object' errors.
  $bagit_library_dir = variable_get('islandora_bagit_library_dir', 'BagItPHP');
  if ($bagit_library_path = libraries_get_path($bagit_library_dir)) {
    require_once $bagit_library_path . '/lib/bagit.php';
  }

  // Now that we have completed the batch and added all objects, get the
  // Bag from the session so we can alter it, serialize it, et.
  $bag = islandora_bagit_unserialize_bag_object();

  // Add octetstream sum and clean up the related session variables.
  if (variable_get('islandora_bagit_payload_octetstream_sum', FALSE)) {
    // Variables to hold the number of files in the batch bag and their
    // summed size.
    $octetstream_num_files = $_SESSION['islandora_bagit_batch_octetstream_sum_count'];
    $octetstream_total_size = $_SESSION['islandora_bagit_batch_octetstream_sum_size'];
    $bag->setBagInfoData('Payload-Oxum', $octetstream_total_size .
      '.' . $octetstream_num_files);
    $bag->update();
    unset($_SESSION['islandora_bagit_batch_octetstream_sum_count']);
    unset($_SESSION['islandora_bagit_batch_octetstream_sum_size']);
  }

  // Allow other modules to modify the Bag using
  // mymodule_islandora_bagit_alter($bag, $islandora_object).
  drupal_alter('islandora_bagit', $bag, $islandora_object);

  if (variable_get('islandora_bagit_serialize_collection_bags', 0)) {
    // Write out the serialized (i.e., compressed) Bag.
    $serialized_bag_path = variable_get('islandora_bagit_bag_output_dir', '/tmp') .
      DIRECTORY_SEPARATOR . $bag_file_name;
    $compression_type = variable_get('islandora_bagit_compression_type', 'tgz');
    $bag->package($serialized_bag_path, $compression_type);

    if (variable_get('islandora_bagit_show_messages', 1)) {
      $serialized_bag_path .= '.' . $compression_type;
      drupal_set_message(t('Serialized collection Bag created and saved at <a href="/@path">@path</a>', array(
        '@path' => $serialized_bag_path,
      )));
    }

    if (variable_get('islandora_bagit_delete_unserialized_bag', 1)) {
      rrmdir($bag_output_path);
    }
  }
  else {
    drupal_set_message(t("Collection Bag created at %path", array(
      '%path' => $bag_output_path)));
  }

  // Delete the temp directory created by file create plugins, if it exists.
  $bag_tmp_dir = variable_get('islandora_bagit_bag_tmp_dir', file_directory_temp()) .
    DIRECTORY_SEPARATOR . $pid;
  if (file_exists($bag_tmp_dir)) {
    rrmdir($bag_tmp_dir);
  }

  // Clean up the temp director where we downloaded the datastreams.
  if (file_exists($tmp_ds_directory)) {
    rrmdir($tmp_ds_directory);
  }

  $serialized_all_added_files = implode(', ', $all_added_files);
  if (variable_get('islandora_bagit_log_bag_creation', 1)) {
    watchdog('islandora_bagit', 'Bag created for PID !pid (!files).',
      array('!pid' => $pid, '!files' => $serialized_all_added_files));
  }

  if (variable_get('islandora_bagit_show_messages', 1)) {
    drupal_set_message(t("Files added: %files",
      array('%files' => $serialized_all_added_files)));
  }

  // Allow other modules to fire the post-Bag creation hook.
  if (variable_get('islandora_bagit_serialize_collection_bags', 0)) {
    $post_create_data = module_invoke_all('islandora_bagit_post_create', $pid, $serialized_bag_path);
  }
  else {
    $post_create_data = module_invoke_all('islandora_bagit_post_create', $pid, $bag_output_path);
  }

  if (variable_get('islandora_bagit_provide_download_link', 1) && variable_get('islandora_bagit_serialize_collection_bags', 0)) {
    // file_build_uri() needs a relative path.
    if (variable_get('islandora_file_default_scheme') == 'private') {
      $drupal_files_path = variable_get('islandora_file_private_path');
    }
    else {
      $drupal_files_path = variable_get('islandora_file_public_path', conf_path() . '/files');
    }
    $relative_bag_path = preg_replace("#$drupal_files_path#", '', $serialized_bag_path);
    $download_path = file_create_url(file_build_uri($relative_bag_path));
    return l(t('Download the Bag'), $download_path);
  }
  else {
    return array();
  }
}

/*****************************************************************
 * Islandora hooks for detecting object and datastream ingestion *
 * and modification.                                             *
 * **************************************************************/

/**
 * Implements hook_islandora_object_ingested().
 *
 * Creates a Bag when an object is ingested.
 */
function islandora_bagit_islandora_object_ingested(FedoraObject $islandora_object) {
  if (islandora_bagit_access_check($islandora_object)) {
    if (variable_get('islandora_bagit_create_on_ingest', 0)) {
      // Don't create a Bag if we are dealing with a collection.
      if (!isset($islandora_object["COLLECTION_POLICY"])) {
        islandora_bagit_create_bag($islandora_object);
      }
    }
  }
}

/**
 * Implements hook_islandora_object_modified().
 *
 * This hook applies to object properties only, and not to modifications
 * to datastreams.
 *
 * We store a copy of the Islandora object in the session table to avoid
 * multiple Bags being created on datastream modification, since whenever
 * you update a datastream, the object gets updated as well, firing this hook.
 */
function islandora_bagit_islandora_object_modified(FedoraObject $islandora_object) {
  if (islandora_bagit_access_check($islandora_object)) {
    if (variable_get('islandora_bagit_create_on_modify', 0)) {
      // Don't create a Bag if we are dealing with a collection.
      if (!isset($islandora_object["COLLECTION_POLICY"])) {
        $_SESSION['islandora_object'] = $islandora_object;
      }
    }
  }
}

/**
 * Implements hook_islandora_datastream_modified().
 *
 * Fired when the properties or content of any datastream are about
 * to be modified.
 *
 * We store a copy of the Islandora object in the session table to avoid
 * multiple Bags being created on datastream modification.
 */
function islandora_bagit_islandora_datastream_modified(AbstractObject $islandora_object, AbstractDatastream $datastream) {
  if (islandora_bagit_access_check($islandora_object)) {
    if (variable_get('islandora_bagit_create_on_modify', 0)) {
      // Don't create a Bag if we are dealing with a collection.
      if (!isset($islandora_object["COLLECTION_POLICY"])) {
        $_SESSION['islandora_object'] = $islandora_object;
      }
    }
  }
}

/**
 * Implements hook_islandora_datastream_modified().
 */
function islandora_bagit_islandora_datastream_ingested(AbstractObject $islandora_object, AbstractDatastream $datastream) {
  if (islandora_bagit_access_check($islandora_object)) {
    if (variable_get('islandora_bagit_create_on_modify', 0)) {
      // Don't create a Bag if we are dealing with a collection.
      if (!isset($islandora_object["COLLECTION_POLICY"])) {
        $_SESSION['islandora_object'] = $islandora_object;
      }
    }
  }
}

/**
 * Implements hook_islandora_datastream_purged().
 */
function islandora_bagit_islandora_datastream_purged(AbstractObject $islandora_object, $dsid) {
  if (islandora_bagit_access_check($islandora_object)) {
    if (variable_get('islandora_bagit_create_on_modify', 0)) {
      // Don't create a Bag if we are dealing with a collection.
      if (!isset($islandora_object["COLLECTION_POLICY"])) {
        $_SESSION['islandora_object'] = $islandora_object;
      }
    }
  }
}

/**
 * Placeholder line to meet Coder style checks.
 *
 * Checks session to see if either islandora_bagit_islandora_object_modified()
 * or islandora_bagit_islandora_datastream_modified() saved a copy of the
 * Islandora object to the session. See additional details at those function
 * definitions.
 */
function islandora_bagit_create_bag_on_shutdown() {
  if (isset($_SESSION['islandora_object'])) {
    islandora_bagit_create_bag($_SESSION['islandora_object']);
    // Clear this variable in the session so the Bag isn't generated again.
    unset($_SESSION['islandora_object']);
  }
}

