带案例的 SQL GroupBy

Question

我对 SQL 还很陌生，我正在尝试根据条件编写一个对 groupBy 的查询。 我想要的是 -> 对于市场 CA 和 MX，总金额基于 MARKETPLACE 和 TRANSACTION_TYPE（代码始终相同） -> 对于市场 BR，根据 MARKETPLACE & TRANSACTION_TYPE & CONSUMPTION_TYPE 计算金额和本地金额的总和

但它给了我例外：服务：AmazonAthena；状态代码：400；错误代码：当我运行以下查询时出现 InvalidRequestException。

SELECT 
    CASE rae.TRANSACTION_MARKETPLACE_ID 
        WHEN '123456' THEN 'CA'
        WHEN '234567' THEN 'MX'
        WHEN '345678' THEN 'BR'
        END as MARKETPLACE,
    TRANSACTION_TYPE, SUM(AMOUNT) as AMOUNT, CODE,
    IF (TRANSACTION_MARKETPLACE_ID = '345678',
        SUM(rae.LOCAL_AMOUNT)
        ) as LOCAL_AMOUNT,
    IF (
        rae.TRANSACTION_MARKETPLACE_ID = '345678',
        LOCAL_CODE
        ) as LOCAL_CODE,
    IF (
        rae.TRANSACTION_MARKETPLACE_ID = '345678',
        CONSUMPTION_TYPE
        ) as CONSUMPTION_TYPE
FROM xxxTable as rae
WHERE MODEL = 'MODEL_1'
    AND TRANSACTION_MARKETPLACE_ID IN ('123456', '234567', '345678')
    AND (TRANSACTION_TYPE = 'TYPE_1' OR TRANSACTION_TYPE = 'TYPE_2' OR (TRANSACTION_TYPE = 'TYPE_3' AND ALLOW_CLAWBACK = TRUE))
    AND TRANSACTION_DATE BETWEEN timestamp '2023-09-01' AND timestamp '2023-09-30'
GROUP BY (CASE WHEN MARKETPLACE = 'BR'
         THEN TRANSACTION_MARKETPLACE_ID, TRANSACTION_TYPE, CODE, LOCAL_CODE, CONSUMPTION_TYPE END),
         (CASE WHEN MARKETPLACE IN ('BR', 'CA') THEN TRANSACTION_MARKETPLACE_ID, TRANSACTION_TYPE, CODE END);

预期输出会喜欢

市场	交易类型	金额	代码	消费类型	LOCAL_AMOUNT	本地代码
CA	TYPE_1	xxx	美元
MX	TYPE_1	xxx	美元
CA	TYPE_2	xxx	美元
MX	TYPE_2	xxx	美元
BR	TYPE_1	xxx	美元	CON_TYPE_1	xxx	美元
BR	TYPE_1	xxx	美元	CON_TYPE_2	xxx	美元
BR	TYPE_2	xxx	美元	CON_TYPE_1	xxx	美元
BR	TYPE_2	xxx	美元	CON_TYPE_2	xxx	美元

在 groupby 中添加案例后，它开始出现异常。谁能给出一些想法为什么？或者分成两个查询会更好吗？ 另外，当 TRANSACTION_MARKETPLACE_ID = '345678' 时，我想再选择三列，我分别设置它们，有没有办法将它们设置在一起？

Answer 1

我认为您在单个查询中尝试执行的操作没有意义。我从未见过像您尝试做的那样对案例进行条件分组，更不用说添加更多列了。

我建议将其分成两个不同的查询。然后，您甚至可以为具有较少列的查询使用空值来均衡缺少的列，并在两个结果之间执行 UNION。比如：

SELECT 
    CASE rae.TRANSACTION_MARKETPLACE_ID 
        WHEN '123456' THEN 'CA'
        WHEN '234567' THEN 'MX'
        END as MARKETPLACE,
    TRANSACTION_TYPE, SUM(AMOUNT) as AMOUNT, CODE,
    0 as LOCAL_AMOUNT, '' as LOCAL_CODE, '' as CONSUMPTION_TYPE
from rae as rae
WHERE MODEL = 'MODEL_1'
    AND TRANSACTION_MARKETPLACE_ID IN ('123456', '234567')
    AND (TRANSACTION_TYPE = 'TYPE_1' OR TRANSACTION_TYPE = 'TYPE_2' OR (TRANSACTION_TYPE = 'TYPE_3' AND ALLOW_CLAWBACK = TRUE))
    AND TRANSACTION_DATE BETWEEN timestamp '2023-09-01' AND timestamp '2023-09-30'
 GROUP BY TRANSACTION_MARKETPLACE_ID, TRANSACTION_TYPE, CODE
UNION
 SELECT 
    'BR' as MARKETPLACE,
    TRANSACTION_TYPE, SUM(AMOUNT) as AMOUNT, CODE,
    SUM(rae.LOCAL_AMOUNT) as LOCAL_AMOUNT,LOCAL_CODE, CONSUMPTION_TYPE
from rae as rae
WHERE MODEL = 'MODEL_1'
    AND TRANSACTION_MARKETPLACE_ID ='345678'
    AND (TRANSACTION_TYPE = 'TYPE_1' OR TRANSACTION_TYPE = 'TYPE_2' OR (TRANSACTION_TYPE = 'TYPE_3' AND ALLOW_CLAWBACK = TRUE))
    AND TRANSACTION_DATE BETWEEN timestamp '2023-09-01' AND timestamp '2023-09-30'
 GROUP BY TRANSACTION_MARKETPLACE_ID, TRANSACTION_TYPE, CODE, LOCAL_CODE, CONSUMPTION_TYPE