Best way to index user specific products on store index

[roman731 (RC)]

I have an index with ~20000 products for an e-commerce website. I would like to implement a "previously purchased feature" for specific users.. so only specific products will return for specific users in this case... but also with a sort on their most frequently purchased item.

If there was no sorting, I think the best way to do this would be just to a store a keyword array of user_ids on each product.

So my products would look something like:

{
       "name": "Product A",
       .......
       "bought_by_user_ids": [1,2,3,4,5,8,10]
}

And then my query would be something like

{
"query": {
    "bool": {
      "must": [
        { "match":  { "name" : "Product" } }
      ],
      "filter: [
        { "match": { "bought_by_user_ids": 1 } }
      ]
      }
  }
}
Where 1 is the current user's integer ID.

The kicker is I also want to sort on most commonly purchased by each user. So my thinking was instead of just keeping user_ids on the product, I can store an object field with user_ids as keys and the number of times they purchased the product as the values.

So the product would look like:

{
    "name": "Product A",
     .......
     "bought_by_user_ids": {
         "1": 5,
         "2": 10
      }
}

Meaning user with ID 1 has bought the product 5 times, and user with ID 2 has bought the product 10 times.

And the query would be an exists query.

{
     "query": {
           "exists": {
	            "field": "bought_by_user_ids.1"
	    }
	  },
	  "sort" : [{
             "bought_by_user_ids.1" : {
                  "order" : "desc",
                  "unmapped_type" : "float"
             }
       }]
}
Where 1 is the current user's integer ID.

The solution works but I'm wondering if there's a better way to do this? I noticed I may eventually run into a total_fields limit because apparently every entry in the object is its own field? Any help is appreciated.

Thanks.