Function calls passing column names?

[kgutwin]

I have been working on trying to create PRQL functions to replace some of the boilerplate operations in our project. One of the operations is a very simple "mapping" operation which uses a key-value lookup table to transform a column from keys to values. The function currently looks like this:

let map_into = func
  map_tbl <relation>
  column <scalar>
  map_tpl <tuple>
  tbl <relation>
  -> (
  join side:left map=map_tbl (column == that.key) tbl
  derive map_tpl
  select !{ map.key, map.value }
)

and it can be used like this:

let postcodes = [
  { key = "35032", value = "Bon Air" },
  { key = "35046", value = "Clanton" },
  { key = "35070", value = "Garden City" },
]

from employees
select { last_name, first_name, location }
map_into postcodes location { location = that.value } 

Resulting SQL, as expected (click to expand)

WITH table_0 AS (
  SELECT
    '35032' AS "key",
    'Bon Air' AS value
  UNION
  ALL
  SELECT
    '35046' AS "key",
    'Clanton' AS value
  UNION
  ALL
  SELECT
    '35070' AS "key",
    'Garden City' AS value
),
postcodes AS (
  SELECT
    "key",
    value
  FROM
    table_0
)
SELECT
  employees.last_name,
  employees.first_name,
  map.value AS location
FROM
  employees
  LEFT JOIN postcodes AS map ON employees.location = map."key"

-- Generated by PRQL compiler version:0.11.2 (https://prql-lang.org)

However, I'm interested in reducing the duplication of the location term in the map_into function call. I would like to be able to simplify it to just map_into postcodes location. However, when I move the definition of the last tuple into the function, as below:

let map_into = func
  map_tbl <relation>
  column <scalar>
  tbl <relation>
  -> (
  join side:left map=map_tbl (column == that.key) tbl
  derive { column = map.value }
  select !{ map.key, map.value }
)

let postcodes = [
  { key = "35032", value = "Bon Air" },
  { key = "35046", value = "Clanton" },
  { key = "35070", value = "Garden City" },
]

from employees
select { last_name, first_name, location }
map_into postcodes location

This results in the wrong SQL:

-- snip...
SELECT
  employees.last_name,
  employees.first_name,
  employees.location,    -- this should not be present
  map.value AS "column"  -- this should be `map.value AS location`
FROM
  employees
  LEFT JOIN postcodes AS map ON employees.location = map."key"

-- Generated by PRQL compiler version:0.11.2 (https://prql-lang.org)

This is because the word column has now become the name of the new column being defined through the mapping operation -- not the name of the column from the call of the function.

Is there any way to define the map_into function to avoid the double-reference to the mapped column? For example, would it possible to pass the name of the original column through the function parameter, and use it on the left-hand side of a tuple definition inside the function? I hope this question is clear, I apologize if I'm not explaining myself well.