ryanhouston / elasticsearch-documents Goto Github PK

It would be nice to have a directory with a simple set of model classes, associated Documents, and a couple Queryable classes to show a quick overview of real usage.

Client code may need to pass options in to the query such as 'size' and 'from' in order to paginate results. Queryable#execute should take these options and merge them into the hash that ultimately gets passed through Elasticsearch::Client.search.

Currently Queryable classes define a #to_hash method that requires the index and body hash keys are present:

def to_hash
  {
    index: 'test_index',
    body: {
      query: {
        query_string: {
          analyzer: "snowball",
          query:    "something to search for"
        }
      }
    }
  }
end

The index name is already passed to the configuration so this should not be required here.

There are many options that can be passed to the elasticseach client search method. These options are also currently set in the hash provided by #as_hash. Perhaps there is a better way to abstract the query options so #as_hash becomes focused only on serializing the Query DSL to be assigned to the body attribute.

Memoizing the Elasticsearch::Client instance on a module level is not threadsafe and could be an issue for multi-threaded applications.

The current specs mock and stub all actual interactions with Elasticsearch::Transport::Client. This means we never get any real validation that the requests we are constructing are valid.