# Aggregating on an Array of Objects

**URL:** <https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973>\
**Category:** Elasticsearch\
**Created:** [June 26, 2016, 4:53pm UTC](https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973 "2016-06-26T16:53:46Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![rolandino](https://avatars.discourse-cdn.com/v4/letter/r/e36b37/32.png) [@rolandino](https://discuss.elastic.co/u/rolandino)\
**Post date:** [June 26, 2016, 4:53pm UTC](https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973/1 "2016-06-26T16:53:46Z")

</div>

Wonder if anyone can help. I've got the following mapping:

```
curl -XPOST localhost:9200/test_index -d '{
    "mappings": {
        "test_type" : {
            "dynamic": "strict",
            "properties": {
                "id": {
                    "type": "string",
                    "index": "not_analyzed"
                },
                "source_data": {
                    "type": "nested",
                    "properties": {
                        "source": {"type": "string", "index": "not_analyzed"}
                    }
                }
            }
        }
    }
}'

```

Insert some sample content:

```
curl -XPUT 'http://localhost:9200/test_index/test_type/1' -d '{
    "source_data" : [{"source": "feed"}]
}'

curl -XPUT 'http://localhost:9200/test_index/test_type/2' -d '{
    "source_data" : [{"source": "supermarket"}]
}'

curl -XPUT 'http://localhost:9200/test_index/test_type/3' -d '{
    "source_data" : [{"source": "supermarket"}, {"source": "feed"}]
}'

```

So source\_data is an array of objects, currently each object only having one field: source.  
(in reality there are many fields, but for the purpose of this post, there's just one)

If I filter for those articles where source\_data.source contains both supermarket and feed, it works,  
and successfully identifies just the one article (article id '3'):

```
curl -XGET 'localhost:9200/test_index/_search' -d '
{
  "query": {
    "bool": {
      "filter": {
        "bool": {
          "must": [
            {
              "nested": {
                "path": "source_data",
                "query": {
                  "query_string": {
                    "query": "feed",
                    "default_field": "source_data.source"
                  }
                }
              }
            },
            {
              "nested": {
                "path": "source_data",
                "query": {
                  "query_string": {
                    "query": "supermarket",
                    "default_field": "source_data.source"
                  }
                }
              }
            }
          ]
        }
      }
    }
  }
}
'

```

However, I now want to run a filtered aggregation, to find the count of the  
number of source\_data's where the 'source' array contains 'feed' and/or 'supermarket':

```
curl -XGET 'localhost:9200/test_index/_search' -d '
{
  "query": {
    "match_all": {}
  },
  "aggs": {
    "source_data": {
      "aggs": {
        "filtered_agg": {
          "aggs": {
            "num_articles": {
              "terms": {
                "field": "source_data.source",
                "size": 0
              }
            }
          },
          "filters": {
            "filters": {
              "both_or": {
                "terms": {
                  "source_data.source": [
                    "feed",
                    "supermarket"
                  ]
                }
              },
              "both_and": {
                "query": {
                  "bool": {
                    "must": [
                      {
                        "query": {
                          "query_string": {
                            "query": "feed",
                            "default_field": "source_data.source"
                          }
                        }
                      },
                      {
                        "query": {
                          "query_string": {
                            "query": "supermarket",
                            "default_field": "source_data.source"
                          }
                        }
                      }
                    ]
                  }
                }
              }
            }
          }
        }
      },
      "nested": {
        "path": "source_data"
      }
    }
  },
  "size": 0
}
'

```

'feed or supermarket' works fine.

However, I cannot get 'feed AND supermarket' to work, despite the fact that I'm using the same filter structure as in the above filter.

Remember what I'm trying to count is the number of instances where the ARRAY source\_data is as follows: "source\_data" = [{"source": "feed"}, {"source": "supermarket"}] (ordering does not matter).

Can anyone help?

I think what the filtered agg is doing (though I'm not completely sure) is trying to find matches where the _string_ source\_data.source contains "feed" and "supermarket" - of which there aren't any ....

---

<div class="post-metadata">

**Author:** ![rolandino](https://avatars.discourse-cdn.com/v4/letter/r/e36b37/32.png) [@rolandino](https://discuss.elastic.co/u/rolandino)\
**Post date:** [June 26, 2016, 10:17pm UTC](https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973/2 "2016-06-26T22:17:37Z")

</div>

I do wonder whether what I want to do is actually possible given this mapping.  
Wonder whether I need to change the mapping at all ....  
Any help would be most welcome!

---

<div class="post-metadata">

**Author:** ![rolandino](https://avatars.discourse-cdn.com/v4/letter/r/e36b37/32.png) [@rolandino](https://discuss.elastic.co/u/rolandino)\
**Post date:** [June 29, 2016, 7:46am UTC](https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973/3 "2016-06-29T07:46:28Z")

</div>

I have come to the conclusion that it is not.  
Reason being is that there will never be a bucket of "feed" and "supermarket".  
Sadly.

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\
**Post date:** [July 5, 2017, 10:39pm UTC](https://discuss.elastic.co/t/aggregating-on-an-array-of-objects/53973/4 "2017-07-05T22:39:43Z")

</div>