dkan_dataset.module

<?php

/**
 * @file
 * Custom elements for datasets.
 */

include_once 'dkan_dataset.features.inc';
include_once 'dkan_dataset.theme.inc';
include_once 'includes/getRemoteFileInfo.php';
include_once 'dkan_dataset.forms.inc';

define('DKAN_REMOTE_FILE_SOURCE_HINT_TEXT', 'http://example.com/files/file.zip');

/**
 * Implements hook_preprocess_html().
 */
function dkan_dataset_preprocess_html(&$vars) {
  $path = drupal_get_path('module', 'dkan_dataset');
  drupal_add_css($path . '/css/dkan_dataset.css');
  // Add body class for resource teaser type.
  if ($node = menu_get_object()) {
    if ($node->type == 'dataset') {
      $preview = variable_get('dkan_dataset_teaser_preview', 'explore');
      $vars['classes_array'][] = $preview;
    }
  }
  drupal_add_css($path . '/css/dkan-flaticon.css');
}

/**
 * Implements hook_menu().
 */
function dkan_dataset_menu() {
  $path = drupal_get_path('module', 'dkan_dataset');
  $items['node/%node/resource'] = array(
    'title' => 'Add Resource',
    'page callback' => 'dkan_dataset_add_resource',
    'page arguments' => array(1),
    'access callback' => 'dkan_dataset_add_resource_access',
    'access arguments' => array(1),
    'type' => MENU_LOCAL_TASK,
    'file' => 'dkan_dataset.pages.inc',
    'file path' => $path,
  );
  $items['node/%node/dataset'] = array(
    'title' => 'Back to dataset',
    'page callback' => 'dkan_dataset_back',
    'page arguments' => array(1),
    'access callback' => 'dkan_dataset_back_access',
    'access arguments' => array(1),
    'type' => MENU_LOCAL_TASK,
    'file' => 'dkan_dataset.pages.inc',
    'file path' => $path,
  );
  $items['node/%node/dataset/download'] = array(
    'title' => 'Download Dataset',
    'page callback' => 'dkan_dataset_download',
    'page arguments' => array(1),
    'access callback' => 'dkan_dataset_download_access',
    'access arguments' => array(1),
    'file' => 'dkan_dataset.pages.inc',
    'file path' => $path,
  );
  $items['admin/dkan'] = array(
    'title' => 'DKAN',
    'description' => 'DKAN Settings and Administration.',
    'weight' => -8,
    'position' => 'left',
    'page callback' => 'system_admin_menu_block_page',
    'access arguments' => array('access administration pages'),
    'file' => 'system.admin.inc',
    'file path' => drupal_get_path('module', 'system'),
  );
  $items['admin/dkan/dataset_preview'] = array(
    'title' => 'DKAN Dataset Previews',
    'description' => 'Settings for DKAN Dataset local and external previews.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('dkan_dataset_preview_settings'),
    'access arguments' => array('administer site configuration'),
    'file' => 'dkan_dataset.admin.inc',
  );
  $items['admin/dkan/dataset_forms'] = array(
    'title' => 'DKAN Dataset Forms',
    'description' => 'Form settings for DKAN Datasets.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('dkan_dataset_form_settings'),
    'access arguments' => array('administer site configuration'),
    'file' => 'dkan_dataset.admin.inc',
  );
  if (module_exists('open_data_schema_map_dkan')) {
    $items['admin/dkan/dataset_api'] = array(
      'title' => 'DKAN Dataset API',
      'description' => 'Settings for DKAN Dataset APIs.',
      'page callback' => 'dkan_dataset_api_menu_settings',
      'access arguments' => array('administer site configuration'),
      'file' => 'dkan_dataset.admin.inc',
    );
  }
  return $items;
}

/**
 * Implements hook_node_view().
 */
function dkan_dataset_node_view($node, $view_mode, $langcode) {
  if ($view_mode == 'full' && $node->type == 'dataset') {

    $node->content['resources'] = array(
      '#theme' => 'dkan_dataset_resource_view',
      '#node' => $node,
      '#enabled' => TRUE,
      '#prefix' => '<div id="data-and-resources">',
      '#suffix' => '</div>',
      '#view_mode' => $view_mode,
    );
    $node->content['#attached']['css'][] = drupal_get_path('module', 'dkan_dataset') . '/css/dkan_dataset.css';
    $node->content['#attached']['js'][] = drupal_get_path('module', 'dkan_dataset') . '/js/dkan_tooltip.js';
    $node->content['#attached']['js'][] = array('data' => drupal_get_path('module', 'dkan_dataset') .'/js/dataset.js');
    $node->content['release_date'] = array(
      '#theme' => 'dkan_dataset_release_date_view',
      '#node' => $node,
      '#enabled' => TRUE,
      '#title' => t('Release Date'),
      '#label_display' => 'above',
      '#items' => array(),
      '#field_name' => '',
      '#field_type' => '',
      '#bundle' => $node->type,
      '#view_mode' => $view_mode,
    );
    $node->content['modified_date'] = array(
      '#theme' => 'dkan_dataset_modified_date_view',
      '#node' => $node,
      '#enabled' => TRUE,
      '#title' => t('Modified Date'),
      '#label_display' => 'above',
      '#items' => array(),
      '#field_name' => '',
      '#field_type' => '',
      '#bundle' => $node->type,
      '#view_mode' => $view_mode,
    );
    $node->content['identifier'] = array(
      '#theme' => 'dkan_dataset_identifier_view',
      '#node' => $node,
      '#enabled' => TRUE,
      '#title' => t('Identifier'),
      '#label_display' => 'above',
      '#items' => array(),
      '#field_name' => '',
      '#field_type' => '',
      '#bundle' => $node->type,
      '#view_mode' => $view_mode,
    );
  }
  elseif (($view_mode == 'teaser' || $view_mode == 'search_result') && ($node->type == 'dataset' || $node->type == 'resource')) {
    if (isset($node->content['links']['node']['#links']['node-readmore'])) {
      unset($node->content['links']['node']['#links']['node-readmore']);
    }
    $node->content['body']['#weight'] = '-10';
    $node->content['resources'] = array(
      '#type' => 'item',
      '#theme' => 'dkan_dataset_resource_teaser_view',
      '#node' => $node,
      '#enabled' => TRUE,
      '#title' => '',
      '#label_display' => 'above',
      '#prefix' => '<div class="data-and-resources">',
      '#suffix' => '</div>',
      '#weight' => '20',
    );
    if (isset($node->content['field_topic'])) {
      $node->content['field_topic']['#weight'] = '5';
    }
  }
  elseif ($view_mode == 'full' && $node->type == 'resource') {
    $node->content['#attached']['css'][] = drupal_get_path('module', 'dkan_dataset') . '/css/dkan_dataset.css';
    $node->content['#attached']['js'][] = drupal_get_path('module', 'dkan_dataset') . '/js/dkan_tooltip.js';
  }
}

/**
 * Implements hook_field_extra_fields().
 */
function dkan_dataset_field_extra_fields() {
  $extra['node']['dataset'] = array(
    'form' => array(
      'resources' => array(
        'label' => t('Data and Resources'),
        'description' => t('Shows book children of dataset'),
        'weight' => 0,
      ),
      'release_date' => array(
        'label' => t('Release Date'),
        'description' => t('Node published date available as a field'),
        'weight' => 0,
      ),
      'modified_date' => array(
        'label' => t('Modified Date'),
        'description' => t('Node changed date available as a field'),
        'weight' => 0,
      ),
      'identifier' => array(
        'label' => t('Identifier'),
        'description' => t('Unique identifier for dataset'),
        'weight' => 0,
      ),
    ),
    'display' => array(
      'resources' => array(
        'label' => t('Data and Resources'),
        'description' => t('Shows book children of dataset'),
        'weight' => 0,
      ),
      'release_date' => array(
        'label' => t('Release Date'),
        'description' => t('Node published date available as a field'),
        'weight' => 0,
      ),
      'modified_date' => array(
        'label' => t('Modified Date'),
        'description' => t('Node changed date available as a field'),
        'weight' => 0,
      ),
      'identifier' => array(
        'label' => t('Identifier'),
        'description' => t('Unique identifier for dataset'),
        'weight' => 0,
      ),
    ),
  );

  return $extra;
}

/**
 * Helper to get Resource nodes linked to a Dataset.
 */
function dkan_dataset_get_resource_nodes($nid) {
  $nodes = array();
  $nids = array();
  $query = new EntityFieldQuery();

  $results = $query
    ->entityCondition('entity_type', 'node')
    ->fieldCondition('field_dataset_ref', 'target_id', $nid, '=')
    ->execute();

  if ($results) {
    foreach ($results as $node) {
      foreach ($node as $nid => $obj) {
        $nids[] = $nid;
      }
    }
  }
  $nodes = isset($nids) ? node_load_multiple($nids) : array();
  return $nodes;
}

/**
 * Helper to get Visualization entities that are created from a Resource.
 */
function dkan_dataset_get_visualization_entities($uuid) {
  // The EFQ will throw an error if the uuid_resource field doesn't exist.
  if (!field_info_instance('visualization', 'field_uuid_resource','ve_chart')) {
  return array();
  }
  $nodes = array();
  $nids = array();
  $query = new EntityFieldQuery();

  $results = $query
    ->entityCondition('entity_type', 'visualization')
    ->fieldCondition('field_uuid_resource', 'target_uuid', $uuid, '=')
    ->execute();

  if ($results) {
    foreach ($results as $node) {
      foreach ($node as $uuid => $obj) {
        $nids[] = $uuid;
      }
    }
  }
  $nodes = isset($nids) ? node_load_multiple($nids) : array();
  return $nodes;
}

/**
 * Helper to get the format string for a resource node
 *
 * @param $node
 *   A resource node
 *
 * @return
 *   The format string (eg, "csv", "html", etc.). Defaults to "data".
 */
function dkan_dataset_get_resource_format($node) {
  $node_wrapper = entity_metadata_wrapper('node', $node);
  if ($node_wrapper->getBundle() != 'resource') {
    return FALSE;
  }
  if (isset($node->field_format['und']) && $node_wrapper->field_format->value()->name) {
    return $node_wrapper->field_format->value()->name;
  }
  return 'data';
}

/**
 * Helper to get title from nid.
 */
function dkan_dataset_get_title($nid) {
  $result = db_select('node', 'n')
    ->fields('n', array('title'))
    ->condition('nid', $nid, '=')
    ->execute()
    ->fetchAssoc();

  if (isset($result['title'])) {
    return $result['title'];
  }
  else {
    return FALSE;
  }
}

