Question

我有两个桌子

表1

col1    col2    col3    col4
v1      v2      v3      x
v4      v5      v6      y
v4      v5      v6      z
v7      v8      v9      w
v13     v14     v15     u

表2

col1    col2    col3    col5
v1      v2      v3      a
v1      v2      v3      b
v4      v5      v6      c
v10     v11     v12     d
v13     v14     v15     e

我想检查col1，col2和col3的那些表中的值是否完全相同。还有其他我不想考虑的列，例如col4。两个表中都没有主键。并且在col1，col2和col3中还会重复一些值。例如，在表1中，（v1，v2，v3）对（col1，col2，col3）重复两次。因此，这些也应在表2中重复相同的次数。

以上（仅v13，v14，v15）匹配。
（v1，v2，v3）在表2中出现两次，但在表1中出现一次。
（v4，v5，v6）在表1中出现两次，但在表2中出现一次。
（v7，v8，v9）仅出现在表1中，而不出现在表2中。
（v10，v11，v12）仅出现在表2中，而不出现在表1中。

如何在单个查询中完成此操作？

Answer 1

您可以使用full join获取例外：

select coalesce(t1.col1, t2.col1) as col1,
       coalesce(t1.col2, t2.col2) as col2,
       coalesce(t1.col3, t2.col3) as col3,
       t1.cnt, t2.cnt
from (select col1, col2, col3, count(*) as cnt
      from table1
      group by col1, col2, col3
     ) t1 inner join  
     (select col1, col2, col3, count(*) as cnt
      from table2
      group by col1, col2, col3
    ) t2
    on t2.col1 = t1.col1 and t2.col2 = t1.col2 and t2.col3 = t1.col3 
where (t2.cnt <> t1.cnt) or (t1.cnt is null) or (t2.cnt is null);

如果只希望与cnt上的那些匹配，则可以使用内部联接。

Answer 2

您可以创建两个子查询，这些子查询按1、2、3列分组并带有一个计数，然后将它们连接到列值和计数上

        select * 
  from (select col1, col2, col3, count(*) c
          from table1
         group by col1, col2, col3) t1
  join (select col1, col2, col3, count(*) c
          from table2
         group by col1, col2, col3) t2 
    on t2.col1 = t1.col1
   and t2.col2 = t2.col2
   and t2.col3 = t3.col3
   and t2.c = t1.c

Answer 3

您可以通过此查询找到所有匹配的元组：

select col1, col2, col3
  from (select col1, col2, col3, count(*) cnt 
          from table1
         group by col1, col2, col3) t1
  join (select col1, col2, col3, count(*) cnt
          from table2
         group by col1, col2, col3) t2
 using (col1, col2, col3, cnt)

Results ：

| col1 | col2 | col3 |
|------|------|------|
|  v13 |  v14 |  v15 |

然后，您可以采用上述查询的结果来分别确定原始表中不匹配的记录：

select *
  from table1
 where (col1,col2,col3) not in (  
    select col1, col2, col3
      from (select col1, col2, col3, count(*) cnt 
              from table1
             group by col1, col2, col3) t1
      join (select col1, col2, col3, count(*) cnt
              from table2
             group by col1, col2, col3) t2
      using (col1, col2, col3, cnt));

Results ：

| col1 | col2 | col3 | col4 |
|------|------|------|------|
|   v1 |   v2 |   v3 |    x |
|   v4 |   v5 |   v6 |    y |
|   v4 |   v5 |   v6 |    z |
|   v7 |   v8 |   v9 |    w |

和

select *
  from table2
 where (col1,col2,col3) not in (  
    select col1, col2, col3
      from (select col1, col2, col3, count(*) cnt 
              from table1
             group by col1, col2, col3) t1
      join (select col1, col2, col3, count(*) cnt
              from table2
             group by col1, col2, col3) t2
      using (col1, col2, col3, cnt));

Results ：

| col1 | col2 | col3 | col5 |
|------|------|------|------|
|   v1 |   v2 |   v3 |    a |
|   v1 |   v2 |   v3 |    b |
|   v4 |   v5 |   v6 |    c |
|  v10 |  v11 |  v12 |    d |

或全部在一起：

select * from (
    select 't1' src, t1.*
      from table1 t1
    union all
    select 't2' src, t1.*
      from table1 t1) u1
 where (col1,col2,col3) not in (  
    select col1, col2, col3
      from (select col1, col2, col3, count(*) cnt 
              from table1
             group by col1, col2, col3) t1
      join (select col1, col2, col3, count(*) cnt
              from table2
             group by col1, col2, col3) t2
      using (col1, col2, col3, cnt));

Results ：

| src | col1 | col2 | col3 | col4 |
|-----|------|------|------|------|
|  t1 |   v1 |   v2 |   v3 |    x |
|  t1 |   v4 |   v5 |   v6 |    y |
|  t1 |   v4 |   v5 |   v6 |    z |
|  t1 |   v7 |   v8 |   v9 |    w |
|  t2 |   v1 |   v2 |   v3 |    x |
|  t2 |   v4 |   v5 |   v6 |    y |
|  t2 |   v4 |   v5 |   v6 |    z |
|  t2 |   v7 |   v8 |   v9 |    w |

通过SQL Fiddle

观看所有操作

在SQL中查找多个列的不匹配值

3 个答案: