Why does adding keyword make my query not work?-CodePudding

Im making a query that can filter out content based on certain fields and I had this.

{
  "from" : 0,
  "size" : 10,
  "query" : {
    "bool" : {
      "filter" : [
        {
          "bool" : {
            "must_not" : [
              {
                "terms" : {
                  "metadata.subtype" : [
                    "Type1"
                  ],
                  "boost" : 1.0
                }
              }
            ],
          }
        }
      ],
    }
  },
}

With a object like this

{
  id: "SomeId" <- Text
    metadata: {
      subtype: "Type1" <- Keyword
    }
}

And that works fine for that one field, but if instead of metadata.subtype I used id it no longer works unless I make it id.keyword.

But then metadata.subtype.keyword does not work and I dont understand why. Can you not use .keyword with fields that are indexed as keywords?

CodePudding user response：

terms query works with fields of keyword type, so it won't match your id field if it is text.

If you really want to filter against text field, use another query, e.g. match. But if you want to actually use terms query, you need to change the type of id field to keyword.

Another common approach is to define a field of text type, but add a keyword subfield so that it can also be matched by term and terms queries.

See: https://www.elastic.co/guide/en/elasticsearch/reference/current/multi-fields.html

Here's an example index mapping that would probably match your needs.

{
  "mapping": {
    "properties": {
      "id": {
        "type": "text",
        "fields": {
          "raw": {
            "type": "keyword"
          }
        }
      },
      "metadata.subtype": {
        "type": "keyword"
      }
    }
  }
}

Then you can filter with any of the following queries:

{ "match": { "id": "text" } }
{ "terms": { "id.raw": "text" } }
{ "terms": { "metadata.subtype": "value" } }