SQL Server:根据其他两列的条件更新布尔列

当我的

InvoiceFileId

只有一个

InvoiceId时，我需要将列
PrimaryInvoiceFile
中的所有行设置为1（true）

但是，如果
InvoiceId
有多个
InvoiceFileId
，我需要将所有
PrimaryInvoiceFile
行设置为0（false），除了根据添加日期添加的最新
InvoiceFileId
之外

例如，它应该如下所示：

|CreatedDate|InvoiceId|InvoiceFileId|PrimaryInvoiceFile|
+-----------+---------+-------------+------------------+
|2019-01-16 | 1       | 1           | 1                |
|2019-01-17 | 2       | 2           | 1                |
|2019-01-18 | 3       | 3           | 0                |
|2019-01-19 | 3       | 4           | 0                |
|2019-01-20 | 3       | 5           | 1                |
|2019-01-21 | 4       | 6           | 1                |

我刚刚添加了
PrimaryInvoiceFile
列迁移，并将默认值设置为0

在此方面的任何帮助都将不胜感激！我一直在绞尽脑汁试图让我的更新语句执行此更新。
请尝试以下操作：

;WITH Data AS (
    SELECT t.CreatedDate,t.InvoiceId,t.InvoiceFieldId,t.PrimaryInvoiceFile
        ,COUNT(*)OVER(PARTITION BY t.InvoiceId) AS [cnt]
    FROM [YourTableName] t
)
UPDATE d SET d.PrimaryInvoiceFile = CASE WHEN d.cnt = 1 THEN 1 ELSE 0 END
FROM Data d
;

要玩的查询：

DROP TABLE IF EXISTS #YourTableName;
CREATE TABLE #YourTableName(CreatedDate DATETIME2,InvoiceId INT, InvoiceFieldId INT,PrimaryInvoiceFile BIT);
INSERT INTO #YourTableName(CreatedDate,InvoiceId,InvoiceFieldId)VALUES
    ('2019-01-16',1,1),('2019-01-17',2,2),('2019-01-18',3,3),('2019-01-19',3,4),('2019-01-20',3,5),('2019-01-21',4,6)

;WITH Data AS (
    SELECT t.CreatedDate,t.InvoiceId,t.InvoiceFieldId,t.PrimaryInvoiceFile,COUNT(*)OVER(PARTITION BY t.InvoiceId) AS [cnt]
    FROM #YourTableName t
)
UPDATE d SET d.PrimaryInvoiceFile = CASE WHEN d.cnt = 1 THEN 1 ELSE 0 END
FROM Data d
;

SELECT t.CreatedDate,t.InvoiceId,t.InvoiceFieldId,t.PrimaryInvoiceFile
FROM #YourTableName t
;

DROP TABLE IF EXISTS #YourTableName;

您可以在执行更新时使用rownumber来实现所需的结果。此外，按降序排序，以便获得最近的日期

让我们创建一个表，将PrimaryInvoiceFile保持为null，然后稍后更新

select '2019-01-16' as CreatedDate, 1 as invoiceID,         1 as Invoicefield, null 
as PrimaryInvoiceFile 
into #temp 
union all 
select '2019-01-17' as CreatedDate, 2 as invoiceID,         2 as Invoicefield, null 
as Primaryinvoicefile union all 
select '2019-01-18' as CreatedDate, 3 as invoiceID,         3 as Invoicefield, null 
as Primaryinvoicefile union all 
select '2019-01-19' as CreatedDate, 3 as invoiceID,         4 as Invoicefield, null 
as Primaryinvoicefile union all 
select '2019-01-20' as CreatedDate, 3 as invoiceID,         5 as Invoicefield, null 
 as Primaryinvoicefile union all 
 select '2019-01-21' as CreatedDate, 4 as invoiceID,         6 as Invoicefield, null 
 as Primaryinvoicefile

update t 
set Primaryinvoicefile = tst.Rownum 
from #temp t 
join  
(Select  invoiceID, Invoicefield,CreatedDate, 
case when ROW_NUMBER() over (partition by invoiceID order by createddate desc) = 1 
then 1 else 0 end as Rownum   from #temp) tst 
on  tst.CreatedDate = t.CreatedDate 
      and tst.invoiceID = t.invoiceID 
      and tst.Invoicefield =  t.Invoicefield 

Case statement would make sure that you are value as 1 for only the rows where you have 1 row for invoice ID or for the most recent data. 

select * from #temp 

输出：

 CreatedDate    invoiceID   Invoicefield    PrimaryInvoiceFile
 2019-01-16        1          1                    1
 2019-01-17        2          2                    1
 2019-01-18        3          3                    0
 2019-01-19        3          4                    0
 2019-01-20        3          5                    1
 2019-01-21        4          6                    1

如果默认PrimaryInvoiceFile设置为0，则需要更新按InvoiceId分组的最大CreatedDate的PrimaryInvoiceFile

UPDATE inv1 
SET inv1.PrimaryInvoiceFile = 1
FROM invoices inv1 JOIN 
    (SELECT max(CreatedDate) as maxDate, InvoiceId 
     FROM invoices 
     GROUP BY InvoiceId ) as inv2  
WHERE inv1.CreatedDate=inv2.maxDate and inv1.InvoiceId= inv2.InvoiceId