uofm_maintenance.module

<?php
/**
 * @file
 * Contains hooks and functions
 */

/**
 * Fix tiffs queue
 */
define('UOFM_MAINTENANCE_FIX_TIFFS_QUEUE', 'uofm_maintenance_fix_tiff_mimetype');
define('UOFM_MAINTENANCE_FIX_TIFFS_TIMER', UOFM_MAINTENANCE_FIX_TIFFS_QUEUE);

/**
 * Regenerate JP2 datastream queue and timer
 */
define('UOFM_MAINTENANCE_REGEN_JP2_QUEUE', 'uofm_maintenance_regenerate_jp2');
define('UOFM_MAINTENANCE_REGEN_JP2_TIMER', UOFM_MAINTENANCE_REGEN_JP2_QUEUE);

/**
 * Batch index queue
 */
define('UOFM_MAINTENANCE_BATCH_INDEX_QUEUE', 'uofm_maintenance_batch_reindex');

/**
 * Derivative Fixer queue and timer.
 */
define('UOFM_MAINTENANCE_DERIVATIVE_FIXER_QUEUE', 'uofm_maintenance_derivative_fixer');
define('UOFM_MAINTENANCE_DERIVATIVE_FIXER_TIMER', UOFM_MAINTENANCE_DERIVATIVE_FIXER_QUEUE);

/**
 * Update parent queue.
 */
define('UOFM_MAINTENANCE_UPDATE_PARENT_QUEUE', 'uofm_maintenance_update_parent');

/**
 * Propogate XACML.
 */
define('UOFM_MAINTENANCE_COPY_XACML_QUEUE', 'uofm_maintenance_copy_xacml');

/**
 * Correct any incorrect image/tiff mimetypes.
 *
 * @param string|AbstractObject $object
 *   the string PID or a Fedora Object
 */
function uofm_maintenance_fix_tiff_worker($object) {
  $obj = FALSE;
  if (is_string($object)) {
    module_load_include('inc', 'islandora', 'includes/utilities');
    if (islandora_is_valid_pid($object)) {
      $obj = islandora_object_load($object);
    }
  }
  if (is_a($object, 'AbstractObject')) {
    $obj = $object;
  }
  if (isset($obj['OBJ']) && $obj['OBJ']->mimetype == "image/tif") {
    $obj['OBJ']->mimetype = "image/tiff";
  }
  if (isset($obj['TIFF']) && $obj['TIFF']->mimetype == "image/tif") {
    $obj['TIFF']->mimetype = "image/tiff";
  }
  return TRUE;
}

/**
 * Regenerate JP2 derivative from OBJ/TIFF.
 *
 * @param string|AbstractObject $object
 *   the string PID or a Fedora Object
 */
function uofm_maintenance_regen_jp2_worker($object) {
  $obj = FALSE;
  if (is_string($object)) {
    module_load_include('inc', 'islandora', 'includes/utilities');
    if (islandora_is_valid_pid($object)) {
      $obj = islandora_object_load($object);
    }
  }
  if (is_a($object, 'AbstractObject')) {
    $obj = $object;
  }
  if (isset($obj['OBJ']) || isset($obj['TIFF'])) {
    if ((isset($obj['OBJ']) && isset($obj['JP2']) && ($obj['OBJ']->size * 0.001) > $obj['JP2']->size) ||
      (isset($obj['TIFF']) && isset($obj['JP2']) && ($obj['TIFF']->size * 0.001) > $obj['JP2']->size)) {
      module_load_include('inc', 'islandora_large_image', 'includes/derivatives');
      drush_log("Regenerate JP2 for " . $obj->id, "ok");
      islandora_large_image_create_jp2_derivative($obj);
    }
  }
  return TRUE;
}

/**
 * Custom queue generation for derivative fixer.
 *
 * @var $pid string
 *   An object pid to process
 * @return array
 *   An array for the derivative fixer worker.
 */
function uofm_maintenance_derivative_fixer_queue($pid) {
  $force = drush_get_option('force', NULL);
  $item = array(
    'pid' => $pid,
    'force' => (bool) (!is_null($force)),
  );
  if (!is_null(drush_get_option('source_dsid', NULL))) {
    $item['source_dsid'] = drush_get_option('source_dsid', '');
  }
  if (!is_null(drush_get_option('destination_dsid', NULL))) {
    $item['destination_dsid'] = drush_get_option('destination_dsid', '');
  }
  return $item;
}

/**
 * Generate any missing derivatives.
 *
 * @param array $data
 *   array with keys 'pid' and 'force'
 */
function uofm_maintenance_derivative_fixer_worker($data) {
  $obj = FALSE;
  if (array_key_exists('pid', $data) && is_string($data['pid'])) {
    module_load_include('inc', 'islandora', 'includes/utilities');
    if (islandora_is_valid_pid($data['pid'])) {
      $obj = islandora_object_load($data['pid']);
    }
  }
  if ($obj) {
    module_load_include('inc', 'islandora', 'includes/derivatives');
    $options = array_diff_key($data, array('pid' => 1));
    if (!isset($options['force']) || (!is_bool($options['force']))) {
      $options['force'] = FALSE;
    }
    $results = islandora_do_derivatives($obj, $options);
    islandora_derivative_logging($results);
    return TRUE;
  }
  return FALSE;
}

/**
 * Custome queue item builder.
 */
function uofm_maintenance_update_parent_custom_queue($pid) {
  $relation_uri = drush_get_option('parent_relationship_uri', 'info:fedora/fedora-system:def/relations-external#');
  $relation_pred = drush_get_option('parent_relationship_pred', 'isMemberOf');
  $old_parent = drush_get_option('old_parent', NULL);
  $new_parent = drush_get_option('new_parent', NULL);
  $update_xacml = (bool) (!is_null(drush_get_option('update_xacml', NULL)));
  return array(
    'relation_uri' => $relation_uri,
    'relation_pred' => $relation_pred,
    'old_parent' => $old_parent,
    'new_parent' => $new_parent,
    'update_xacml' => $update_xacml,
    'pid' => $pid,
  );
}

/**
 * Replace/add a parent to the object.
 *
 * @param array $data
 *   array with keys:
 *
 *   'pid' => pid to act on,
 *   'relation_uri' => relation URI to use,
 *   'relation_pred' => relation predicate to use,
 *   'old_parent' => parent to replace (if exists),
 *   'new_parent' => parent to add,
 *   'update_xacml' => copy POLICY from the new parent,
 */
function uofm_maintenance_update_parent_worker($data) {
  $key_template = array(
    'pid' => 'test',
    'relation_uri' => 'test',
    'relation_pred' => 'test',
    'new_parent' => 'test',
    'update_xacml' => TRUE,
  );
  if (count(array_intersect_key($data, $key_template)) != 5) {
    drush_log("Invalid array passed to the update_parent_worker", "error");
    return FALSE;
  }
  $pid = $data['pid'];
  if (is_string($pid)) {
    module_load_include('inc', 'islandora', 'includes/utilities');
    if (islandora_is_valid_pid($pid)) {
        $object = islandora_object_load($pid);
    }
  }
  if (!is_a($object, 'AbstractObject')) {
    drush_log("Did not load object properly in update_parent_worker", "error");
    return FALSE;
  }
  $new_parent = $data['new_parent'];
  if (!(strpos($new_parent, "info:fedora/") == 0)) {
    $new_parent = "info:fedora/$new_parent";
  }

  if (isset($data['old_parent'])) {
    $old_parent = $data['old_parent'];
    if (!(strpos($old_parent, "info:fedora/") == 0)) {
      $old_parent = "info:fedora/$old_parent";
    }
    if (count($object->relationships->get($data['relation_uri'], $data['relation_pred'])) > 0) {
      $object->relationships->remove($data['relation_uri'], $data['relation_pred'], $old_parent);
    }
  }
  $object->relationships->add($data['relation_uri'], $data['relation_pred'], $new_parent);
  $result = TRUE;
  if (module_exists('islandora_xacml_editor') && $data['update_xacml']) {
    $result = uofm_maintenance_update_xacml_worker(array('from' => $data['new_parent'], 'to' => $pid));
  }
  return $result;
}

/**
 * XACML queue custom builder.
 *
 * @param string $pid
 *   the PID for this action.
 */
function uofm_maintenance_copy_xacml_custom_queue($pid) {
  $from = drush_get_option('from_pid', NULL);
  if (is_null($from)) {
    drush_set_error("--from_pid must be set.");
  }
  return array('to' => $pid, 'from' => $from);
}

/**
 * Copies XACML policy from one object to another.
 *
 * @param array $data
 *   array with keys ('from' and 'to') and values as the objects or PIDs.
 * @return bool
 *   Was it successful.
 */
function uofm_maintenance_update_xacml_worker($data) {
  if (!module_exists('islandora_xacml_editor')) {
    return FALSE;
  }
  if (!is_array($data) || !array_key_exists('from', $data) || !array_key_exists('to', $data)) {
    return FALSE;
  }
  $from = $data['from'];
  $to = $data['to'];
  $loader = function($object) {
    if (is_string($object)) {
      module_load_include('inc', 'islandora', 'includes/utilities');
      if (islandora_is_valid_pid($object)) {
        return islandora_object_load($object);
      }
    }
    if (is_a($object, 'AbstractObject')) {
      return $object;
    }
  };
  $from_obj = $loader($from);
  if (!isset($from_obj['POLICY'])) {
    return FALSE;
  }
  module_load_include('inc', 'islandora_xacml_editor', 'includes/batch');
  $to_obj = $loader($to);
  $xml = $from_obj['POLICY']->content;
  $policy_update = new IslandoraUpdatePolicy($to_obj->id, $xml);
  $success = $policy_update->updatePolicy();
  return $success;
}

