识别 SQL 查询中的重复字段

Question

我有一个返回如下列的 SQL 查询：

foo
-----------
1200
1200
1201
1200
1200
1202
1202
1202

它已经以特定方式排序，我想对该结果集执行另一个查询以标识重复数据，如下所示：

foo    ID
----   ----
1200   1
1200   1
1201   2
1200   3
1200   3
1202   4
1202   4
1202   4

重要的是，第二组 1200 与第一组是分开的。 OVER/PARTITION 的每一个变体似乎都想把两个组放在一起。有没有办法将分区窗口化到这些重复的组？

编辑： 这适用于 Microsoft SQL Server 2012

Answer 1

不确定这是最快的结果...

select main.num, main.id from
(select x.num,row_number() 
over (order by (select 0)) as id 
from (select distinct num from num) x) main
join 
(select num, row_number() over(order by (select 0)) as ordering
 from num) x2 on 
x2.num=main.num
order by x2.ordering

假设表“num”有一个包含您的数据的列“num”，按顺序 - 当然 num 可以成为原始查询的视图或“with”。

请看下面sqlfiddle

Answer 2

这是一种不用CURSOR的方法

-- Create a temporay table
DECLARE @table TABLE 
(
    SeqID INT IDENTITY(1,1) NOT NULL PRIMARY KEY,
    foo INT,    
    id int null 
)


DECLARE @i INT
DECLARE @j INT
DECLARE @k INT
declare @tFoo INT
declare @oldFoo INT

SET @k = 0
set @oldFoo = 0



-- Insert data into the temporary table
INSERT INTO @table(foo) 
SELECT 1200
INSERT INTO @table(foo) 
SELECT 1200
INSERT INTO @table(foo) 
SELECT 1201
INSERT INTO @table(foo) 
SELECT 1200
INSERT INTO @table(foo) 
SELECT 1200
INSERT INTO @table(foo) 
SELECT 1202
INSERT INTO @table(foo) 
SELECT 1202
INSERT INTO @table(foo) 
SELECT 1202

-- Get the max and min SeqIDs to loop through
SELECT @i = MIN(SeqID) FROM @table
SELECT @j = MAX(SeqID) FROM @table


-- Loop through the temp table using the SeqID indentity column
WHILE (@i <= @j)
BEGIN
  SELECT @tFoo = foo FROM @table WHERE SeqID = @i

  if @oldFoo <> @tFoo
    set @k = @k + 1 

  update @table set id = @k where SeqID = @i    

  SET @oldFoo = @tFoo
  -- Increment the counter
  SET @i = @i + 1
END

SELECT * from @table

Answer 3

你可以不使用游标来做到这一点，但它看起来不太好（至少我想出的）。所以 1）我假设你有 PK 列来排序你的主要值。 然后 2) 我假设您有一个要设置的 ID 列。

    create table tbl(foo int, pk int, id int);

    insert into tbl(foo, pk) values (1100, 5);
    insert into tbl(foo, pk) values (1200, 10);
    insert into tbl(foo, pk) values (1200, 20);
    insert into tbl(foo, pk) values (1201, 30);
    insert into tbl(foo, pk) values (1200, 40);
    insert into tbl(foo, pk) values (1200, 50);

    insert into tbl(foo, pk) values (1202, 60);
    insert into tbl(foo, pk) values (1202, 70);
    insert into tbl(foo, pk) values (1202, 80);
    insert into tbl(foo, pk) values (1202, 90);

这里的 SQL 小提琴：http://sqlfiddle.com/#!6/fdaaa/2

    update tbl
    set
    ID = 1 



    update t

    set

    t.ID = m.RN2 

    from

    tbl t 

    join 

    (


    select

    y1.RN as RN1, y1.PK as PK1,
    y2.RN as RN2, y2.PK as PK2

    FROM

    (

    SELECT

    ROW_NUMBER() OVER(ORDER BY x.pk1 ASC) AS rn,

    x.pk1 AS pk

    FROM
    (
    SELECT t1.pk AS pk1, t2.pk AS pk2
    FROM
    tbl t1
    LEFT JOIN tbl t2 ON
    (
    (t1.pk < t2.pk AND t1.foo = t2.foo)
    AND
    (
      NOT EXISTS
      (
        SELECT tMid.pk FROM
        tbl tMid WHERE
        tMid.pk < t2.pk
        AND
        tMid.pk > t1.pk
      )
    )
    )
    ) x WHERE x.pk2 IS NULL


    ) y1


    left join 




    (

    SELECT

    ROW_NUMBER() OVER(ORDER BY x.pk1 ASC) AS rn,

    x.pk1 AS pk

    FROM
    (
    SELECT t1.pk AS pk1, t2.pk AS pk2
    FROM
    tbl t1
    LEFT JOIN tbl t2 ON
    (
    (t1.pk < t2.pk AND t1.foo = t2.foo)
    AND
    (
      NOT EXISTS
      (
        SELECT tMid.pk FROM
        tbl tMid WHERE
        tMid.pk < t2.pk
        AND
        tMid.pk > t1.pk
      )
    )
    )
    ) x WHERE x.pk2 IS NULL


    ) y2 on y1.RN = y2.RN - 1 

    ) m on 

    (
    (t.pk > m.pk1 and ((m.pk2 is not null ) and (t.pk <= m.pk2)))
    -- or
    -- (t.pk<=m.pk1)
    )

Answer 4

这是我使用游标和临时表保存结果的解决方案。

DECLARE @foo INT
DECLARE @previousfoo INT = -1
DECLARE @id INT = 0
DECLARE @getid CURSOR

DECLARE @resultstable TABLE 
(
    primaryId INT IDENTITY(1, 1) NOT NULL PRIMARY KEY,
    foo INT,    
    id int null 
)

SET @getid = CURSOR FOR
SELECT originaltable.foo
FROM   originaltable

OPEN @getid
FETCH NEXT
FROM @getid INTO @foo
WHILE @@FETCH_STATUS = 0
BEGIN
    IF (@foo <> @previousfoo)
    BEGIN
        SET @id = @id + 1
    END

    INSERT INTO @resultstable VALUES (@foo, @id)
    SET @previousfoo = @foo

    FETCH NEXT
    FROM @getid INTO @foo
END

CLOSE @getid
DEALLOCATE @getid