Filtering on multiple fields in an Elasticsearch ObjectField()-CodePudding

I am having trouble figuring out the filtering syntax for ObjectFields() in django-elasticsearch-dsl. In particular, when I try to filter on multiple subfields of the same ObjectField(), I'm getting incorrect results.

For example, consider the following document

class ItemDocument(Document):
    product = fields.ObjectField(properties={
        'id': fields.IntegerField(),
        'name': fields.TextField(),
        'description': fields.TextField()
    })
    details = fields.ObjectField(properties={
        'category_id': fields.IntegerField(),
        'id': fields.IntegerField(),
        'value': fields.FloatField()
    })
    description = fields.TextField()

I want to find an Item with a detail object that has both category_id == 3 and value < 1.5, so I created the following query

x = ItemDocument.search().filter(Q("match",details__category_id=3) & Q("range",details__value={'lt':1.5})).execute()

Unfortunately, this returns all items which have a detail object with category_id==3 and a separate detail object with value < 1.5 e.g.

{
  "product": ...
  "details": [
    {
      "category_id": 3,
      "id": 7,
      "value": 20.0
    },
    {
      "category_id": 4,
      "id": 7,
      "value": 1.0
    },
    ...
]
}

instead of my desired result of all items that have a detail object with both category_id==3 AND value < 1.5 e.g.

{
  "product": ...
  "details": [
    {
      "category_id": 3,
      "id": 7,
      "value": 1.0
    },
    ...
]
}

How do I properly format this query using django-elasticsearch-dsl?

CodePudding user response：

You can use the nested query in Elasticsearch to filter on multiple subfields of the same ObjectField. Here is an example of how you can do this in django-elasticsearch-dsl:

x = ItemDocument.search().query(
    "nested",
    path="details",
    query=Q("match", details__category_id=3) & Q("range", details__value={'lt':1.5})
).execute()

The nested query allows you to filter on multiple subfields of the details object, and only return documents that have a detail object with both category_id==3 and value < 1.5.

You can also use the inner_hits option with the nested query to get the details of the matching detail objects:

x = ItemDocument.search().query(
    "nested",
    path="details",
    query=Q("match", details__category_id=3) & Q("range", details__value={'lt':1.5}),
    inner_hits={}
).execute()

This will add a nested field to each search result, which contains the details of the matching detail objects. You can access this field in your code like this:

results = x.to_dict()
for result in results['hits']['hits']:
    nested_results = result['inner_hits']['details']['hits']['hits']
    # do something with nested_results