tombstone

#!/usr/bin/env php
<?php

/**
 * Script for creating tombstone entries for Bags in ElasticSearch.
 *
 * Run './tombstone --help' for usage.
 */

require 'vendor/autoload.php';
use GuzzleHttp\Client;
use GuzzleHttp\Psr7\Request;
use GuzzleHttp\Exception\ClientException;
use GuzzleHttp\Psr7;
use \wapmorgan\UnifiedArchive\UnifiedArchive;

$cmd = new Commando\Command();
$cmd->option('p')
  ->aka('path')
  ->describedAs('Absolute or relative path to the Bag filename to create the tombstone for. Use either this option or --id.')
  ->default('');
$cmd->option('e')
  ->aka('elasticsearch_url')
  ->describedAs('URL (including port number) of your Elasticsearch endpoint. Default is "http://localhost:9200".')
  ->default('http://localhost:9200');
$cmd->option('x')
  ->aka('elasticsearch_index')
  ->describedAs('Elasticsearch index. Default is "bags".')
  ->default('bags');
$cmd->option('i')
  ->aka('id')
  ->describedAs('The ID of the bag to create the tombstone for. Use either this option or --path.');

if (strlen($cmd['p'])) {
    $bag_id = get_id_from_path($cmd['e'], $cmd['x'], $cmd['p']);
}
else {
    $bag_id = $cmd['id'];
}


if (document_exists($cmd['elasticsearch_url'], $cmd['elasticsearch_index'], $bag_id)) {
    $client = new Client([
      'base_uri' => $cmd['elasticsearch_url'],
      ]);
    try {
      $update['doc'] = array('tombstone' => true, 'document_timestamp' => gmdate("Y-m-d\TH:i:s\Z")); 
      $response = $client->post($cmd['elasticsearch_index'] . '/bag/' . $bag_id . '/_update', ['json' => $update]);
    }
    catch (ClientException $e) {
      print Psr7\str($e->getRequest());
      print Psr7\str($e->getResponse());
    }
}
else {
  print "Bag " . $bag_id . " not found in " . $cmd['elasticsearch_url'] . '/' . $cmd['elasticsearch_index'] . PHP_EOL;
  exit;
}

print "Tombstone for bag " . $bag_id . " added to " . $cmd['elasticsearch_url'] . '/' . $cmd['elasticsearch_index'] . PHP_EOL;

/**
 * See if the document exists.
 *
 * @param string $base_url
 *   The Elasticsearch server's URL.
 * @param string $bag_id
 *   The ID of the bag in the Elasticsearch index.
 *
 * @return bool
 *   True if the index exists, false if not.
 */
function document_exists($base_url, $index, $bag_id) {
    $client = new Client(['base_uri' => $base_url]);
    try {
      $document_exists_response = $client->request('GET', $index . '/bag' . '/' . $bag_id);
    }
    catch (ClientException $e) {
      if ($e->getResponse()->getStatusCode() == 404) {
        return false;
      }
    }

    return true;
}

/**
 * Given a bag's path, get its ID in the index.
 *
 * @param string $base_url
 *   The Elasticsearch server's URL.
 * @param string $bag_id
 *   The ID of the bag in the Elasticsearch index.
 * @param string $bag_path
 *   The absolute path of the bag in the Elasticsearch index.
 *
 * @return string
 *   The bag's ID.
 */
function get_id_from_path($base_url, $index, $bag_path) {
    $client = new Client(['base_uri' => $base_url]);
    try {
      $query = array('query' => array('term' => array('bag_location_exact' => $bag_path)));
      $response = $client->request('GET', '/'. $index . '/bag' . '/_search', ['json' => $query]);
      $result = json_decode($response->getBody());
      if ($result->hits->total > 0) {
      return $result->hits->hits[0]->_id;
      }
      else {
          return false;
      }
    }
    catch (ClientException $e) {
      if ($e->getResponse()->getStatusCode() == 404) {
        return false;
      }
    }

    return true;
}