Elastic Painless Scripted Field On Null/Missing Value

This post shows how to use elastic painless language in scripted field to work on documents’ keys which might not exist in some documents.

Parsing analyzed field in Painless

Suppose we have following 2 documents in elastic:

[{
    "kye1": "value1",
    "key2": {
        "key22": "value22"
    }
}, {
    "key1": "valuex"
}]

The key key22 in the first document can be accessed by doc['key2.key22'].value. If we use this script in the scripted field, we will see a null value for all the documents. This is because the second document doesn’t have the key key22, painless language will throw an error. This github issue is discussing how to return a default value if it is missing.

To workaround this, I found a solution from this github issue. We should check the null value each time.

The script should be:

(params._source.key2 == null) ? '' : ( (params._source.key2.key22 == null) ? '' : (params._source.key2.key22) ))

Parsing documents by params._source is very slow. It’s not cached, and is calculated in real-time each time.

The fields calculated by the scripted field is not searchable.