/**
 * Returns trimmed text using views_trim_text().
 */
function dkan_dataset_text_trim($title, $number = 25) {
  $alter = array(
    'max_length' => $number,
    'ellipsis' => TRUE,
    'word_boundary' => TRUE,
    'trim' => TRUE,
  );
  return views_trim_text($alter, $title);
}

/**
 * Implements hook_libraries_info().
 */
function dkan_dataset_libraries_info() {
  $libraries['slugify'] = array(
    'name' => 'Slugify',
    'vendor url' => 'https://github.com/pmcelhaney/jQuery-Slugify-Plugin',
    'download url' => 'https://github.com/pmcelhaney/jQuery-Slugify-Plugin/archive/master.zip',
    'files' => array('js' => array('jquery.slugify')),
    'version' => 'master',
  );

  return $libraries;
}

/**
 * Creates the part on the node edit form that says what stage you are on.
 */
function dkan_form_create_stages($op, $dataset_nid = NULL, $resource_nid = NULL) {
  // TODO: should be theme function.
  $stages = '<div id="dkan-stages">';
  // Omit "Additional Info" depending on variable.
  $stages_additional_1 = $stages_additional_2 = $stages_additional_3 = $stages_additional_4 = '';
  if (variable_get('dkan_dataset_form_additional_info', 1)) {
    $stages_additional_1 = '
     <li class="last uncomplete">
        <span class="highlight">' . l(t('Additional data'), 'node/' . $dataset_nid . '/edit', array('query' => array('additional' => TRUE))) . '</span>
    </li>';
    $stages_additional_2 = '
    <li class="last active">
      <span class="highlight" name="save" value="go-metadata" type="submit">' . l(t('Additional data'), 'node/' . $dataset_nid . '/edit', array('query' => array('additional' => TRUE))) . '</span>
    </li> ';
    $stages_additional_3 = '
        <li class="last uncomplete">
          <span class="highlight">' . t('Additional data') . '</span>
        </li> ';
    $stages_additional_4 = '
    <li class="last complete">
        <span class="highlight">' . l(t('Additional data'), 'node/' . $dataset_nid . '/edit', array('query' => array('additional' => TRUE))) . '</span>
    </li> ';

  }
  if ($op == 'resource-edit' || $op == 'resource-create') {
    $stages .= '<ol class="stages stage-3">
      <li class="first complete"><span class="highlight">' . l(t('Edit dataset'), 'node/' . $dataset_nid . '/edit') . '</span></li>
      <li class="middle active">
          <span class="highlight">Add data</span>
      </li> ' . $stages_additional_1 . '
    </ol>';
  }
  if ($op == 'dataset-additional') {
    $stages .= '<ol class="stages stage-3">
      <li class="first complete">
          <span class="highlight">' . l(t('Edit dataset'), 'node/' . $dataset_nid . '/edit') . '</span>
      </li>
      <li class="middle complete">
          <span class="highlight">' . l(t('Add dataset'), 'node/add/resource', array('query' => array('dataset' => $dataset_nid))) . '</span>
      </li> ' . $stages_additional_2 . '
    </ol>';
  }
  if ($op == 'dataset-create') {
    $stages .=
        '<ol class="stages stage-1">
          <li class="first active">
            <span class="highlight">' . t('Create dataset') . '</span>
          </li>
          <li class="middle uncomplete">
            <span class="highlight">' . t('Add data') . ' </span>
          </li> ' . $stages_additional_3 . '
        </ol>';
  }
  if ($op == 'dataset-edit') {
    $stages .= '<ol class="stages stage-1">
        <li class="first active">
            <span class="highlight">' . t('Create dataset') . '</span>
      </li>
      <li class="middle complete">
          <span class="highlight">' . l(t('Add data'), 'node/add/resource', array('query' => array('dataset' => $dataset_nid))) . '</span>
      </li> ' . $stages_additional_4 . '
    </ol>';
  }
  $stages .= '</div>';
  return $stages;
}

/**
 * Access callback for back link.
 */
function dkan_dataset_back_access($node) {
  if ($node->type != 'resource') {
    return FALSE;
  }
  else {
    return node_access('view', $node);
  }
}

/**
 * Access callback for 'Download Dataset' tab.
 */
function dkan_dataset_download_access($node) {
  if ($node->type != 'dataset') {
    return FALSE;
  }
  else {
    $node = entity_metadata_wrapper('node', $node);
    $resources = $node->field_resources->value();
    $file_count = 0;
    foreach ($resources as $resource) {
      if (isset($resource->field_upload[LANGUAGE_NONE])) {
        $file_count++;
      }
    }
    if ($file_count > 1) {
      return TRUE;
    }
  }
}

/**
 * Access callback for 'Add Resource' tab.
 */
function dkan_dataset_add_resource_access($node) {
  if ($node->type != 'dataset') {
    return FALSE;
  }
  elseif (node_hook($node->type, 'form') && node_access('update', $node)) {
    return TRUE;
  }
  else {
   return FALSE;
  }
}

/**
 * Discovers extension based off of mimetype.
 *
 * @param string $mimetype
 *   Mimetype to search for.
 *
 * @return string
 *   Extension or empty string.
 */
function dkan_dataset_discovery_extension($mimetype) {
  include_once DRUPAL_ROOT . '/includes/file.mimetypes.inc';
  $type = '';
  $mapping = file_mimetype_mapping();
  $offset = array_search($mimetype, $mapping['mimetypes']);
  if ($offset) {
    $type = array_search($offset, $mapping['extensions']);
  }
  if (!$type) {
    // Allows for custom mappings despite file_mimetype_mapping only allowing one
    // mimetype per extension
    switch ($offset) {
      case 998:
        $type = 'xml';
        break;
      default:
        $type = $mimetype;
        break;
    }
  }
  return $type;
}

/**
 * Implements hook_file_mimetype_mapping_alter().
 */
function dkan_dataset_file_mimetype_mapping_alter(&$mapping) {
  $mapping['extensions']['json'] = 999;
  $mapping['mimetypes'][999] = 'application/json';
  // Text/xml would map to the same extention as application/xml, so we'll
  // fix trhat in dkan_dataset_discovery_extension()
  $mapping['mimetypes'][998] = 'text/xml';
  // 'Htm' and 'html' have the same mapping number but 'htm' is after
  // content types of 'text/html' are alway 'htm' without this.
  unset($mapping['extensions']['htm']);
  // Same with 'shtml'.
  unset($mapping['extensions']['shtml']);
}

/**
 * Implements hook_node_update().
 */
function dkan_dataset_node_update($node) {

  global $user;

  switch ($node->type) {
    case 'dataset':
      // Update resource's groups.
      dkan_dataset_sync_groups($node);
      break;

    case 'resource':
      // Add changelog on associated datasets.
      $resource = entity_metadata_wrapper('node', $node);
      $datasets = $resource->field_dataset_ref->raw();
      if (count($datasets)) {
        $queue = DrupalQueue::get('dataset_changelog');
        foreach ($datasets as $dataset) {
          $item = array(
            'dataset' => $dataset,
            'message' => 'Update to resource \'' . $node->title . '\'',
            'user' => $user->uid,
          );
          $queue->createItem($item);
        }
      }
      break;

    default:
      break;
  }
}

/**
 * Syncs groups on resources across groups.
 *
 * Resources attached to datasets need to add or remove groups based on whether
 * groups are added to a dataset or resources are added or removed from
 * a dataset.
 */
function dkan_dataset_sync_groups($dataset) {

  // This only makes sense if the 'og' module is enabled.
  if (module_exists('og')) {

    // Wrap dataset nodes as entities for easy usage.
    $wrapped_dataset_old = entity_metadata_wrapper('node', $dataset->original);
    $wrapped_dataset_new = entity_metadata_wrapper('node', $dataset);

    // Get all (new + old) groups associated with the dataset.
    $dataset_groups_old = dkan_dataset_get_dataset_groups($wrapped_dataset_old);
    $dataset_groups_new = dkan_dataset_get_dataset_groups($wrapped_dataset_new);

    // Get all (new + old) resources associated with the dataset.
    $dataset_resources_old = dkan_dataset_get_dataset_resources($wrapped_dataset_old);
    $dataset_resources_new = dkan_dataset_get_dataset_resources($wrapped_dataset_new);

    // Check changes on the dataset
    $groups_changed = ($dataset_groups_old != $dataset_groups_new) ? true : false;
    $resources_changed = ($dataset_resources_old != $dataset_resources_new) ? true : false;

    // If groups changed or resources changed then we need to update the groups on all affected resources.
    if ($groups_changed || $resources_changed) {

      // Process resources and separate them based on 'added', 'removed', 'unmodified' status.
      $resources = dkan_dataset_process_resources($dataset_resources_old, $dataset_resources_new);

      // Update 'added' resources. Append the current dataset's groups to their other groups.
      if (isset($resources['added'])) {
        dkan_dataset_update_resource_groups($resources['added'], $wrapped_dataset_new, $dataset_groups_new);
      }

      // Update 'removed' resources. Don't append the current dataset's groups to their other groups.
      if (isset($resources['removed'])) {
        dkan_dataset_update_resource_groups($resources['removed'], $wrapped_dataset_new, NULL);
      }

      // If the groups changed we need to also update the resources that were already
      // associated with the dataset. Append the current dataset's groups to their other groups.
      if ($groups_changed && isset($resources['unmodified'])){
        dkan_dataset_update_resource_groups($resources['unmodified'], $wrapped_dataset_new, $dataset_groups_new);
      }
    }
  }
}

/**
 * Process resource and group them based on their 'added', 'removed' or 'unmodified' status.
 *
 * @param $resources_old
 *    The list of old resources associated with the dataset that is being updated.
 * @param $resources_new
 *    The list of new resources associated with the dataset that is being updated.
 */
function dkan_dataset_process_resources($resources_old, $resources_new) {
  $processed_resources = array();

  // Join all resources (old + new). Remove duplicates.
  $all_resources = $resources_old;
  foreach($resources_new as $resource_new_key => $resource_new) {
    $all_resources[$resource_new_key] = $resource_new;
  }

  // Organize resources separating them by their 'added', 'removed', 'unmodified' status.
  foreach ($all_resources as $resource_key => $resource) {

    $status = 'added';
    if (isset($resources_new[$resource_key])) {
      if (isset($resources_old[$resource_key])) {
        $status = 'unmodified';
      }
    }
    else {
      $status = 'removed';
    }

