Multi-layer query + limiting results of a leaf query

[felipeecst (Felipe Teixeira)]

Hi there,

I am new to ElasticSearch and still not much familiar with the capabilities of the query language. Am I trying to perform a specific query and would appreciate some help on how to build it, even the answer is that I cannot achieve that using ElasticSearch.

I am storing documents with the following format on my ElasticSearch cluster:

{
      "id" : 996,
      "price" : 320,
      "company" : "Company A",
      "created_at" : "2020-11-21T17:02:15.707Z",
      "updated_at" : "2020-11-21T17:02:15.707Z",
      "departments" : [
        {
          "id" : 10,
          "name" : "Insurance"
        },
        {
          "id" : 129,
          "name" : "Information Technology and Services"
        }
      ],
      "source" : "database"
}

And these are my mappings:

      settings do
        mapping do
          indexes :company, type: :text
          indexes :price, type: :integer
          indexes :departments, type: :object do
            indexes :id, type: :keyword
          end
        end
      end

Given a search term and an array of department ids, I want to create a score function that achieves the following behavior:

1. First, return all the records where the search term is an exact match with the company field and the array of department ids has at least one match with departments.id fields.
2. After these, come all the records where the search term has an exact match with the company field.
3. In the sequence, return at maximum 5 elements with source = api.
4. At the end, come all the other records ordered by price (ascending).

I started with something like this:

{
        query: {
          bool: {
            must: [
              { match_all: {} }
            ],
            should: [
              { term: { 'company.keyword': { value: <searchterm> } }, boost: 4 },
              { terms: { 'departments.id': [<department_ids>], boost: 3 } },
              { term: { source: { value: 'api', boost: 2 } }
            ]
          }
        },
        sort: [
          '_score',
          { price: { order: 'desc' } }
        ]
      }

Is there a way to limite the results of a specific leaf query? Something like this:

           { term: { source: { value: 'api', boost: 2, size: 5 } }

Can someone please help me clarify the major misconceptions I am making? I'm sure there are several.

Also, any help to make my query come closer to the desired result is welcome.

Thanks.