/*******************
 * Misc. functions *
 * ****************/

/**
 * Placeholder line to meet Coder style checks.
 *
 * Iterates through the Islandora object, saves each specified datastream
 * as a file in a temporary directory using the DSID as the filename,
 * and returns a list of all the files saved.
 *
 * @param object $islandora_object
 *   The Islandora object to create a Bag for.
 *
 * @param array $datastreams
 *   List of DSIDs to retrieve. If empty, retrieve all datastreams.
 *
 * @return array
 *   List of all the files saved.
 */
function islandora_bagit_retrieve_datastreams($islandora_object, $tmp_ds_directory, $datastreams = array()) {
  $ds_files = array();
  $mime_detect = new MimeDetect();
  foreach ($islandora_object as $ds) {
    if (islandora_datastream_access(ISLANDORA_VIEW_OBJECTS, $islandora_object[$ds->id])) {
      // If $datastreams is empty, retrieve all datastreams.
      if ((count($datastreams) == 0) || in_array($ds->id, $datastreams)) {
        $extension = $mime_detect->getExtension($ds->mimetype);
        $ds_content_file_path = $tmp_ds_directory . DIRECTORY_SEPARATOR . $ds->id . '.' .
          $extension;
        // Only get the datastream if its file doesn't already exist.
        if (!file_exists($ds_content_file_path)) {
          try {
            $ds->getContent($ds_content_file_path);
            if (!in_array($ds_content_file_path, $ds_files)) {
              $ds_files[] = $ds_content_file_path;
            }
          }
          catch (RepositoryException $e) {
            drupal_set_message(t('Cannot save datastream file.'), 'warning');
            watchdog('bagit', 'Could not save datastream file for datastream !dsid to !path.',
              array('!dsid' => $id->id, '!path' => $ds_content_file_path));
          }
        }
        else {
          // If the file already exists, add its path to the return array so it
          // will be registered with the plugin.
          if (strlen($ds_content_file_path)) {
            if (!in_array($ds_content_file_path, $ds_files)) {
              $ds_files[] = $ds_content_file_path;
            }
          }
        }
      }
    }
  }
  return $ds_files;
}

/**
 * Serialize and write to a file the Bag object.
 *
 * Saving the Bag to a $_SESSION variable resulted in unavoidable
 * 'incomplete object' errors. Note: This has nothing to do with
 * serializing (zipping) a Bag, it's a replacement for using Drupal's
 * $_SESSION.
 *
 * @param object $bag
 *   The Bag object.
 *
 * @param string $pid
 *   The PID of the collection object that the Bag is being created for.
 *   This PID should already be sanitized so that it can be used in file paths.
 *
 * @param string $tmp_ds_directory
 *   The randomly-generated directory path to the temporary files used in
 *   creation of this Bag.
 */
function islandora_bagit_serialize_bag_object($bag, $pid, $tmp_ds_directory) {
  // We can't pass any arguments to islandora_bagit_unserialize_bag_object()
  // so save the path to the serialized data in the session.
  $_SESSION['islandora_bagit_tmp_ds_directory'] = $tmp_ds_directory;
  $serialized_bag_blob = serialize($bag);
  file_put_contents($tmp_ds_directory . DIRECTORY_SEPARATOR .
    'serialized_bag_object.dat', $serialized_bag_blob
  );
}

/**
 * Placeholder line to meet Coder style checks.
 *
 * Unserialize a Bag object serialized by
 * islandora_bagit_serialize_bag_object(). Note: The BagitPHP library
 * will need to be loaded in the scope that this function is called in,
 * otherwise we get fatal 'incomplete object' errors.
 *
 * @return object
 *   The unserialized Bag object.
 */
function islandora_bagit_unserialize_bag_object() {
  $tmp_ds_directory = $_SESSION['islandora_bagit_tmp_ds_directory'];
  $path = $tmp_ds_directory . DIRECTORY_SEPARATOR . 'serialized_bag_object.dat';
  $serialized_bag_string = file_get_contents($path);
  $bag = unserialize($serialized_bag_string);
  return $bag;
}

/**
 * Placeholder line to meet Coder style checks.
 *
 * Generates the value for the Payload-Oxum metadata tag for
 * an object-level Bag. We generate this value for collection-
 * level Bags in islandora_bagit_collection_batch_finish_bag().
 *
 * @param array $files
 *   Array of source paths for files added to the Bag.
 *
 * @return string
 *   The Payload-Oxum value.
 */
function islandora_bagit_get_octetstream_sum($files) {
  $file_counter = 0;
  $filesize_sum = 0;
  foreach ($files as $file) {
    $file_counter++;
    $filesize_sum = $filesize_sum + (int) filesize($file);
  }
  return $filesize_sum . '.' . $file_counter;
}