    $processed_resources[$status][$resource_key] = $resource;
  }

  return $processed_resources;
}

/**
 * Get the list of groups associated with the dataset.
 *
 * @param $dataset
 *    Entity wrapped version of the dataset node.
 * @return array $groups
 *    An array containing the list of groups associated with the dataset.
 */
function dkan_dataset_get_dataset_groups($dataset) {

  $groups = array();

  $field_groups_values = $dataset->__isset('og_group_ref') ? $dataset->og_group_ref->value() : '';
  foreach ($field_groups_values as $field_groups_value) {
    $groups[$field_groups_value->nid] = $field_groups_value;
  }

  return $groups;
}

/**
 * Get the list of resources associated with the dataset.
 *
 * @param $dataset
 *    Entity wrapped version of the dataset node.
 * @return array $resources
 *    An array containing the list of resources associated with the dataset.
 */
function dkan_dataset_get_dataset_resources($dataset) {

  $resources = array();

  $field_resources_values = $dataset->__isset('field_resources') ? $dataset->field_resources->value() : '';
  foreach($field_resources_values as $field_resources_value) {
    $resources[$field_resources_value->nid] = $field_resources_value;
  }

  return $resources;
}

/**
 * Get the full list of groups associated with all the datasets where the
 * resource was added except the ones from the dataset that is being updated.
 *
 * @param $resource
 *    Entity wrapped version of the resource node.
 * @param $updated_dataset_id
 *    ID of the dataset that is currently being updated.
 * @return array $groups
 *    Array containing the full list of groups associated with all the datasets where the
 *    resource was added except the ones from the dataset that is being updated.
 */
function dkan_dataset_get_resource_groups($resource, $updated_dataset_id) {

  $groups = array();

  // Check if the resource is associated with any datasets.
  if (isset($resource->field_dataset_ref)) {

    $referenced_datasets = $resource->field_dataset_ref->value();

    // Get the groups from all the datasets associated with the resource.
    // A resource can be associated with multiple datasets.
    foreach ($referenced_datasets as $referenced_dataset_key => $referenced_dataset) {

      // Get groups of all the associated datasets except the one that's currently
      // being edited. Those groups will be added after if needed.
      if ($referenced_dataset->nid != $updated_dataset_id) {
        // Load the latest version of the referenced dataset.
        $dataset = node_load($referenced_dataset->nid, NULL, TRUE);
        // Wrap node as entity for easy usage.
        $dataset_wrapper = entity_metadata_wrapper('node', $dataset);

        // Get the list of groups associated with the dataset and append those to the result list.
        $dataset_groups = $dataset_wrapper->og_group_ref->value();
        foreach ($dataset_groups as $group_id => $group) {
          $groups[$group->nid] = $group;
        }
      }
    }
  }

  return $groups;
}

/**
 * Update the associated groups on all the resources that were modified as part of the
 * dataset update.
 *
 * @param $resources
 *    The list of resources whose resources need to be updated.
 * @param $updated_dataset
 *    Entity wrapped version of the dataset that was updated.
 * @param $updated_dataset_groups
 *    The current groups associated with the dataset that is being updated.
 * @param int $use_queue_limit
 *    A limit indicating when a Drupal queue should be used to update the resources.
 */
function dkan_dataset_update_resource_groups($resources, $updated_dataset, $updated_dataset_groups, $use_queue_limit = 20) {

  $use_queue = false;
  $queue = '';

  $num_resources = count($resources);

  // Check if a Drupal queue should be used to update the resources.
  if ($num_resources > $use_queue_limit) {
    $use_queue = true;
    $queue = DrupalQueue::get('dkan_dataset_group_update');
    $queue->createQueue();
  }

  foreach($resources as $resource) {

    // Load latest version of the resource.
    $resource = node_load($resource->nid, NULL, TRUE);
    $wrapper_resource = entity_metadata_wrapper('node', $resource);

    // Get all the resource groups from all the other datasets the resource was associated with.
    $resource_groups = dkan_dataset_get_resource_groups($wrapper_resource, $updated_dataset->nid->value());
    // If present then append the groups of the current dataset.
    if ($updated_dataset_groups) {
      $resource_groups = array_merge($resource_groups, $updated_dataset_groups);
    }

    // Update resource groups.
    if ($use_queue) {
      $queue->createItem(array(
        'wrapper' => $wrapper_resource,
        'new_groups' => $resource_groups
      ));
    }
    else {
      $wrapper_resource->og_group_ref->set($resource_groups);
      $wrapper_resource->save();
    }
  }

  // Log messages to the user.
  if ($num_resources) {
    if ($use_queue) {
      drupal_set_message(t("%d resources were queued to be updated with their new groups. They will be processed during upcoming cron runs.", array(
        '%d' => $num_resources
      )));
    }
    else {
      drupal_set_message(t("Groups were updated on %d resource(s).", array(
        '%d' => $num_resources
      )));
    }
  }
}

/**
 * Implements hook_cron_queue_info().
 */
function dkan_dataset_cron_queue_info() {
  $queues = array();
  $queues['dkan_dataset_group_update'] = array(
    'worker callback' => 'dkan_dataset_update_groups',
    'time' => 30,
  );
  $queues['dataset_changelog'] = array(
    'worker callback' => 'dkan_dataset_dataset_changelog_run',
    'time' => 20,
  );
  return $queues;
}

