雅典娜：将列转换为不同的值

Question

I would like to convert a column to another value, kind of like having a map. I am currently doing something like

SELECT col1,
    col2,
    CASE
        data = 'data1' THEN 'd1'
        WHEN data = 'data2' THEN 'd2' ELSE data
    END AS converted_data
FROM some_table

Which works fine, but I really have a bunch of other possibilities, maybe 20 or more. It can potentially even get to hundreds in the future.

Is it possible to have the map as a table, and convert the values from that table? I believe you can use JOIN but that seems it might be pretty expensive? Is there a way to use CASE without writing each statement and just basing the values from that table?

Answer 1

There really is not a way to use a case statement without writing each condition to do what you are trying to accomplish.

If your end goal is a mapping to change the value of a column to something else and that could change or is a large number of items; for easy maintenance you would want to use another table which then is going to require a join or the use of a sub-query.

Here are 2 working examples with test data to show how those work:

Left Join:

#CTEs with test data
with convert_data as (
    select *
    From (
            values('data1', 'd1'),
                ('data2', 'd2')
        ) as convertdata (ddatavalue, datareplace)
),
some_table as(
    select *
    From (
            values('1', '1', 'data1'),
                ('2', '2', 'data2'),
                ('3', '3', 'data3')
        ) as sametable (col1, col2, ddata)
)
#query showing how to do the join
select col1,
    col2,
    coalesce(datareplace, ddata) as converted_data
from some_table st
    left join convert_data cd on cd.ddatavalue = st.ddata

Sub-query:

#CTEs with test data
with convert_data as (
    select *
    From (
            values('data1', 'd1'),
                ('data2', 'd2')
        ) as convertdata (ddatavalue, datareplace)
),
some_table as(
    select *
    From (
            values('1', '1', 'data1'),
                ('2', '2', 'data2'),
                ('3', '3', 'data3')
        ) as sametable (col1, col2, ddata)
)
#query showing how to do the sub-query
select col1,
    col2,
    coalesce(
        (
            select cd.datareplace
            from convert_data cd
            where cd.ddatavalue = st.ddata
        ),
        ddata
    ) as converted_data
from some_table st

Which one will perform better over the other will really be dependent on your data, structure and data file format. Test both and see which one works better for you.