Question

我正在努力尝试几天前提出的 SO 用户之一的建议，并使其与以下各项一起使用：

WITH your_table(ID,STOREDATE,VALUE,INFO)
AS
(
SELECT 1122,'1/1/2020',2,'DONE' UNION ALL
SELECT 1122,'1/2/2020',1,'DONE' UNION ALL
SELECT 1122,'1/3/2020',7,'DONE' UNION ALL
SELECT 1122,'1/4/2020',8,'DONE'
),

CTE AS
(
    SELECT ID,
    STOREDATE,
    VALUE,
    CASE 
        WHEN VALUE = 8 THEN 0
        WHEN VALUE + LAG(VALUE) OVER(ORDER BY STOREDATE) = 8 THEN 0
        WHEN VALUE + LEAD(VALUE) OVER(ORDER BY STOREDATE) = 8 THEN 0
        ELSE VALUE
    END VALUE2,
    INFO
    FROM your_table
)

SELECT *,
CASE 
    WHEN 
        (
            SELECT COUNT(*) FROM CTE A 
            WHERE A.VALUE2 = 0 AND A.STOREDATE < B.STOREDATE
        ) >= 1 AND B.VALUE = 8 THEN B.VALUE
    ELSE B.VALUE2
END VALUE3
FROM CTE B

NB ：我的想法是在任何给定的行中获得8的总和。因此，对于上面的输入，我的预期输出是这样。

输出：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    0           //1 + 7 = 8; it'll update both the rows with zero
1122    1/3/2020    0           //For this, it's just fine
1122    1/4/2020    8       DONE

现在我的问题是以下输入内容，我无法弄清楚如何获得预期的结果集。

输入：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    1       DONE
1122    1/3/2020    2       DONE
1122    1/4/2020    7       DONE

预期产量：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    0       
1122    1/3/2020    2       DONE
1122    1/4/2020    0

输入：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    2       DONE
1122    1/3/2020    2       DONE
1122    1/4/2020    2       DONE

预期产量：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    0       
1122    1/2/2020    0       
1122    1/3/2020    0       
1122    1/4/2020    0

我可能有几个ID来获得如下所示的准确结果，但是对查询进行调整并不能使我了解所需的内容。任何建议将不胜感激-谢谢。

WITH your_table(ID,STOREDATE,VALUE,INFO)
AS
(
SELECT 1122,'1/1/2020',2,'DONE' UNION ALL
SELECT 1122,'1/2/2020',7,'DONE' UNION ALL
SELECT 1122,'1/3/2020',1,'DONE' UNION ALL
SELECT 1122,'1/4/2020',8,'DONE' UNION ALL
SELECT 4466,'1/1/2020',2,'DONE' UNION ALL
SELECT 4466,'1/2/2020',7,'DONE' UNION ALL
SELECT 4466,'1/3/2020',1,'DONE' UNION ALL
SELECT 4466,'1/4/2020',8,'DONE'
),

CTE AS
(
    SELECT ID,
    STOREDATE,
    VALUE,
    CASE 
        WHEN VALUE = 8 THEN 0
        WHEN VALUE + LAG(VALUE) OVER(ORDER BY STOREDATE) = 8 THEN 0
        WHEN VALUE + LEAD(VALUE) OVER(ORDER BY STOREDATE) = 8 THEN 0
        ELSE VALUE
    END VALUE2,
    INFO
    FROM your_table
)

SELECT *,
CASE 
    WHEN 
        (
            SELECT COUNT(*) FROM CTE A 
            WHERE A.VALUE2 = 0 AND A.STOREDATE < B.STOREDATE
        ) >= 1 AND B.VALUE = 8 THEN B.VALUE
    ELSE B.VALUE2
END VALUE3
FROM CTE B

预期产量：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    0           
1122    1/3/2020    0           
1122    1/4/2020    8       DONE
4466    1/1/2020    2       DONE
4466    1/2/2020    0           
4466    1/3/2020    0           
4466    1/4/2020    8       DONE

更新1 ：请检查以下内容

输入：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    1       DONE
1122    1/3/2020    2       DONE
1122    1/4/2020    7       DONE

预期产量：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    0       
1122    1/3/2020    2       DONE
1122    1/4/2020    0

输入：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    2       DONE
1122    1/2/2020    2       DONE
1122    1/3/2020    2       DONE
1122    1/4/2020    2       DONE

预期产量：

ID      STOREDATE   VALUE   INFO
1122    1/1/2020    0       
1122    1/2/2020    0       
1122    1/3/2020    0       
1122    1/4/2020    0

Answer 1

您的第一个大错误是'stordate'不是日期，而是日期的字符串表示形式。因此，所有比较将是字符串的比较，而不是日期。作为字符串，首先是'01 / 02/2020'或'02 / 01/2019'。

import torch
torch.manual_seed(2020)

m = torch.tensor([[0, 1, 0]]).to(torch.int32)
x = torch.rand((1, 3, 2))
o = torch.rand((1, 3, 2))

print(o)
# tensor([[[0.5899, 0.8105],
#          [0.2512, 0.6307],
#          [0.5403, 0.8033]]])
print(x)
# tensor([[[0.4869, 0.1052],
#          [0.5883, 0.1161],
#          [0.4949, 0.2824]]])

o[:, m[0].to(torch.bool), :] = x[:, m[0].to(torch.bool), :]
# or 
# o[:, m[0] == 1, :] = x[:, m[0] == 1, :]

print(o)
# tensor([[[0.5899, 0.8105],
#          [0.5883, 0.1161],
#          [0.5403, 0.8033]]])

Answer 2

能否请您查看以下解决方案-

DEMO HERE

DEMO2 HERE

1
2020-03-29 01:45:37.099114
2
2020-03-29 01:45:37.099114
c
2020-03-29 01:45:52.114733
6
2020-03-29 01:45:52.114733
10
2020-03-29 01:45:52.114733
1
2020-03-29 01:45:52.114733
e
2020-03-29 01:46:07.115963
11
2020-03-29 01:46:07.115963
78
2020-03-29 01:46:07.115963
a
2020-03-29 01:46:22.117232
b
2020-03-29 01:46:37.118046

查询结果未按预期返回

2 个答案: