How to use the JavaScript Elasticsearch client errors?

Question

I'm trying to parse the error from the Elasticsearch client called through a nodeJs express routes. I my case it is the timeout error I would like to identify.

I more or less implement something like that:

var client = new elasticsearch.Client( {
  host: process.env.ELASTICSEARCHLOCATION,
  requestTimeout:500,
  maxRetries:1
});

function returnResponse( response ) {
  return function ( resp ) {
    //stringExtract 
    if ( resp.hits && resp.hits.hits ) {
      response.status( 200 ).send( resp.hits.hits );
    }else{
      response.status( 200 ).send( [] );
    }
  };
}

function handleError( response ) {
  return function( err ){
    if ( err.statusCode == 404 ) {
      response.status( 200 ).send( [] );
    } else {
      console.error( err.message );
      response.status( err.statusCode ).send( err );
    }
  }
}
router.get( "/test", function ( req, res, next ) {
  log.data( "route GET: ", "/wynsureSearch/test" );
  log.request( req.url );

  client .search( {
    index: process.env.ELASTICSEARCHINDEX, //target the wynsure version aka the ES DB index
    type: [], //target the wynsure types
    body: req.body
  }).then( returnResponse( res ), handleError( res ) );
})   

All works fine when there are no errors. But when a timeout error occurs, the breakpoint in the function returned by handleError() shows me this:

I read from elastic.co errors documentation that there are standars errors. How I have my function match matching the errors from documentation? I would like to be able to match RequestTimeout or InternalServerError for example.

Answer 1

res parameter passed in returnResponse and handleError refers to the parameter passed in router.get( "/test", function ( req, res, next ) { therefore it is undefined .

Quote from Elasticsearch

When a callback is passed to any of the API methods, it will be called with (err, response, status). If you prefer to use promises, don't pass a callback and a promise will be returned. The promise will either be resolved with the response body, or rejected with the error that occured (including any 300+ response for non "exists" methods).

client .search( {
  index: process.env.ELASTICSEARCHINDEX, //target the wynsure version aka the ES DB index
  type: [], //target the wynsure types
  body: req.body }, function callback(err, response, status){
  if (err) {
    if(status == 404) res.status(200).send();
    else console.error(err.message);
  }
  else res.status(200).send(response); // res variable is refering to HTTP response whereas response is the result returned from elasticsearch.
 });

Note: status codes in ElasticSearch have their own meaning, don't confuse them with HTTP status codes.

Note: Javascript supports enclosing scopes, not block scopes.

Answer 2

Concerning the undefined status code, there seems to be a bug with the status code definition in ElasticSearch.JS .

@Spalger offers a solution about how using the JavaScript Elasticsearch client errors?:

// require the Error class
var EsClientRequestTimeout = require('elasticsearch').errors.RequestTimeout;

// ... somewhere else in your code

client.search(..., function (err) {
  if (err instanceof EsClientRequestTimeout) {
    console.log('timeout')
  } else {
    console.log('not a timeout')
  }
})

Thank you a lot @Spalger for helping!