/**
 * Processes resources.
 */
function dkan_dataset_update_groups($data) {
  $data['wrapper']->og_group_ref->set($data['new_groups']);
  $data['wrapper']->save();
}

/**
 * Callback for dkan_dataset_dataset_changelog cron worker.
 *
 * @param array $item
 *   An associative array containing the dataset nid and a log message.
 */
function dkan_dataset_dataset_changelog_run($item) {
  global $user;
  $original_user = $user;
  drupal_save_session(FALSE);
  // Increment Revision.
  if (isset($item['user']) && isset($item['dataset']) && isset($item['message'])) {
    $user = user_load($item['user']);
    ctools_include('Dataset', 'dkan_dataset');
    $dataset = Dataset::instance($item['dataset']);
    if ($dataset->created) {
      $dataset->incrementRevision($item['message']);
      $dataset->save();
    }
  }
  // Restore Anonymous user.
  $user = $original_user;
  drupal_save_session(TRUE);
}

/**
 * Implements hook_page_alter().
 */
function dkan_dataset_page_alter(&$page) {
  $breadcrumbs = drupal_get_breadcrumb();
  if (arg(0) == 'dataset' && is_array($breadcrumbs)) {
    $breadcrumbs[1] = t('Datasets');
    $breadcrumbs[2] = t('Search');
    // We can return here because page_title is handled by views.
    drupal_set_breadcrumb($breadcrumbs);
  }
}

/**
 * Retrieves all external preview links for a given dataset node. The
 * dkan_dataset module ships with three previews - dkan, arcgis and cartodb.
 * Additional previews can be added by implementing
 * hook_dkan_dataset_external_previews_alter().
 *
 * @param $node
 *   The dataset node object.
 *
 * @return array
 *   An array of associative arrays, keyed by "machine name" and containing:
 *     - name: Full name of preview type.
 *     - link_callback: Callback function for generating the preview link, given
 *       a node object
 *     - suggested_formats: array of formats suggested for use with this preview.
 *       Will populate the format vocabulary and be used in preview definitions.
 *     - description: Full description of the preview type.
 *     - format: Array of the formats this preview is currently enabled for.
 */
function dkan_dataset_teaser_get_external_previews() {
  $previews = array();
  $previews['dkan'] = array(
    'name' => variable_get('dkan_dataset_teaser_preview_label', '') . ' ' . t('Preview'),
    'link_callback' => 'dkan_dataset_preview_url_local',
    'suggested_formats' => array('csv', 'xls'),
    'description' => t('Local preview. Preview tabular data in a grid, basic chart or map. This preview type is not optional and will load only for the suggested formats listed.'),
  );
  // For DKAN, we don't add the format mannually, so copy the values from suggested_formats
  $previews['dkan']['format'] = $previews['dkan']['suggested_formats'];

  $previews['arcgis'] = array(
    'name' => t('ArcGIS Preview'),
    'link_callback' => 'dkan_dataset_preview_url_arcgis',
    'suggested_formats' => array('rest', 'esri rest', 'arcgis'),
    'description' => t('ArcGIS preview for ESRI endpoints. Requires a URL in the resource API field; will not work with resource files.'),
  );
  $previews['cartodb'] = array(
    'name' => t('CartoDB Preview'),
    'link_callback' => 'dkan_dataset_preview_url_cartodb',
    'suggested_formats' => array('csv', 'xls', 'excel', 'openxml', 'kml', 'geojson'),
    'description' => t('Previews resources with geodata in the CartoDB mapping service.'),
  );

  drupal_alter('dkan_dataset_external_previews', $previews);

  // Load format vocabulary into array of strings
  $format_vocabulary = taxonomy_vocabulary_machine_name_load('format');
  $formats = taxonomy_get_tree($format_vocabulary->vid);
  $format_names = array();
  // Get an array of just term names
  foreach ($formats as $format) {
    $format_names[] = $format->name;
  }

  // Make sure there is a taxonomy entry for every "suggested" format, and an
  // array for each $preview[format].
  $suggested_formats = array();
  // Collect all suggested formats across preview types.
  foreach ($previews as $machine_name => $preview) {
    $suggested_formats = array_merge($suggested_formats, $preview['suggested_formats']);
    if (!isset($preview['format'])) {
      $previews[$machine_name]['format'] = array();
    }
  }
  $suggested_formats = array_unique($suggested_formats);
  foreach ($suggested_formats as $suggested_format) {
     if (!in_array($suggested_format, $format_names)) {
        $new_format = new stdClass();
        $new_format->name = $suggested_format;
        $new_format->vid = $format_vocabulary->vid;
        taxonomy_term_save($new_format);
     }
  }

  // Now populate the formats array
  foreach ($formats as $format) {
    $var_name = "dkan_dataset_format_previews_tid{$format->tid}";
    $format_settings = variable_get($var_name, array());
    if (!empty($format_settings)) {
      foreach ($format_settings as $machine_name => $value) {
        if ($value) {
          $previews[$machine_name]['format'][] = $format->name;
        }
      }
    }
  }
  return $previews;
}

/**
 * Get array of previews for a specific node
 * @param  object $node The resource node object
 * @return array       Array of preview arrays.
 */
function dkan_dataset_teaser_external_previews_for_resource($node) {
  $resource_wrapper = entity_metadata_wrapper('node', $node);
  $previews = dkan_dataset_teaser_get_external_previews();
  // Get resource format
  $format = t("data");
  if ($resource_wrapper->field_format->value()) {
    $format = $resource_wrapper->field_format->value()->name;
  }
  foreach ($previews as $provider => $preview) {
    // Filter by format and availability of arguments (if any argument is NULL, no dice)
    // @todo allow individual previews to be disabled in settings
    if (!in_array($format, $preview['format'])) {
      unset($previews[$provider]);
    }
    else {
      if ($url = $preview['link_callback']($node)) {
        $previews[$provider]['url'] = $url;
      }
      else {
        unset($previews[$provider]);
      }
    }
  }
  return $previews;
}

/**
 * Callback function for local preview URLs
 * @param  object $node Resource node object
 * @return string       Preview URL
 */
function dkan_dataset_preview_url_local($node) {
  if ($node->type == 'resource') {
    return 'node/' . $node->nid;
  }
}

/**
 * Callback function for ArcGIS preview URLs
 * @param  object $node Resource node object
 * @return string       Preview URL
 */
function dkan_dataset_preview_url_arcgis($node) {
  // Get resource API link URL
  $resource_wrapper = entity_metadata_wrapper('node', $node);
  if ($resource_wrapper->field_link_api->value()) {
    return 'http://www.arcgis.com/home/webmap/viewer.html?url=' . $resource_wrapper->field_link_api->url->value();
  }
}

/**
 * Callback function for CartoDB preview URLs
 * @param  object $node Resource node object
 * @return string       Preview URL
 */
function dkan_dataset_preview_url_cartodb($node) {
  // Get resource file URL
  $resource_file_url = NULL;
  $resource_wrapper = entity_metadata_wrapper('node', $node);
  $pattern = 'http://oneclick.cartodb.com/?file=%s';

  if ($resource_wrapper->field_upload->value()) {
    $resource_file_url = file_create_url($resource_wrapper->field_upload->value()->uri);
  }
  elseif ($resource_wrapper->field_link_remote_file->value()) {
    $resource_file_url = $resource_wrapper->field_link_remote_file->value()->uri;
  }
  if ($resource_file_url) {
    return sprintf($pattern, $resource_file_url);
  }
}

/**
 * Implements hook_filefield_sources_info().
 */
function dkan_dataset_filefield_sources_info() {
  $source = array();
  $source['dkan_remotefile'] = array(
    'name' => t('Open Data Remote File'),
    'label' => t('Open Data Remote File'),
    'description' => t('Link to a file located on a remote server.'),
    'process' => 'dkan_dataset_remotefile_process',
    'value' => 'dkan_dataset_remotefile_value',
    'weight' => 4,
  );
  return $source;
}

/**
 * Implements hook_filefield_sources_element_validate().
 */
function dkan_filefield_sources_element_validate($element, $file) {
  $validators = $element['#upload_validators'];
  $errors = array();

    foreach ($validators as $function => $args) {
      // Add the $file variable to the list of arguments and pass it by
      // reference (required for PHP 5.3 and higher).
      array_unshift($args, NULL);
      $args[0] = &$file;
      $errors = array_merge($errors, call_user_func_array($function, $args));
    }

  // Check for validation errors.
  if (!empty($errors)) {
    $message = t('The selected file %name could not be referenced.', array('%name' => $file->filename));
    if (count($errors) > 1) {
      $message .= '<ul><li>' . implode('</li><li>', $errors) . '</li></ul>';
    }
    else {
      $message .= ' ' . array_pop($errors);
    }
    form_error($element, $message);
    return 0;
  }

  return 1;
}

/**
 * A #process callback to extend the filefield_widget element type.
 */
function dkan_dataset_remotefile_process($element, &$form_state, $form) {

  $element['filefield_dkan_remotefile'] = array(
    '#weight' => 100.5,
    '#theme' => 'filefield_source_dkan_remotefile_element',
    '#filefield_source' => TRUE, // Required for proper theming.
    '#filefield_sources_hint_text' => DKAN_REMOTE_FILE_SOURCE_HINT_TEXT,
  );

  $element['filefield_dkan_remotefile']['url'] = array(
    '#type' => 'textfield',
    '#maxlength' => NULL,
  );

  $element['filefield_dkan_remotefile']['select'] = array(
    '#name' => implode('_', $element['#array_parents']) . '_select',
    '#type' => 'submit',
    '#value' => t('Select'),
    '#validate' => array(),
    '#submit' => array('filefield_sources_field_submit'),
    '#limit_validation_errors' => array($element['#parents']),
    '#ajax' => array(
      'path' => 'file/ajax/' . implode('/', $element['#array_parents']) . '/' . $form['form_build_id']['#value'],
      'wrapper' => $element['#id'] . '-ajax-wrapper',
      'effect' => 'fade',
      'method' => 'replace',
    ),
    '#suffix' => '<div class="description">' . t('This field will ignore the file size limitation') . '</div>',
  );

  return $element;
}

/**
 * A #filefield_value_callback function.
 */
function dkan_dataset_remotefile_value($element, &$item) {
  // A remote url has been entered.
  if (isset($item['filefield_dkan_remotefile']['url'])
    && drupal_strlen($item['filefield_dkan_remotefile']['url']) > 0
    && $item['filefield_dkan_remotefile']['url'] != DKAN_REMOTE_FILE_SOURCE_HINT_TEXT) {

