Get contacts present on Hrinfo but not on HID export files

hr_contacts.drush.inc

<?php

/**
 * @file
 * hr_contacts module drush integration.
 */

/**
 * Implements hook_drush_command().
 *
 * @return array
 *  An associative array describing your command(s).
 *
 * @see drush_parse_command()
 */
function hr_contacts_drush_command() {
  $items = array();

  $items['hr-contacts-get-non-hid'] = array(
    'description' => "Get the contacts in HRinfo but not in HID",
    'drupal dependencies' => array(),
    'aliases' => array(),
  );

  return $items;
}

/**
 * Callback to get the contacts that aren't in HID.
 */
function drush_hr_contacts_get_non_hid() {

  $query = new EntityFieldQuery();
  $results = $query->entityCondition('entity_type', 'node')
    ->entityCondition('bundle', 'hr_contact')
    ->execute();
  if (empty($results['node'])) {
    return;
  }
  $contacts = array_keys($results['node']);

  $profiles = db_query("SELECT uid, cid FROM {hid_profiles}")->fetchAllAssoc('uid');

  $global_hid_file_path = file_directory_temp() . '/global_hid_contacts.csv';
  $local_hid_file_path = file_directory_temp() . '/local_hid_contacts.csv';

  $batch = array(
    'operations' => array(
      array(
        'drush_hr_contacts_get_non_hid_operation',
        array(
          $contacts,
          $profiles,
          $global_hid_file_path,
          $local_hid_file_path,
        ),
      ),
    ),
    'finished' => 'drush_hr_contacts_get_non_hid_finished',
    'title' => t('Calculating'),
    'init_message' => t('Preparing contacts calculation...'),
    'progress_message' => t('Calculating contacts...'),
    'error_message' => t('Contacts process resulted in an error.'),
  );

  batch_set($batch);
  drush_backend_batch_process();
}

function drush_hr_contacts_get_non_hid_finished($success, $results, $operations) {
  drush_log('Process finished');
}

/**
 * Iterates the contacts and checks if they are in the HID files.
 */
function drush_hr_contacts_get_non_hid_operation($contacts, $profiles, $global_hid_file_path, $local_hid_file_path, &$context) {
  $limit = 150;
  $context['finished'] = 0;

  if (!isset($context['sandbox']['file'])) {
    $headers = array(
      'Contact id (Hrinfo)',
      'Contact url',
      'First Name',
      'Last Name',
      'Email',
    );
    // Create the file and print the labels in the header row.
    $file_path = file_directory_temp() . '/contacts-report.csv';
    $handle = fopen($file_path, 'w');
    fputcsv($handle, $headers);
    fclose($handle);
    $context['sandbox']['contacts'] = $contacts;
    $context['sandbox']['total_contacts'] = count($contacts);
    $context['sandbox']['file'] = $file_path;
    $context['results']['count'] = 0;
  }

  $handle = fopen($context['sandbox']['file'], 'a');

  $global_hid_handle = fopen($global_hid_file_path, 'r');
  $local_hid_handle = fopen($local_hid_file_path, 'r');
  if ($contacts_pending = count($context['sandbox']['contacts'])) {
    $actual_limit = min($contacts_pending, $limit);
    $contact_info = array_slice(
      $context['sandbox']['contacts'],
      $context['results']['count'],
      $actual_limit,
      TRUE
    );

    // Load all the contacts for the iteration.
    $contact_nodes = node_load_multiple($contact_info);

    foreach ($contact_nodes as $contact_node) {
      $wrapper = entity_metadata_wrapper('node', $contact_node);
      $email = $wrapper->field_email->value();

      // We might not have the email of the contact.
      if (!empty($email)) {
        // Step 1. If the contact is present in the profiles table, skip it.
        if (drush_hr_contacts_email_is_in_profile($email, $profiles)) {
          continue;
        }

        // Step 2. If the contact is present in the global list, skip it.
        if (drush_hr_contacts_email_is_in_file($email, $global_hid_handle)) {
          continue;
        }

        // Step 3. If the contact is present in the local list, skip it.
        if (drush_hr_contacts_email_is_in_file($email, $local_hid_handle)) {
          continue;
        }
      }

      // Otherwise, add it to the csv.
      $row = array();
      $row[] = $wrapper->getIdentifier();
      $row[] = url('node/' . $wrapper->getIdentifier());
      $row[] = $wrapper->field_first_name->value();
      $row[] = $wrapper->field_last_name->value();
      $row[] = $email;
      fputcsv($handle, $row);

      rewind($global_hid_handle);
      rewind($local_hid_handle);
    }


    $context['results']['count'] += $actual_limit;
    $context['finished'] = $context['results']['count'] / $context['sandbox']['total_contacts'];
  }
  else {
    $context['finished'] = 1;
  }

  fclose($handle);
  fclose($global_hid_handle);
  fclose($local_hid_handle);

  $context['message'] = t(
    'Processed @count of @total contacts.',
    array(
      '@count' => $context['results']['count'],
      '@total' => $context['sandbox']['total_contacts'],
    )
  );
}

/**
 * Checks the email in the profiles.
 */
function drush_hr_contacts_email_is_in_profile($email, $profiles) {
  foreach ($profiles as $profile) {
    if (strstr($profile->cid, $email) !== FALSE) {
      return TRUE;
    }
  }

  return FALSE;
}

/**
 * Reads the file and looks for the email.
 */
function drush_hr_contacts_email_is_in_file($email, $handle) {
  while ($row = fgetcsv($handle)) {
    if (in_array($email, $row)) {
//      $meta_data = stream_get_meta_data($handle);
//      $filename = $meta_data['uri'];
//      drush_log("Found $email in $filename");
      return TRUE;
    }
  }

  return FALSE;
}