计数不同和大小写问题

Question

我在 Snowflake 演示数据库上重现了实时数据问题。 case "Mark 1" 被评估为 NULL（没关系）。我假设基于此条件 ("Mark 2") 的 count distinct 应该 return 0 但它 returns 1.count distinct 和 count 之间没有区别。

我错过了什么？为什么在这种情况下 null 被计算在内？

代码：

WITH _table1 as(
    SELECT 
        O_CUSTKEY,
        COUNT(DISTINCT CASE WHEN O_TOTALPRICE>=50000 AND O_TOTALPRICE<100000 THEN O_ORDERKEY END) AS NO_SALES_OVER_50_TO_100,
        COUNT(DISTINCT CASE WHEN O_TOTALPRICE>=100000 THEN O_TOTALPRICE END) AS NO_SALES_OVER_100
    FROM SNOWFLAKE_SAMPLE_DATA.TPCH_SF1.ORDERS
    GROUP BY 1
)
SELECT 
    O_CUSTKEY,
    CASE WHEN NO_SALES_OVER_50_TO_100>0 THEN O_CUSTKEY END, --Mark 1
    COUNT(DISTINCT CASE WHEN NO_SALES_OVER_50_TO_100>0 THEN O_CUSTKEY end) --Mark 2
FROM _table1
WHERE O_CUSTKEY=8117
GROUP BY 1,2

Answer 1

这似乎是一个错误，因为如果用简单的逻辑检查案例，你会得到标记 2 的 0

WITH table1 as (
    SELECT 
        100 as column1,
        --0 as column2
        COUNT(DISTINCT CASE WHEN O_TOTALPRICE>=50000 AND O_TOTALPRICE<100000 THEN O_ORDERKEY END) as column2
    FROM SNOWFLAKE_SAMPLE_DATA.TPCH_SF1.ORDERS
    WHERE O_CUSTKEY = 8117
    GROUP BY 1
)
select column1, column2,
    case when column2 >0 then column1 end as mark1,
    count(distinct mark1) as mark2
FROM table1
WHERE column1 = 100
group by 1,2;

给予

COLUMN1 COLUMN2 MARK1   MARK2
100     0       NULL    1

是

WITH table1 as (
    SELECT 
        100 as column1,
        0 as column2
        --COUNT(DISTINCT CASE WHEN O_TOTALPRICE>=50000 AND O_TOTALPRICE<100000 THEN O_ORDERKEY END) as column2
    FROM SNOWFLAKE_SAMPLE_DATA.TPCH_SF1.ORDERS
    WHERE O_CUSTKEY = 8117
    GROUP BY 1
)
select column1, column2,
    case when column2 >0 then column1 end as mark1,
    count(distinct mark1) as mark2
FROM table1
WHERE column1 = 100
group by 1,2;

给予

COLUMN1 COLUMN2 MARK1   MARK2
100     0       NULL    0

因此你应该 post 把它变成雪花一样的错误。

计数不同和大小写问题

Count distinct and case issue

sql

snowflake-cloud-data-platform