    $value = $item['filefield_dkan_remotefile']['url'];

    if (!valid_url($value, TRUE)) {
      form_error($element, t('Invalid Remote File URL.'));
      return;
    }
    elseif (!file_stream_wrapper_valid_scheme(file_uri_scheme($value))) {
      // Check that the scheme is supported.
      form_error($element, t('Remote File URLs with the %scheme scheme are not supported.', array('%scheme' => $scheme)));
      return;
    }
    else {
      // Check that the file exists.
      $request = drupal_http_request($value, array('method' => 'HEAD'));
      if (!empty($request->error)) {
        if (isset($request->headers['x-error-message']) && $request->headers['x-error-message'] != 'HEAD is not supported') {
          form_error($element, t('Unable to fetch file from Remote File URL %url (error @code: %error).', array('%url' => $value, '@code' => $request->code, '%error' => $request->error)));
          return;
        }
      }
    }

    try {
      $file = remote_stream_wrapper_file_create_by_uri($value);

      $file_info = new dkanDataset\getRemoteFileInfo($value, t('!site_name crawler', array('!site_name' => variable_get('site_name'))) , TRUE, file_directory_temp());
      if ($filename = $file_info->getName()) {
        $file->filename = $filename;
      }
      if ($filemime = $file_info->getType()) {
        $file->filemime = $filemime;
      }

      $file->status = FALSE;
      // With file uploads if the file already exists a copy is created and
      // stored as the unique URI. We can't copy the file since it exists on
      // another server.
      //
      // Instead we update an existing record if there is one.
      $fid = db_query("SELECT fid from {file_managed} WHERE uri = '" . check_plain($value) . "' LIMIT 1")->fetchField();
      if ($fid) {
        $file->fid = $fid;
      }

      $file = file_save($file);
    }
    catch (Exception $e) {
      form_set_error('url', $e->getMessage());
      return;
    }

    if (empty($file->fid)) {
      form_set_error($element, t('Unable to add file from URL %file.', array('%file' => $value)));
      return;
    }

    // Run all the normal validations, minus file size restrictions.
    if (isset($element['#upload_validators']['file_validate_size'])) {
      unset($element['#upload_validators']['file_validate_size']);
      unset($element['#upload_validators']['file_validate_extensions']);
    }

    if (dkan_filefield_sources_element_validate($element, (object) $file)) {
      $item = array_merge($item, (array) $file);
    }
  }
}

/**
 * Implements hook_entity_info_alter().
 */
function dkan_dataset_entity_info_alter(&$entity_info) {
  $entity_info['node']['view modes']['search_result'] = array(
      'label' => t('Search Result'),
      'custom settings' => TRUE,
  );
}

/**
 * Implements hook_preprocess_node().
 */
function dkan_dataset_preprocess_node(&$vars) {
  if($vars['view_mode'] == 'search_result') {
    $vars['theme_hook_suggestions'][] = 'node__search_result';
  }
}

/**
 * Return node modified date.
 */
function dkan_dataset_modified_date($node) {
  $node = ($node instanceof EntityMetadataWrapper) ? $node : entity_metadata_wrapper('node', $node);
  $harvested_date = $node->field_modified_source_date->value();

  // Harvested times can be dates or periods then we don't format them.
  return ($harvested_date) ? $harvested_date : format_date($node->changed->value(), 'custom', 'Y-m-d');
}

/**
 * Implements hook_field_formatter_info().
 */
function dkan_dataset_field_formatter_info() {
  return array(
    'harvest_info_formatter' => array(
      'label' => t('Harvest Info'),
      'field types' => array('entityreference'),
      'settings'  => array(),
    ),
  );
}

/**
 * Implements hook_field_formatter_view().
 */
function dkan_dataset_field_formatter_view($entity_type, $entity, $field, $instance, $langcode, $items, $display) {
  $element = array();
  $rows = array();

  if (isset($entity->field_harvest_source_ref) && isset($entity->field_harvest_source_ref[LANGUAGE_NONE][0]['target_id'])) {
    $harvest_node = node_load($entity->field_harvest_source_ref[LANGUAGE_NONE][0]['target_id']);
    if ($harvest_node) {
      $harvestSource = HarvestSource::getHarvestSourceFromNode($harvest_node);
      $migrate_table = 'migrate_map_dkan_harvest_migrate_' . $harvestSource->machine_name;
      $migrate_info = db_query("SELECT * FROM {$migrate_table} WHERE destid1=:destid1", array(':destid1' => $entity->nid))->fetchObject();
      $sourceid = isset($migrate_info->sourceid1) ? $migrate_info->sourceid1 : '';
      $rows[] = array(t('<strong>Harvest Source Title</strong>'), $harvest_node->title);
      $rows[] = array(t('<strong>Harvest Source URI</strong>'), l($harvest_node->field_dkan_harvest_source_uri[LANGUAGE_NONE][0]['value'], $harvest_node->field_dkan_harvest_source_uri[LANGUAGE_NONE][0]['value']));
      $rows[] = array(t('<strong>Last Harvest Performed</strong>'), dkan_harvest_get_last_event_log($harvest_node));
      $caption = t('Harvested from !link', array('!link' => l($harvest_node->title, 'node/' . $harvest_node->nid)));
      $markup = theme('table', array('rows' => $rows, 'caption' => $caption));
    }
  }
  else {
    $markup = '';
  }

  $element[0]['#markup'] = $markup;

  return $element;
}