Question

我有一个名为temptable的表，我必须根据'trantype'和'tradedate'更新每个'cusip'的'quantity'。逻辑为sum（BUY）== sum（SELL），当SELL之后再次出现BUY时，重复相同的逻辑。因此，我想到了一个随机生成所有BUY值并根据这些值操纵SELL值的想法。

这是表脚本和一些示例数据：

newDBName = "${env.accountName}"

这就是我想要获得的：

示例1：

newDBName = env.accountName

扩展结果：

CREATE TABLE temptable(
TRANTYPE  VARCHAR(4) NOT NULL 
,CUSIP     VARCHAR(9) NOT NULL
,TRADEDATE DATE  NOT NULL
,QUANTITY  INTEGER 
);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ003$03','2018-08-01',3517150);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ004$04','2018-08-01',2575357);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ102$02','2018-08-01',6523548);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ006$06','2018-08-02',9429433);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ006$06','2018-08-02',9240844);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ006$06','2018-08-02',9056027);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ007$07','2018-08-02',4266903);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ104$04','2018-08-02',6914360);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ004$04','2018-08-03',2523849);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ007$07','2018-08-08',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('BUY','XYZ007$07','2018-08-15',4097932);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ006$06','2018-09-04',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ102$02','2018-10-05',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ003$03','2018-10-18',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ003$03','2018-10-24',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ003$03','2018-10-26',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ003$03','2018-10-29',NULL);
INSERT INTO temptable(TRANTYPE,CUSIP,TRADEDATE,QUANTITY) VALUES ('SELL','XYZ007$07','2018-08-16',NULL);

示例2：

SELECT * FROM [dbo].temptable WHERE CUSIP ='SCD007$07';

扩展结果：

-- +----------+-----------+------------+----------+
-- | TRANTYPE |   CUSIP   | TRADEDATE  | QUANTITY |
-- +----------+-----------+------------+----------+
-- | BUY      | XYZ007$07 | 2018-08-02 |  4266903 |
-- | SELL     | XYZ007$07 | 2018-08-08 |  4266903 |
-- | BUY      | XYZ007$07 | 2018-08-15 |  4097932 |
-- | SELL     | XYZ007$07 | 2018-08-16 |  4097932 |
-- +----------+-----------+------------+----------+

这是我尝试过的：

SELECT * FROM temptable WHERE CUSIP = 'SCD003$03';`

我该怎么做？

Answer 1

您要更新出售记录（quantity的{{1}}，并将其设置为之前 buy < / em>（trantype = 'SELL'）的同一客户（quantity），其中记录按trantype = 'BUY'排序。当两次购买之间有多个销售记录时，您希望将先前购买中的cusip平均分配到销售记录中。

这是一个产生预期输出的SQLServer 2017查询。基本上，它会自动tradedate与相应的销售记录一起购买记录；使用条件为quantity的相关子查询来查找相关的销售记录（第一个记录）。然后，窗口函数JOIN用于根据连续销售记录的数量划分数量。该查询还会考虑现有的销售数量（如果有）。

NOT EXISTS

现在，如果您要更新源表，可以将查询转到CTE。可以简化查询，因为它只需要影响空的销售记录。

COUNT(*) OVER(...)

Demo on DB Fiddle ：

TRANTYPE | CUSIP | TRADEDATE | QUANTITY :------- | :-------- | :------------------ | -------: BUY | SCD003$03 | 01/08/2018 00:00:00 | 3517150 SELL | SCD003$03 | 18/10/2018 00:00:00 | 879287 SELL | SCD003$03 | 24/10/2018 00:00:00 | 879287 SELL | SCD003$03 | 26/10/2018 00:00:00 | 879287 SELL | SCD003$03 | 29/10/2018 00:00:00 | 879287 BUY | SCD004$04 | 01/08/2018 00:00:00 | 2575357 BUY | SCD004$04 | 03/08/2018 00:00:00 | 2523849 BUY | SCD006$06 | 02/08/2018 00:00:00 | 9429433 BUY | SCD006$06 | 03/08/2018 00:00:00 | 9240844 BUY | SCD006$06 | 04/08/2018 00:00:00 | 9056027 SELL | SCD006$06 | 04/09/2018 00:00:00 | 9056027 BUY | SCD007$07 | 02/08/2018 00:00:00 | 4266903 SELL | SCD007$07 | 08/08/2018 00:00:00 | 4266903 BUY | SCD007$07 | 15/08/2018 00:00:00 | 4097932 SELL | SCD007$07 | 16/08/2018 00:00:00 | 4097932 BUY | SCD102$02 | 01/08/2018 00:00:00 | 6523548 SELL | SCD102$02 | 05/10/2018 00:00:00 | 6523548 BUY | SCD104$04 | 02/08/2018 00:00:00 | 6914360

根据其他值更新单元格中的值

1 个答案: