Elastic script_score, only count specific array elements

Question

I have a bunch of objects like this one in Elastic:

{
          "id" : "b397ab9c-2bab-4cce-a419-ba6b23545772",
          "name": "Alice",
          // other fields omitted
          "foo" : [
            {
              "id" : "1",
              "bar" : 20
            },
            {
              "id" : "2",
              "bar" : 20
            }
           ]
},
{
          "id" : "826784cb-7dcd-4d2c-b444-a466032a7b06",
          "name": "Bob",
          // other fields omitted
          "foo" : [
            {
              "id" : "1",
              "bar" : 15
            }
           ]
}

I am trying to make a score based on any elements in foo of which the id is 2. So in the above objects, Alice has a score of 20 and Bob has a score of 0 (as there's no element with id 2 in his foo array. I'm however a bit stuck in how to have my script_score look for the specific element. Here is my query:

  "query": {
    "function_score": {
      "score_mode": "max", 
      "functions": [
        {
          "filter": {
            "term": {
              "foo.id": {
                "value": "2"
              }
            }
          },
          "script_score": {
            "script": {
              "source": "doc['foo'].values /* what goes here? */",
              "params": {
              }
            }
          }
        }
      ]
    }
  }

No, the foo key is in the root of the object. It is an array that contains simple objects. — yesman
– yesman, Commented Nov 30, 2020 at 14:55

Jozef - Spatialized.io · Accepted Answer · 2020-11-30 16:56:42Z

1

When foo is not of type nested, you'll have lost the connection between id and bar due to value flattenning.

After you've reconfigured your mapping to be nested, you can do

{
  "query": {
    "nested": {
      "path": "foo",
      "query": {
        "function_score": {
          "score_mode": "max",
          "functions": [
            {
              "script_score": {
                "script": {
                  "source": "if (doc['foo.id.keyword'].value == '2') { def bar = doc['foo.bar'].value; return bar } return 0",
                  "params": {}
                }
              }
            }
          ]
        }
      }
    }
  }
}

but without the filter because that'd have filtered out doc#2 since it does not meet that condition and you probably want to do some sort of a scripted fallback.

Do note that once you're in the nested context, you don't have access to other object groups in your foo array. In other words, when you're in id: 2, you don't see the id: 1 group. More info here.

answered Nov 30, 2020 at 16:56

Jozef - Spatialized.io

17k4 gold badges29 silver badges79 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

yesman Over a year ago

Thanks Joe, I'll try reconfiguring my mapping tomorrow.

yesman Over a year ago

Got it working, thanks. I'll buy you a coffee tonight!

Alvaromon Over a year ago

coming here from the other question posted by @yesman and answered by joe im just here because i actually found a nice interaction in the comments

Jozef - Spatialized.io Over a year ago

Cheers @Alvaromon! Which question though?

Collectives™ on Stack Overflow

Elastic script_score, only count specific array elements

1 Answer 1

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related