为 SQL 中的一组列生成相同的 UUID

Question

Can someone pls suggest a way to write SQL query which would generate a unique UUID which would be same for a group of columns in amazon athena.

For example, i have a table like this, where i want to create a UUID for columns1, columns2 and columns3.

column1 | column2 | column3 | column 4
2016    | 101     | 1       | 25
2016    | 101     | 1       | 59
2017    | 105     | 2       | 57
2017    | 105     | 2       | 78

Output 1 must look like

ID      | column1 | column2 | column3 | column 4
UUID-1  | 2016    | 101     | 1       | 25
UUID-1  | 2016    | 101     | 1       | 59
UUID-2  | 2017    | 105     | 2       | 57
UUID-2  | 2017    | 105     | 2       | 78

Output 2: should look like

ID      | count |column1 | column2 | column3
UUID-1  | 2     |2016    | 101     | 1
UUID-2  | 2     |2017    | 105     | 2

I understand that grouping can be done on output 1 to generate output 2. Can someone suggest, how i can generate output 1?

Thanks.

Answer 1

You can try to use uuid() function.

SELECT uuid() id,
       COUNT(*),
       column1 ,
       column2 ,
       column3
FROM T
GROUP BY column1 ,
       column2 ,
       column3

EDIT

I saw you edit your question, you can try to use subquery and self join get output1

SELECT t1.*,t2.column4
FROM (
    SELECT DISTINCT uuid() id,
           column1 ,
           column2 ,
           column3
    FROM T
) t1 INNER JOIN T t2 
ON t1.column1 = t2.column1
AND t1.column2 = t2.column2
AND t1.column3 = t2.column3

another way you can try to use max window function to get only one GUID per column1 , column2 , column3 columns.

select max(id) over (partition by column1,column2,column3) as id, 
       column1,
       column2,
       column3,
       column4
from 
(
    SELECT uuid() id,*
    FROM T
) t1