/**
 * Adds metadata to the bag-info.txt metadata.
 *
 * Makes no attempt to wrap lines at 79 characters as recommended by the spec.
 *
 * @return array
 *   An array containing the tags' name => value pairs.
 */
function islandora_bagit_create_baginfo() {
  $bag_info = array();
  if (strlen(variable_get('islandora_bagit_transferring_organization', ''))) {
    $bag_info['Source-Organization'] = variable_get('islandora_bagit_transferring_organization', '');
  }
  if (strlen(variable_get('islandora_bagit_transferring_organization_address', ''))) {
    $bag_info['Organization-Address'] = variable_get('islandora_bagit_transferring_organization_address', '');
  }
  if (strlen(variable_get('islandora_bagit_contact_name', ''))) {
    $bag_info['Contact-Name'] = variable_get('islandora_bagit_contact_name', '');
  }
  if (strlen(variable_get('islandora_bagit_contact_phone', ''))) {
    $bag_info['Contact-Phone'] = variable_get('islandora_bagit_contact_phone', '');
  }
  if (strlen(variable_get('islandora_bagit_contact_email', ''))) {
    $bag_info['Contact-Email'] = variable_get('islandora_bagit_contact_email', '');
  }
  if (strlen(variable_get('islandora_bagit_profile_uri', ''))) {
    $bag_info['BagIt-Profile-Identifier'] = variable_get('islandora_bagit_profile_uri', '');
  }
  if (variable_get('islandora_bagit_bagging_date', 1)) {
    $bag_info['Bagging-Date'] = date("Y-m-d");
  }
  return $bag_info;
}

/**
 * Query the RI index to get PIDs of all child objects in the collection.
 *
 * @param object $islandora_object
 *   The Islandora collection object to create the Bags for.
 *
 * @return array
 *   List of PIDs.
 */
function islandora_bagit_get_batch_members($islandora_object) {
  // List of objects to create Bags for.
  $objects_to_bag = array();

  // Query the rindex to get all the objects that have a 'isMemberOfCollection'
  // relationship with the specified collection and add all their PIDs to the
  // list.
  $tuque = islandora_get_tuque_connection();
  $ri_query = <<<EOQ
SELECT ?object
FROM <#ri>
WHERE {
  ?object <fedora-rels-ext:isMemberOfCollection> <info:fedora/{$islandora_object->id}>
}
EOQ;
  $members = $tuque->repository->ri->sparqlQuery($ri_query, 'unlimited');
  foreach ($members as $member) {
    // Allow other modules to filter which objects get included in a batch.
    // Implementations of hook_islandora_bagit_filter_batch($pid) must
    // return TRUE if they want to filter out an object, FALSE otherwise.
    $filter = module_invoke_all('islandora_bagit_filter_batch', $member['object']['value']);
    // If any modules implement hook_islandora_bagit_filter_batch()...
    if (count($filter)) {
      if (!$filter[0]) {
        $objects_to_bag[] = $member['object']['value'];
      }
    }
    // If none do.
    else {
      $objects_to_bag[] = $member['object']['value'];
    }
  }

  return $objects_to_bag;
}

/**
 * Get list of plugin filenames.
 *
 * @param string $type
 *   Either 'object' or 'collection'.
 *
 * @return array
 *   List of plugin filenames.
 */
function islandora_bagit_get_plugins($type = 'object') {
  $plugins = array();
  $module_dir = drupal_get_path('module', 'islandora_bagit');
  $plugin_files = file_scan_directory($module_dir . '/plugins', '/^plugin_' . $type . '.*\.inc$/');
  sort($plugin_files);
  foreach ($plugin_files as $plugin) {
    $plugins[] = $plugin->name;
  }
  return $plugins;
}

/**
 * Implements hook_file_dowload().
 */
function islandora_bagit_file_download($uri) {
  // Get the file's name.
  $file_name = basename($uri);
  // Get the file's size.
  $path = variable_get('islandora_bagit_bag_output_dir', '') . DIRECTORY_SEPARATOR .
    $file_name;
  // If the file is in the Bag output directory, this module controls
  // access to it; if it's not, this module doesn't control access to
  // it so we return NULL.
  if (!file_exists($path)) {
    return NULL;
  }

  if (user_access('create Islandora Bags')) {
    $file_size = stat($path);
    return array(
      'Content-Type' => 'application/octet-stream',
      'Content-Length' => $file_size['size'],
      'Content-Disposition' => 'attachment; filename=' . $file_name,
    );
  }
  else {
    return -1;
  }
}