Given two arrays like a = [10, 20, 30], and b = [9, 21, 32], how can I construct an array that consists of the minimum or maximum element based on index in snowflake, i.e. the desired output for minimum is [9,20,30] and for the maximum is [10,21,32]?

I looked at snowflake's array functions and didn't find a function that does this.

CodePudding user response：

Using numbers table/[] to access elements and ARRAY_AGG to build new arrays:

WITH cte AS (
    SELECT ARRAY_CONSTRUCT(10, 20, 30) AS a, ARRAY_CONSTRUCT(9, 21, 32) AS b
), numbers AS (
  SELECT ROW_NUMBER() OVER(ORDER BY seq4())-1 AS IND 
  FROM TABLE(GENERATOR(ROWCOUNT => 10001))
)
SELECT a,b
    ,ARRAY_AGG(LEAST(a[ind], b[ind])) WITHIN GROUP(ORDER BY n.ind) AS min_array
    ,ARRAY_AGG(GREATEST(a[ind], b[ind]))  WITHIN GROUP(ORDER BY n.ind) AS max_array
FROM cte
JOIN numbers n
  ON n.ind < GREATEST(ARRAY_SIZE(a), ARRAY_SIZE(b))
GROUP BY a,b;

Output:

CodePudding user response：

If the arrays are always the same size (and reusing Lukasz great data cte):

WITH cte AS (
    SELECT ARRAY_CONSTRUCT(10, 20, 30) AS a, ARRAY_CONSTRUCT(9, 21, 32) AS b
)
SELECT a,b
    ,ARRAY_AGG(LEAST(a[n.index], b[n.index])) WITHIN GROUP(ORDER BY n.index) AS min_array
    ,ARRAY_AGG(GREATEST(a[n.index], b[n.index]))  WITHIN GROUP(ORDER BY n.index) AS max_array
FROM cte
    ,table(flatten(a)) n
GROUP BY 1,2;

gives:

And if you have uneven lists:

WITH cte AS (
    SELECT ARRAY_CONSTRUCT(10, 20, 30) AS a, ARRAY_CONSTRUCT(9, 21, 32) AS b
    union all
    SELECT ARRAY_CONSTRUCT(10, 20, 30) AS a, ARRAY_CONSTRUCT(9, 21, 32, 45) AS b
)
SELECT a,b
    ,ARRAY_AGG(LEAST(a[n.index], b[n.index])) WITHIN GROUP(ORDER BY n.index) AS min_array
    ,ARRAY_AGG(GREATEST(a[n.index], b[n.index]))  WITHIN GROUP(ORDER BY n.index) AS max_array
FROM cte
    ,table(flatten(iff(array_size(a)>=array_size(b), a, b))) n
GROUP BY 1,2;

will pick the largest, but given the NULL from the smaller list will cause LEAST/GREATEST to return NULL and ARRAY_AGG drops nulls, you don't even need to size compare, unless you want to NVL/COALESCE that values to safe values for nulls.

SELECT 1 as a, null as b, least(a,b);

gives:

like so:

SELECT a,b
    ,ARRAY_AGG(LEAST(nvl(a[n.index],10000), nvl(b[n.index],10000))) WITHIN GROUP(ORDER BY n.index) AS min_array
    ,ARRAY_AGG(GREATEST(nvl(a[n.index],0), nvl(b[n.index],0)))  WITHIN GROUP(ORDER BY n.index) AS max_array
FROM cte
    ,table(flatten(iff(array_size(a)>=array_size(b), a, b))) n
GROUP BY 1,2;