Question

下面的查询是一个草稿，其中包含我要编写的查询中的相关列，因此不要将其视为解决方案。将其用作表和列名称的指南。我正在尝试删除相同的ORDER_ID和ACCOUNT_ID相互抵消的任何事务。我不认为我可以使用SUM进行聚合，因为这会将所有TX_AMOUNT值加在一起进行分组。请参见TX_ID 6和7.这些都需要在结果集中显示。如何输出下表中的TX_ID，并滤除任何不说“显示”的东西？

SELECT 
T1.ACCOUNT_ID
T1.ORDER_ID,
T1.TX_ID
FROM TRANSACTION AS T1
WHERE
T1.ACCOUNT_ID IN (
SELECT T2.ACCOUNT_ID
FROM TRANSACTION AS T2
GROUP BY T2.ACCOUNT_ID, T2.ORDER_ID
HAVING SUM(T2.TX_AMOUNT) != 0 AND T2.ORDER_ID IS NOT NULL 
)
AND T1.ORDER_ID IN (
SELECT T3.ORDER_ID
FROM TRANSACTION AS T3
GROUP BY T3.ACCOUNT_ID, T3.ORDER_ID
HAVING SUM(T3.TX_AMOUNT) != 0 AND T3.ORDER_ID IS NOT NULL 
)

TX_ID   ORDER_ID ACCOUNT_ID  TX_AMOUNT
------------------------------------
1       A1       200         -3.00  <--------- DON'T SHOW THIS; OFFSET BY #2
2       A1       200         3.00   <--------- DON'T SHOW THIS; OFFSET BY #1
3       A1       200         3.00   <--------- SHOW THIS
4       A2       999         -10.01 <--------- DON'T SHOW THIS; OFFSET BY #5
5       A2       999         10.01  <--------- DON'T SHOW THIS; OFFSET BY #4
6       A2       999         10.01  <--------- SHOW THIS
7       A2       999         5.02   <--------- SHOW THIS

Answer 1

版本2： MUCH 更清洁...... Working DEMO with comments（您可能需要点击运行它！）才能看到所需的结果（或者我有一个缓存问题）

CTE（通用表格表达式）只是您显示的数据设置
CTE2只是添加一个由tx_amount，order_Id，account_Id分区的行号。这里的关键是我们得到每个order_ID的row_number，Account_ID和tax_Amount在这3个值改变时重新启动，但在它们保持不变时递增。这后来允许我们在相反的tx_amounts上排除相似的匹配而不消除那些当一方有多于另一方时（你的$ 3.00例子）
Select只是从基集中提取记录，其中存在具有相同行号order_id和帐户的相反值。如果没有，那么我们就知道这是一个没有匹配tx_Amount的值，因此我们想保留它。
询问您是否有疑问！如果有什么不清楚的话，很乐意提供帮助
最后，如果我们改变CTE2，那么rowNumber（）按tx_ID asc而不是tx_Amount desc排序（除了我需要在row_nubmer上订单之外没有任何其他目的），那么我们就会在FIFO方法之后首先一致地摆脱最低数量的匹配）

With CTE (TX_ID,   ORDER_ID, ACCOUNT_ID,  TX_AMOUNT) as (

SELECT 1,       'A1',       200,         -3.00  UNION ALL
SELECT 2,       'A1',       200,         3.00   UNION ALL
SELECT 3,       'A1',       200,         3.00   UNION ALL
SELECT 4,       'A2',       999,         -10.01 UNION ALL
SELECT 5,       'A2',       999,         10.01  UNION ALL
SELECT 6,       'A2',       999,         10.01  UNION ALL
SELECT 7,       'A2',       999,         5.02 ),

cte2 as (
SELECT A.*, row_number() over (partition by order_ID, Account_ID, Tx_Amount order by tx_Amount desc) RN
FROM cte A)

SELECT * 
FROM cte2 A
WHERE NOT exists (SELECT * 
                  FROM cte2 B
                  WHERE A.Order_ID = B.Order_ID
                    and A.Account_ID = B.Account_Id
                    and A.tx_Amount*-1 = B.tx_Amount
                    and A.RN = B.RN)

给我们:(注意我们应该通过将*更改为所需的字段来消除RN，但此时我太懒了）

+----+-------+----------+------------+-----------+----+
|    | TX_ID | ORDER_ID | ACCOUNT_ID | TX_AMOUNT | RN |
+----+-------+----------+------------+-----------+----+
|  1 |     2 | A1       |        200 |      3,00 |  2 |
|  2 |     7 | A2       |        999 |      5,02 |  1 |
|  3 |     5 | A2       |        999 |     10,01 |  2 |
+----+-------+----------+------------+-----------+----+

版本1 :(抓住这个丑陋的;我的意思是认真的;谁这么想？）我这样做......

做点什么。（第1版）
笑（一个重要的，被忽视的一步）
然后做正确的（参见上面的第2版）
现在让它变得更好。（索引，调整连接拼写，布局注释，在CTE2中的row_number逻辑上使用正确的顺序）

DEMO

With CTE (TX_ID,   ORDER_ID, ACCOUNT_ID,  TX_AMOUNT) as (

SELECT 1,       'A1',       200,         -3.00  UNION ALL
SELECT 2,       'A1',       200,         3.00   UNION ALL
SELECT 3,       'A1',       200,         3.00   UNION ALL
SELECT 4,       'A2',       999,         -10.01 UNION ALL
SELECT 5,       'A2',       999,         10.01  UNION ALL
SELECT 6,       'A2',       999,         10.01  UNION ALL
SELECT 7,       'A2',       999,         5.02 ),
cte2 as (
SELECT * 
FROM (Select A.Tx_Id aTx_ID
           , A.order_ID as AOrderID
           , A.Account_ID as AAccount_ID
           , A.tx_Amount as ATx_Amount
           , Row_number() over (partition by Order_ID, Account_ID, tx_Amount order by tx_Amount asc) ARN

      from cte a 
      WHERE tx_Amount <=0) A
FULL OUTER JOIN (SELECT b.tx_Id
                      , b.order_Id
                      , b.Account_Id
                      , b.tx_Amount
                      ,  Row_number() over (partition by Order_ID, Account_ID, tx_Amount order by tx_Amount desc) BRN 
                 FROM  CTE B 
                 WHERE  tx_Amount>0) B
  on A.AOrderID = B.Order_ID
 and A.AAccount_ID = B.Account_ID
 and A.ATx_Amount*-1 = B.tx_Amount
 and A.ARN=B.BRN
Where a.Atx_ID is null
  or B.tx_ID is null)

  Select ATX_ID, AORDERID, AAccount_ID, ATX_AMOUNT from cte2 where ATX_ID is not null
  UNION ALL
  Select TX_ID, ORDER_ID, Account_ID, TX_AMOUNT from cte2 where TX_ID is not null

查询过滤掉具有确切偏移量

1 个答案: