How to compound terms query and bool query together in Elasticsearch

Question

Originally, I would add a filter in the bool query. However, when I turn to terms filter, the documents indicates that it should be replaced by terms query now. So I understand this as that we need to construct a compound query with both terms query and bool query. If I am correct, how should I write the query?

Note: I use elasticsearch's Python API.

Answer 1

There are two types of context now: query context and filter context. Any query clause inside query context contribute to the score of matching document i.e. how well a document matched a query whereas any query clause inside filter context determines whether or not the document matches and doesn't contribute to the score.

So in the query below I have marked both the contexts:

{
  "query": { 
    "bool": { 
      "must": [
        { "match": { "field1":   "terms1"}},              <------query context
      ],
      "filter": [ 
        { "terms":  { "someId": [3242, 12343, 1234] }},   <-------inside filter block every query clause will act as filter (filter context)
      ]
    }
  }
}

Therefore in order to use terms query as a filter you need to follow the below steps:

1. Create a bool query say boolQuery
2. Create a terms query sat termsQuery
3. Add termsQuery to filter of boolQuery.
4. Set boolQuery as query.

This translates to below:

{
  "query": {
    "bool": {
      "filter": [
        {
          "terms": {
            "field1": ["terms1", "term2", "terms3"]
          }
        }
      ]
    }
  }
}