Querying nested objects in Elasticsearch

Question

I have a Product-Merchant mapping which looks like the following

  catalog_map = {

        "catalog": {
            "properties": {
                "merchant_id": {
                    "type": "string",
                },
                "products": {
                    "type": "object",
                },
                "merchant_name" :{
                    "type" : "string"
                }
            }
        }
    }

"product" has objects, say , product_id , product_name , product_price. Products and merchants are mapped, such that :

    for merchant in Merchant.objects.all() :

        products = [{"product_name" : x.product.name, "product_price" : x.price, "product_id" : x.product.id , "product_category" : x.product.category.name} for x in MerchantProductMapping.objects.filter(merchant=merchant)]

        tab = {
            'merchant_id': merchant.id,
            'merchant_name': merchant.name,
            'product': products
            }

       res = es.index(index="my-index", doc_type='catalog', body=tab)

The data gets indexed smoothly, in the desired form. Now, when I query the data from given index, I do it in the following way :

GET /esearch-index/catalog/_search
{ 
"query": {
"bool" :{
    "must": [
       {"match": {
          "merchant_name": {
              "query": "Sir John"
          }
       }}], 
       "should": [
          {"match": {
             "product_name": {
                 "query": "Vanilla"
             }
          }}
       ]
    }}

This query gives me the result of all the products in the index with merchant name "Sir John" . However, I want it to return the details of the product "Vanilla" sold by "Sir John" instead.

On someone's recommendation, I used "_source" while querying, but that doesn't help.

How can I single out the information of one single object from the entire "catalog" index of the merchant?

Answer 1

Once your bool query has a must clause, all the conditions inside of it are required. The conditions inside of the should clause are not required. They will only boost the results. (http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/query-dsl-bool-query.html#query-dsl-bool-query)

So, going back to your query, it will retrieve all catalogs matching merchant_name "Sir John". This is the only required (must) condition. The name "Vanilla" will only boost results with the name "Vanilla" to the top, because it is not required.

If you want to retrieve "Vanilla" sold by "Sir John", put both conditions inside of the must clause and change your query to this:

{
  "query": {
    "bool": {
      "must": [
        {
          "match": {
            "merchant_name": {
              "query": "Sir John"
            }
          }
        },
        {
          "match": {
            "product_name": {
              "query": "Vanilla"
            }
          }
        }
      ]
    }
  }
}