Sql 两个表的并集,但显示数据来自哪个表

Sql 两个表的并集,但显示数据来自哪个表,sql,union,highest,Sql,Union,Highest,我有两张桌子: TABLE_A TABLE_B Fields: Trans Amend Trans Amend data: 100 0 100 0 100 1 110 0 120 0

我有两张桌子:

           TABLE_A                 TABLE_B
Fields:    Trans   Amend           Trans   Amend 
data:      100       0             100      0
           100       1
           110       0         
                                   120      0
                                   120      1
           130       0             130      0 
                                   130      1
           140       0             140      0
           150       0             150      0
           150       1             150      1
           150       2             
我想要的是一个表(视图),它将这些表合并(并集),但只显示每个Trans的最高修正值

寻找这个答案:

Fields:    Trans   Amend   
data:      100       1
           110       0
           120       1
           130       1
           140       0 
           150       2   
Fields:    Trans   Amend    WhichTBL
data:      100       1      Table_A
           110       0      Table_A
           120       1      Table_B
           130       1      Table_B
           140       0      Table_A
           150       2      Table_A
然后,更难的是,我想知道是否有一种方法可以从哪个表中判断数据来自哪个表。当记录A和记录B相等时,表A总是获胜 寻找这个答案:

Fields:    Trans   Amend   
data:      100       1
           110       0
           120       1
           130       1
           140       0 
           150       2   
Fields:    Trans   Amend    WhichTBL
data:      100       1      Table_A
           110       0      Table_A
           120       1      Table_B
           130       1      Table_B
           140       0      Table_A
           150       2      Table_A

我知道不能通过联合来获得此结果。

在Teradata SQL中,您可以执行以下操作,但不确定SQL Server:

select trans,amend,WhichTBL from
(
select trans,amend,'Table_A' WhichTBL from Table_A
union
select trans,amend,'Table_B' WhichTBL from Table_B
) X
qualify row_number() over(partition by trans order by amend desc, WhichTBL) = 1
order by trans;
如果您的SQL没有限定子句,则使用Lucero建议的版本:

select trans,amend,WhichTBL from
(
   select x.*,row_number() over(partition by trans order by amend desc, WhichTBL) as rn
   (
      select trans,amend,'Table_A' as WhichTBL from Table_A
      union
      select trans,amend,'Table_B' as WhichTBL from Table_B
   ) Derived1 as X
) Derived2
where rn = 1
order by trans;
这样行吗

SELECT 
    trans, MAX(max_amend) as max_max_amend
FROM
    (SELECT
        'a' AS src, trans, MAX(amend) AS max_amend
    FROM
        table_a
    GROUP BY
        trans

    UNION ALL

    SELECT
        'b' AS src, trans, MAX(amend) AS max_amend
    FROM
        table_b
    GROUP BY
        trans) m
GROUP BY
    trans

Lucero下面的观点是正确的,最小值(src)应该在全局集合上,而不是相关的最大值() 我认为您必须将源值和表值合并到一个列中,以便最大化。在您的示例中,只需在值中添加1即可区分源,如:

SELECT trans, Max(amend) AS MaxOfamend, 1+[amend] AS isa, 0 AS isb
FROM TableA
GROUP BY trans
但是你可以加上100,或者乘以一个大的值,或者任何与你的数据相关的东西。其思想是将修正值和来源这两条信息合并到一列中

然后,在合并信息后,获得该值的最大值,然后通过取消组合来去除源标志(减去1,除以100,随便什么)


好的,这是我得到的:

CREATE VIEW [dbo].[viewA]    AS
SELECT trans, MAX(amend + .20) AS srcIsA, 0 AS srcIsb
FROM  dbo.tableA
GROUP BY trans

CREATE VIEW [dbo].[viewB]    AS
SELECT trans, 0 AS srcIsA, MAX(amend + .10) AS srcIsB
FROM  dbo.tableB
GROUP BY trans

