islandora_fpr.module

<?php

/**
 * @file
 * Defines all the hooks this module implements, and some other functions.
 */

/**
 * Implements hook_menu().
 */
function islandora_fpr_menu() {
  $items = array();
  $items['admin/islandora/tools/fpr'] = array(
    'title' => 'Islandora FPR',
    'description' => 'Configure Islandora FPR.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('islandora_fpr_admin_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['islandora/object/%islandora_object/manage/fpr'] = array(
    'title' => 'FPR',
    'weight' => '20',
    'page callback' => 'islandora_fpr_get_fpr_data',
    'page arguments' => array(2),
    'access callback' => 'islandora_fits_metadata_access',
    'access arguments' => array(2),
    'type' => MENU_LOCAL_TASK,
  );
  return $items;
}

/**
 * Admin settings form builder.
 */
function islandora_fpr_admin_settings() {
  $form['islandora_fpr_fpr_endpoint'] = array(
    '#title' => t('FPR endpoint URL'),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_fpr_fpr_endpoint', 'https://fpr.example.com/fpr/api/v2/'),
    '#description' => t("URL of the Archivematica Format Policy Registry endpoint."),
    '#maxlength' => 255,
  );
  $form['islandora_fpr_http_timeout'] = array(
    '#title' => t('Timeout'),
    '#type' => 'textfield',
    '#size' => 5,
    '#default_value' => variable_get('islandora_fpr_http_timeout', '2'),
    '#description' => t("Number of seconds to use as the timeout for HTTP
      requests to the FPR endpoint."),
    '#maxlength' => 5,
  );
  return system_settings_form($form);
}

/**
 * Fetches data from the remote instance of the FPR.
 *
 * @param AbstractObject $islandora_object
 *   The Islandora object.
 *
 * return string
 *   A string containing the data to display.
 */
function islandora_fpr_get_fpr_data(AbstractObject $islandora_object) {
  $dsid = variable_get('islandora_fits_techmd_dsid', 'TECHMD');
  if (!$islandora_object[$dsid]) {
    return t("FITS data not available.");
  }
  else {
    $header = array();
    $header[] = array("data" => t("PUID"));
    $header[] = array("data" => t("Description"));
    $header[] = array("data" => t("Access command"));
    $header[] = array("data" => t("Preservation command"));
    $row = array();

    $puid = islandora_fpr_get_puid($islandora_object[$dsid]->content);
    if (!$puid) {
      return t("No PUID is available in the FITS output, skipping the FPR.");
    }

    $row[] = l($puid, 'http://apps.nationalarchives.gov.uk/pronom/' . $puid,
      array('attributes' => array('target' => '_blank')));
    $endpoint = variable_get('islandora_fpr_fpr_endpoint', 'https://fpr.example.com/fpr/api/v2/');
    $timeout = (int) variable_get('islandora_fpr_http_timeout', '2');

    // Format info request.
    $format_request = $endpoint . 'format-version/?pronom_id=' . $puid . '&format=json';
    $format_response = drupal_http_request($format_request, array('timeout' => $timeout));
    // Check for timeout. We only check here, not on every request.
    if ($format_response->code === HTTP_REQUEST_TIMEOUT) {
      return t("Can't reach FPR server at @endpoint", array('@endpoint' => $endpoint));
    }
    $format_response_data = json_decode($format_response->data);
    if ($format_response_data->meta->total_count === 0) {
      return t("Sorry, the format @format isn't in the FPR.", array('@format' => $puid));
    }
    $format_uuid = $format_response_data->objects[0]->uuid;
    $format_description = $format_response_data->objects[0]->description;
    $row[] = $format_description;

    // Access rule request.
    $access_rule_request = $endpoint .
      'fp-rule/?purpose=access&format=json&fmt=' . $format_uuid;
    $access_rule_response = drupal_http_request($access_rule_request);
    $access_rule_response_data = json_decode($access_rule_response->data);
    if ($access_rule_response_data->meta->total_count === 0) {
      $row[] = t("n/a (no access rule for format @puid in the FPR)",
        array('@puid' => $puid));
    }
    else {
      // Access command request.
      $command_parts = explode('/', $access_rule_response_data->objects[0]->command);
      $access_command_uuid = $command_parts[5];
      $access_command_request = $endpoint . 'fp-command/?format=json&uuid=' .
        $access_command_uuid;
      $access_command_response = drupal_http_request($access_command_request);
      $access_command_response_data = json_decode($access_command_response->data);
      if ($access_command_response_data->meta->total_count === 0) {
        $row[] = t("n/a");
      }
      else {
        $row[] = $access_command_response_data->objects[0]->command;
        islandora_fpr_add_datastream($islandora_object, 'FPR_ACCESS_CMD', $access_command_response_data->objects[0]->command);
      }
    }

    // Preservation rule request.
    $preservation_rule_request = $endpoint .
      'fp-rule/?purpose=preservation&format=json&fmt=' . $format_uuid;
    $preservation_rule_response = drupal_http_request($preservation_rule_request);
    $preservation_rule_response_data = json_decode($preservation_rule_response->data);
    if ($preservation_rule_response_data->meta->total_count === 0) {
      $row[] = t("n/a (no preservation rule for format @puid in the FPR)",
        array('@puid' => $puid));
    }
    else {
      // Preservation command request.
      $command_parts = explode('/', $preservation_rule_response_data->objects[0]->command);
      $preservation_rule_uuid = $command_parts[5];
      $preservation_command_request = $endpoint .
        'fp-command/?format=json&uuid=' . $preservation_rule_uuid;
      $preservation_command_response = drupal_http_request($preservation_command_request);
      $preservation_command_response_data = json_decode($preservation_command_response->data);
      if ($preservation_command_response_data->meta->total_count == '0') {
        $row[] = t("n/a");
      }
      else {
        $row[] = $preservation_command_response_data->objects[0]->command;
        islandora_fpr_add_datastream($islandora_object, 'FPR_PRESERVATION_CMD', $preservation_command_response_data->objects[0]->command);
      }
    }

    $rows[] = $row;
    return theme('table', array('header' => $header, 'rows' => $rows));;
  }
}

/**
 * Adds a datastream to an object.
 *
 * @param AbstractObject $object
 *   The object to add a datastream to.
 * @param string $dsid
 *   The datastream ID to be added.
 * @param string $command
 *   A string containing the command
 *
 * @return bool|string
 *   TRUE if the datastream was added/modified successfully, the error message
 *   otherwise.
 */
function islandora_fpr_add_datastream(AbstractObject $object, $dsid, $command) {
  try {
    $command = preg_replace("/\r|\n/", "", $command);
    $ingest = !isset($object[$dsid]);
    if ($ingest) {
      $ds = $object->constructDatastream($dsid, 'M');
      $ds->label = $dsid;
      $ds->mimeType = 'text/plain';
    }
    else {
      $ds = $object[$dsid];
    }
    $ds->setContentFromString($command);
    if ($ingest) {
      $object->ingestDatastream($ds);
    }
    return TRUE;
  }
  catch (exception $e) {
    $variables = array(
      '@dsid' => $dsid,
      '@ret' => $e->getTraceAsString(),
    );
    watchdog('islandora_fpr', 'Failed to add @dsid datastream.<br/>Error: @ret', $variables, WATCHDOG_ERROR);
    $message = $e->getMessage();
    return $message;
  }
}

/**
 * Parses an object's FITS datastream to get the PUID.
 *
 * @param string $fits_xml
 *   The content of the FITS datastream.
 *
 * return bool|string
 *   The PRONOM ID extracted from the FITS XML. Returns
 *   FALSE if no 'identification' element was found or
 *   if a conflict was explicitly found.
 */
function islandora_fpr_get_puid($fits_xml) {
  $xml = simplexml_load_string($fits_xml);
  $xml->registerXPathNamespace('f', 'http://hul.harvard.edu/ois/xml/ns/fits/fits_output');
  $expression = "/f:fits/f:identification/f:identity/f:externalIdentifier[@type='puid']";
  $puids = $xml->xpath($expression);
  if (count($puids)) {
    // There should be only one.
    return (string) $puids[0];
  }
  else {
    // We probably have a format identification conflict, or FITS has
    // not been able to deterime the PUID and has not added an 'identification'
    // element.
    return FALSE;
  }
}