SQL 服务器：ROW_NUMBER() OVER on table 变量以修改重复列

Question

假设我有 table 变量这样声明...

DECLARE @LocalTable TABLE
(
    IdField NVARCHAR(MAX),
    NameField NVARCHAR(MAX)
)

然后我像这样填充它...

INSERT INTO @LocalTable
SELECT
    IdColumn,
    NameColumn
FROM SourceTable

源 table 中的

NameColumn 可能具有重复值，因此本地 table 中的 NameField 将具有相同的重复值。

假设我想像这样将本地 table 插入目标 table...

INSERT INTO TargetTable (NewIdColumn, NewNameColumn)
    SELECT
        IdField,
        NameField
    FROM 
        @LocalTable

但是：TargetTable 中的 NewNameColumn 有一个 UNIQUE 约束，因此重复会导致异常。

我想应用这个例子，

ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY NameField)

这样 NameField 就是 appended/suffixed 并且有一个数字表示它的重复。

我有这个可以 select 正确附加值的工作示例，但我无法让它在这样的更新语句中工作：

UPDATE localtable 
SET NameField = AppendedNameField 
FROM @LocalTable AS localtable

SELECT 
    CONCAT(Ref.NameField, ROW_NUMBER() OVER (PARTITION BY Ref.NameField 
                                             ORDER BY Source.IdField)), *
FROM 
    @LocalTable AS Source
INNER JOIN 
    @LocalTable AS Ref ON Ref.NameField = Source.NameField  
                       AND Ref.IdField != Source.IdField

提前致谢。

Answer 1

只需将身份字段添加到临时文件 table。

DECLARE @LocalTable TABLE
(
ix int identity primary key,
IdField NVARCHAR(MAX),
NameField NVARCHAR(MAX)
)

Insert into @LocalTable(IdColumn, NameColumn)
SELECT
IdColumn,
NameColumn
FROM SourceTable
-- Make sure same names are consecutive in the table
ORDER BY NameColumn

像这样设置NameColumn：

update lt set 
NameColumn = NameColumn 
-- Add a number based on the ix, minus the lowest ix entry for the same name
+ cast(
    (select lt.ix - min(lt2.ix) + 1
    from @localTable lt2 where lt2.name = lt.name) 
as nvarchar(10))
from @LocalTable lt
-- Only do those with duplicated names
where lt.NameColumn in (
  select NameColumn from @localtable group by NameColumn having count(1) > 1
)

Answer 2

如果我明白你要做什么。

WITH CTE AS
(
SELECT 
CONCAT(NameField, ROW_NUMBER()
    OVER(PARTITION BY NameField ORDER BY IdField)) AS NewName, *
FROM @LocalTable
)
UPDATE
CTE SET Name = NewName

如果您只想对重复的名称执行此操作，您可以将 COUNT(*) OVER (PARTITION BY Name) 添加到 CTE 和使用它的条件逻辑中。

Answer 3

不一定要更新table，插入的时候加个后缀不行吗？

DECLARE @LocalTable TABLE (IdField INT, NameField VARCHAR(50));
INSERT @LocalTable VALUES (1, 'Not Duplicate'), (2, 'Duplicate'), (3, 'Duplicate');

INSERT INTO TargetTable (NewIdColumn, NewNameColumn)
SELECT  IdField,
        CONCAT(NameField, 
            CASE WHEN COUNT(*) OVER(PARTITION BY NameField) > 1 
                THEN ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY IdField) 
                ELSE '' 
            END)
FROM    @LocalTable
ORDER BY IdField;

或者，您可以通过简单地将上述 select 包装在子查询中并更新它来进行更新：

DECLARE @LocalTable TABLE (IdField INT, NameField VARCHAR(50));
INSERT @LocalTable VALUES (1, 'Not Duplicate'), (2, 'Duplicate'), (3, 'Duplicate');

UPDATE  t
SET     NameField = NewNameField
FROM    
(
    SELECT  IdField, NameField,
            NewNameField = CONCAT(NameField, 
                                CASE WHEN COUNT(*) OVER(PARTITION BY NameField) > 1 
                                    THEN ROW_NUMBER() OVER(PARTITION BY NameField ORDER BY IdField) 
                                    ELSE '' 
                                END)
    FROM    @LocalTable
) AS t;

SELECT * FROM @LocalTable;

Answer 4

首先，考虑使用 Temp table 而不是临时变量。
其次，尝试将 NVARCHAR(MAX) 更改为更小的值，例如 INT
以下是在 NameField 中仅包含唯一值的代码：

CREATE TABLE #LocalTable 
(
    IdField NVARCHAR(MAX),
    NameField NVARCHAR(MAX)
)

INSERT INTO #LocalTable
VALUES (1,'A'), (2,'B'), (3,'B')

INSERT INTO TargetTable
(
    NewIdColumn,
    NewNameColumn
)
SELECT IdField, NameField
FROM #LocalTable 
WHERE IdField in (
    SELECT MIN(IdField) FROM #LocalTable 
    GROUP BY NameField
);

请注意 "IdField" 的重复记录将被忽略，不会插入目标 table。

Answer 5

尝试使用以下代码：

WITH TargetTable AS(
SELECT  *,
        ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField) AS UniqueID

FROM @LocalTable AS L
),
UpdatedData AS(
    SELECT Source.NameField, 
           ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField) AS UniqueID,
           CONCAT(Ref.NameField, ROW_NUMBER() OVER(PARTITION BY Ref.NameField ORDER BY Source.IdField)) AS AppendedNameField
    FROM @LocalTable AS Source
    INNER JOIN @LocalTable AS Ref ON Ref.NameField = Source.NameField AND Ref.IdField != Source.IdField
)
UPDATE T
SET NameField=U.AppendedNameField
FROM TargetTable AS T
JOIN UpdatedData AS U ON T.NameField=U.NameField AND T.UniqueID=U.UniqueID;

SQL 服务器：ROW_NUMBER() OVER on table 变量以修改重复列

SQL Server : ROW_NUMBER() OVER on table variable to amend duplicate columns

sql

sql-server

duplicates

row-number