CREATE VIEW [dbo].[viewU]    AS
SELECT * from viewA
union all
select *
FROM  viewb

CREATE VIEW [dbo].[viewv]    AS
SELECT trans, srcIsA, srcIsb, srcIsA + srcIsb AS total
FROM  dbo.viewU

CREATE VIEW [dbo].[vieww]    AS
SELECT trans, MAX(total) AS max_total
FROM  dbo.viewv
GROUP BY trans

CREATE VIEW [dbo].[viewx] AS
SELECT trans, 
max_total, 
CAST(max_total AS int) AS maxval, 
CASE WHEN (max_total - CAST(max_total AS int)) = .1 THEN 'a' ELSE 'b' END AS src
FROM  dbo.vieww

我想告诉您,您可以使用标准SQL通过连接和聚合实现这一点:

select coalesce(a.trans, b.trans) as trans,
       (case when coalesce(max(b.amend), -1) > coalesce(max(a.amend), -1)
             then max(b.amend)
             else max(a.amend)
        end) as amend,
       (case when coalesce(max(b.amend), -1) > coalesce(max(a.amend) , -1)
             then 'B' else 'A'
        end) as whichTable
from Table_A a full outer join
     Table_B b
     on a.trans = b.trans
group by coalesce(a.trans, b.trans)

如果修正值仅为1和0 那么第一个问题就可以解决了

select Trans,sum(Amend) AmendMax from (select Trans,Amend from TABLE_A
union select Trans,Amend from TABLE_B) C group by Trans
第二个问题是

select Trans,max(Amend) Amend,case when sum(s)=1 or  sum(s)=2 or  sum(s)=21 
then 'Table-A' when sum(s)=10 or  sum(s)=12 or sum(s)=20  then 'Table-B' 
when sum(s)=11 or  sum(s)=22 then 'Table-A and B' end s from 
(select case when max(Amend)=1 then 1 else 2 end s,Trans,max(Amend) Amend from TABLE_A 
group by Trans union select case when max(Amend)=1 
then 10 else 20 end s,Trans,max(Amend) Amend from TABLE_B group by Trans) C group by Trans

如果在select中添加一个字符串并将其别名为列,该怎么办

SELECT Trans, Amend, 'Table_A' as WhichTBL
FROM (your 1st select query)
UNION
SELECT Trans, Amend, 'Table_B' as WhichTBL
FROM (your 2nd select query)
ORDER BY Trans

GROUP BY
不会将其剪切-
MAX(MAX\u amend)
可能来自与
MIN(src)
不同的行(从而来自不同的源表),因此该行的(真实)原点将丢失(例如,在示例中,您将使用amend 1将trans 130属性化为表a)。并非所有数据库系统中都存在
QUALIFY
子句,但可以通过使用嵌套查询来模拟,该查询公开
行号()
,然后使用
WHERE
子句筛选此查询。这应标记为正确答案。但是,最好将
union
更改为
union all
(因为行从来都不是相同的,所以没有必要对它们进行重复数据消除)。错误:每当
a.amend
为NULL时(由只找到a
b.trans
的完整外部联接引起),您的
MAX(a.amend)
也将为NULL,因此,条件
max(b.amend)>max(a.amend)
将返回false,这将选择“a”表值和标识符,尽管它为NULL(哦,还要注意
'a“
字符串文本上的语法错误)。@Lucero..准确地说,max(a.amend)或max(b.amend)需要为NULL才能发生故障(当amend在数据中为NULL时略有不同)。在任何情况下,我修复了这个问题,假设amend在数据中总是非负的(就像在示例数据中一样)。在
max(b.amend)
为NULL的情况下(无意中?)之所以有效,是因为案例的
ELSE
部分是要使用的正确答案-这就是为什么我只指出了另一部分。接受的答案可能会返回错误的结果。应该使用Carlos A.Ibarra的解决方案。我在实际解决了一个解决方案后更新了我的答案。请重新考虑。