I’ve aggregated multiple calls to different tables into a single query with multiple joins where the is a common value between them (ID). So I have something like:

Table1
|> from(as: :table_1)
|> join(:left, [t], t2 in Table2, on: t.id == t2.id, as: :table_2)
...
|> where([t], i.id == ^id)
|> select([table_1: table_1, ..., table_9: table_9], %{table_1: table_1, ... table_9: table_9})
|> Repo.one()

This works fine, it will have a different structure per field and request all the information I need. Since this can grow in the future, and I decided to request more or fewer tables depending on the use case, I want to make this all dynamically.

I have made this almost work except for the select function, which I cannot find a way to make the same behaviour with select_merge. I’ve changed that to the following structure:

defp apply_table_select(query, tables) do 
  query = select(query, [i], %{table_1: i})

  Enum.reduce(tables, query, fn table, query -> 
    select_merge(query, [{^table, m}], %{^table => m})
  end)
end

But this doesn’t work. I need to add ^ because otherwise (and makes sense since I’m referring to a variable), I have a compile-time error.

** (DBConnection.EncodeError) Postgrex expected a binary, got :table_2. Please make sure the value you are passing matches the definition in your table or in your query or convert the value accordingly.

The issue appears to be related how I added the field to select_merge:

#Ecto.Query<from i0 ..., select: %{table_1: i0, table_2: t1}>

vs

#Ecto.Query<from i0 ..., select: merge(%{table_1: i0}, %{^:table_2 => t1})>

Not sure if this is a limitation from Ecto or I’m doing something wrong.

EDIT: Just found this topic, basically has the same issue.