Question

我具有以下格式的MySQL表

id     c1     c2
---------------------
id1    c1v1   c2v1  -> keep id1 even it has duplicates
id1    c1v1   c2v1
id2    c1v2   c2v2  -> filter out id2 because it has various c1, c2
id2    c1v2   c2v3
id3    c1v3   c2v4
....

预期输出：

id     c1     c2
---------------------
id1    c1v1   c2v1
id3    c1v3   c2v4
....

我只想保留id仅具有唯一的c1，c2值的记录。我确实有解决方案，但是它需要两次全表扫描，包括一次联接，效率非常低，我想知道是否有更好的方法可以做到这一点。

select 
    distinct id, c1, c2
from table 
inner join 
    (select 
        id, 
        count(distinct c1, c2) as counts
    from table 
    group by id
    having counts = 1) tmp
on table.id = tmp.id

Answer 1

仅使用group by：

select id, min(c1) as c1, min(c2) as c2
from t
group by id
having min(c1) = max(c1) and min(c2) = max(c2)

如果要获取所有原始记录，请使用not exists

select t.*
from t
where not exists (select 1
                  from t t2
                  where t2.id = t.id and
                        (t2.c1 <> t.c1 or t2.c2 <> t.c2)
                 );

Answer 2

您可以在不同的c1 / c2值上使用自我LEFT JOIN来执行此操作，仅在第二张表中保留没有匹配行的行（即，同一id的值不同）：

SELECT DISTINCT t1.id, t1.c1, t1.c2
FROM test t1
LEFT JOIN test t2 ON t2.id = t1.id AND (t2.c1 != t1.c1 OR t2.c2 != t1.c2)
WHERE t2.id IS NULL

输出：

id      c1      c2
id1     c1v1    c2v1
id3     c1v3    c2v4

Demo on dbfiddle

SQL筛选出具有多个值的ID

2 个答案: