fetching single child row based on a max value using Django ORM

Question

I have a model, "Market" that has a one-to-many relation to another model, "Contract":

class Market(models.Model):
    name = ...
    ...

class Contract(models.Model):
    name= ...
    market = models.ForeignKey(Market, ...)
    current_price = ...

I'd like to fetch Market objects along with the contract with the maximum price of each. This is how I'd do it via raw SQL:

SELECT M.id as market_id, M.name as market_name, C.name as contract_name, C.price 
as price from pm_core_market M INNER JOIN
    (SELECT market_id, id, name, MAX(current_price) as price 
        FROM pm_core_contract GROUP BY market_id) AS C
ON M.id = C.market_id

Is there a way to implement this without using SQL? If there is, which one should be preferred in terms of performance?

Answer 1

Django 1.1 (currently beta) adds aggregation support to the database API. Your query can be done like this:

from django.db.models import Max, F

Contract.objects.annotate(max_price=Max('market__contract__current_price')).filter(current_price=F('max_price')).select_related()

This generates the following SQL query:

SELECT contract.id, contract.name, contract.market_id, contract.current_price, MAX(T3.current_price) AS max_price, market.id, market.name
FROM contract LEFT OUTER JOIN market ON (contract.market_id = market.id) LEFT OUTER JOIN contract T3 ON (market.id = T3.market_id)
GROUP BY contract.id, contract.name, contract.market_id, contract.current_price, market.id, market.name
HAVING contract.current_price =  MAX(T3.current_price)

The API uses an extra join instead of a subquery (like your query does). It is difficult to tell which query is faster, especially without knowing the database system. I suggest that you do some benchmarks and decide.