I now title is long and confusing. Sorry for that. I have a book database in Elasticsearch 8.2.3 and records like:

{id:"1", "writer": "andy", "genre":"horror", "book_name":"first book of andy"}
{id:"2", "writer": "andy", "genre":"fantasy", "book_name":"second book of andy"}
{id:"3", "writer": "andy", "genre":"romance", "book_name":"thirth book of andy"}
{id:"4", "writer": "mike", "genre":"fantasy", "book_name":"first book of mike"}
{id:"5", "writer": "mike", "genre":"horror", "book_name":"second book of mike"}
{id:"6", "writer": "mike", "genre":"horror", "book_name":"thirth book of mike"}
{id:"7", "writer": "greg", "genre":"romance", "book_name":"first book of greg"}
{id:"8", "writer": "greg", "genre":"horror", "book_name":"second book of greg"}
{id:"9", "writer": "ursa", "genre":"romance", "book_name":"first book of ursa"}
{id:"10", "writer": "ursa", "genre":"fantasy", "book_name":"second book of ursa"}

I want a query that would return authors who have at least one book in both the "horror" and "fantasy" genres.

{"writer": "andy"}
{"writer": "mike"}

Also it would be great if it returns also the total number of books in these genres by these authors.

{"writer": "andy", "book_count": 2}
{"writer": "mike", "book_count": 3}

I can do it with multiple query and some coding with python but it is slow. If there is a way to do it with elsticsearch query I want to use it.

CodePudding user response：

You can use the below query to get expected results

{
    "size": 0,
    "query": {
        "bool": {
            "should": [
                {
                    "term": {
                        "genre.keyword": "horror"
                    }
                },
                {
                    "term": {
                        "genre.keyword": "fantasy"
                    }
                }
            ]
        }
    },
    "aggs": {
        "distinct_writer": {
            "terms": {
                "field": "writer.keyword"
            }
        }
    }
}

And this produce below result

"aggregations": {
        "distinct_writer": {
            "doc_count_error_upper_bound": 0,
            "sum_other_doc_count": 0,
            "buckets": [
                {
                    "key": "mike",
                    "doc_count": 3
                },
                {
                    "key": "andy",
                    "doc_count": 2
                }
            ]
        }
    }

CodePudding user response：

One way to do that is to divide your task in 2 elastic queries. Because that will help you to get book_count as well.

From first query, get all the writers for horror genre -

{
    "size": 0,
    "query": {
        "bool": {
            "filter": [
                {
                    "term": {
                        "genre.keyword": "horror"
                    }
                }
            ]
        }
    },
    "aggs": {
        "horror_genre_writers": {
            "terms": {
                "field": "writer.keyword"
            }
        }
    }
}

Result of First query -

[
  {"writer": "andy", "book_count": 1},
  {"writer": "mike", "book_count": 2},
  {"writer": "greg", "book_count": 1}
] 

From second query, get all the writers for fantasy genre -

{
    "size": 0,
    "query": {
        "bool": {
            "filter": [
                {
                    "term": {
                        "genre.keyword": "fantasy"
                    }
                }
            ]
        }
    },
    "aggs": {
        "fantasy_genre_writers": {
            "terms": {
                "field": "writer.keyword"
            }
        }
    }
}

Result of Second query -

[
  {"writer": "andy", "book_count": 1},
  {"writer": "mike", "book_count": 1},
  {"writer": "ursa", "book_count": 1}
] 

Now take intersection of the results from both queries to get your desired output -

[
  {"writer": "andy", "book_count": 2}, 
  {"writer": "mike", "book_count": 3}
]