Can I write this ecto query better?

Questions / Help
1

I’m attempting to translate a query I’ve written into Ecto. Here’s what I have:

  def select_merged_prs(query \\ GithubPullRequest, period) do
    query
    |> select([github_pull_request], %{
      count: count(github_pull_request.id),
      date: fragment("DATE_TRUNC(?, (?->>'merged_at')::timestamp) as date", ^period, github_pull_request.data),
      name: fragment("?->'user'->>'login' as name", github_pull_request.data),
    })
    |> group_by(fragment("date, name"))
  end

I feel like I’m missing something having to alias the fields using fragment so I can reference them in the group_by. Could this be more elegant?

For reference, the original SQL query is:

SELECT
   count(github_pull_requests.id) as merged_prs,
   DATE_TRUNC('day', 
   (
      github_pull_requests.data ->> 'merged_at'
   )
   ::timestamp) as date,
   github_pull_requests.data ->'user'->>'login' as name
FROM
   github_pull_requests 
WHERE
   github_pull_requests.github_repository_id = 35 
GROUP BY
   date, name
ORDER BY
   date;