How to retrieve more data from aggregation when array of items is aggregated?

Question

This question is related to How to retrieve more data in aggregation?. The difference is that I try to aggregate on field "languages" which is an array and contains many entries like this: { id: 1, name: "English" }.

Here is a mapping:

languages:
                properties:
                    name:
                        type: "string"
                        index: "not_analyzed"
                    id:
                        type: "long"

I try to aggregate like this:

aggs:
    languages:
        terms:
           field: "languages.id"
           aggs:
              languageName:
                 terms:
                    field: "languages.name"

The result is:

{
  "key": 175,
  "doc_count": 1,
  "languageName": {
    "buckets": [
      {
        "key": "Latin",
        "doc_count": 1
      },
      {
        "key": "Lingala",
        "doc_count": 1
      },
      {
        "key": "Quechua",
        "doc_count": 1
      },
      {
        "key": "Tamil",
        "doc_count": 1
      },
      {
        "key": "Walloon",
        "doc_count": 1
      }
    ]
  }
}

For some reason nested aggregation returns all language names for single language id... How I can retrieve correct language name?

Answer 1

You have to use nested aggregation as,

POST _search
{
   "size": 0,
   "aggs": {
      "nestedlang": {
         "nested": {
            "path": "languages"
         },
         "aggs": {
            "languages": {
               "term": {
                  "field": "languages.id"
               },
               "aggs": {
                  "languageName": {
                     "terms": {
                        "field": "languages.name"
                     }
                  }
               }
            }
         }
      }
   }
}

Make sure you have type = nested for "languages"..

eg.

{
    ...

    "languages" : {
        "properties" : {
            "resellers" : { 
                "type" : "nested"
                "properties" : {
                    "name" : { "type" : "string" },
                    "id" : { "type" : "long" }
                }
            }
        }
    }
} 

Learn more @elasticserach

Answer 2

It is easier if you provide mapping, documents and query in json format so it is easier to copy paste and reproduce. Are you really using nested objects in the mapping? If that is the case you also need the nested aggregation: http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/search-aggregations-bucket-nested-aggregation.html

If you are not using nested objects, look up the documentation for inner objects. In the end you could en up with a document that is actually the following:

languages.id=[1,2,3,4]
languages.name=["Latin","Lingala","Dutch","English"]

Hope that helps