/**
 *
 * Fix dates for faceting.
 *
 * Copy full mods:subject/mods:temporal dates, if YYY? make YYY5
 *
 * @param $pid string
 *   The pid of the object to process
 *
 * return boolean
 *   Did it succeed?
 */
function uofm_maintenance_fix_dates_worker($pid) {
  if (is_string($pid)) {
    module_load_include('inc', 'islandora', 'includes/utilities');
    if (islandora_is_valid_pid($pid)) {
      $object = islandora_object_load($pid);
      if (isset($object['MODS'])) {
        $mods = new DOMDocument();
        try {
          $mods->loadXML($object['MODS']->content);
        }
        catch (Exception $e) {
          drush_log(t("Error loading MODS for object @p: @m", array('@p' => $pid, '@m' => $e->getMessage())), 'warning');
          return FALSE;
        }
        $xpath = new DOMXpath($mods);
        $xpath->registerNamespace('mods', 'http://www.loc.gov/mods/v3');
        $mods_top = $xpath->query('/mods:mods');
        $mods_top = $mods_top->item(0);
        $dates = $xpath->query('//mods:subject/mods:temporal');
        $adds = array();
        $today = new DateTime();
        foreach ($dates as $date) {
          $text = $date->textContent;
          $date_string = FALSE;
          // Match the date, if it is only to the decade make it a 5.
          // If the full date is there, use it otherwise append -01-01.
          if (preg_match('/^([0-9]{2,3}\?{1,2}|[0-9]{4})([\-\/][0-9]{1,2})?([\-\/][0-9]{1,2})?/', $text, $matches)) {
            if (strpos($matches[1], '??') === FALSE) {
              $date_string = str_replace('?', 5, $matches[1]) . (isset($matches[2]) ? $matches[2] : '-01') . (isset($matches[3]) ? $matches[3] : '-01');
            }
          }
          // Match MM/DD/YYYY
          elseif (preg_match('/^([0-9\?]{1,2})\/([0-9\?]{1,2})\/([0-9\?]{4})$/', $text, $matches)) {
            if (strpos($matches[3], '??') === FALSE) {
              $date_string = str_replace('?', 5, $matches[3]) . "-" . (strpos($matches[1], '?') === FALSE ? $matches[1] : '01') . "-" . (strpos($matches[2], '?') === FALSE ? $matches[2] : '01');
            }
          }
          // Match ca.YYYY
          elseif (preg_match('/^ca\.?\s*([0-9]{2,3}\?{1,2}|[0-9]{4})/i', $text, $matches)) {
            if (strpos($matches[1], '??') === FALSE) {
              $date_string = str_replace('?', 5, $matches[1]) . "-01-01";
            }
          }
          // Must have at least one digit or skip it.
          else if (preg_match('/\d/', $text)){
            // Try to parse it to a DateTime.
            // Remove commas because of 11 January, 1929 format.
            $date_string = str_replace(',', '', $text);
          }
          if ($date_string) {
            try {
              $dt = new DateTime($date_string);
              if ($dt->format('Y-m-d') != $today->format('Y-m-d')) {
                $adds[] = '<dateCreated>' . $dt->format('Y-m-d') . '</dateCreated>';
              }
            }
            catch (Exception $e) {
              drush_log(t('Could not parse (@t) for object (@p): @e', array(
                '@p' => $pid,
                '@t' => $text,
                '@e' => $e->getMessage(),
              )), 'error');
            }
          }
        }
        // There is a change so update MODS and save back.
        if (count($adds) > 0) {
          $ext = $xpath->query('//mods:extension');
          $extension = '<extension xmlns="http://www.loc.gov/mods/v3">' . implode($adds, ' ') . '</extension>';
          $new_node = dom_import_simplexml(simplexml_load_string($extension));
          $new_node = $mods->importNode($new_node, TRUE);
          if ($ext->length > 0) {
            $mods_top->replaceChild($new_node, $ext->item(0));
          }
          else {
            $mods_top->appendChild($new_node);
          }
          $new_xml = $mods->saveXML();
          $object['MODS']->content = $new_xml;
        }
        else {
          drush_log(t("No date information to add for object @p", array('@p' => $pid)), "notice");
        }
        return TRUE;
      }
      else {
        drush_log(t("No MODS datastream for object @p", array('@p' => $pid)), "notice");
        return TRUE;
      }
    }
    else {
      drush_log(t("@p is not a valid pid", array('@p' => $pid)), "error");
      return FALSE;
    }
  }
  else {
    drush_log("Did not receive a string pid", "error");
    return FALSE;
  }
}