What's the SQL query to list all rows that have 2 column sub-rows as duplicates?

Question

I have a table that has redundant data and I'm trying to identify all rows that have duplicate sub-rows (for lack of a better word). By sub-rows I mean considering COL1 and COL2 only.

So let's say I have something like this:

 COL1   COL2   COL3
 ---------------------
 aa     111    blah_x
 aa     111    blah_j
 aa     112    blah_m
 ab     111    blah_s
 bb     112    blah_d
 bb     112    blah_d
 cc     112    blah_w
 cc     113    blah_p

I need a SQL query that returns this:

 COL1   COL2   COL3
 ---------------------
 aa     111    blah_x
 aa     111    blah_j
 bb     112    blah_d
 bb     112    blah_d

Answer 1

Does this work for you?

select t.* from table t
left join ( select col1, col2, count(*) as count from table group by col1, col2 ) c on t.col1=c.col1 and t.col2=c.col2
where c.count > 1

Answer 2

With the data you have listed, your query is not possible. The data on rows 5 & 6 is not distinct within itself.

Assuming that your table is named 'quux', if you start with something like this:

SELECT a.COL1, a.COL2, a.COL3 
FROM quux a, quux b
WHERE a.COL1 = b.COL1 AND a.COL2 = b.COL2 AND a.COL3 <> b.COL3
ORDER BY a.COL1, a.COL2

You'll end up with this answer:

 COL1   COL2   COL3
 ---------------------
 aa     111    blah_x
 aa     111    blah_j

That's because rows 5 & 6 have the same values for COL3. Any query that returns both rows 5 & 6 will also return duplicates of ALL of the rows in this dataset.

On the other hand, if you have a primary key (ID), then you can use this query instead:

SELECT a.COL1, a.COL2, a.COL3
FROM quux a, quux b
WHERE a.COL1 = b.COL1 AND a.COL2 = b.COL2 AND a.ID <> b.ID
ORDER BY a.COL1, a.COL2

[Edited to simplify the WHERE clause]

And you'll get the results you want:

COL1   COL2   COL3
---------------------
aa     111    blah_x
aa     111    blah_j
bb     112    blah_d
bb     112    blah_d

I just tested this on SQL Server 2000, but you should see the same results on any modern SQL database.

blorgbeard proved me wrong -- good for him!

Answer 3

Join on yourself like this:

SELECT a.col3, b.col3, a.col1, a.col2 
FROM tablename a, tablename b
WHERE a.col1 = b.col1 AND a.col2 = b.col2 AND a.col3 != b.col3

If you're using postgresql, you can use the oid to make it return less duplicated results, like this:

SELECT a.col3, b.col3, a.col1, a.col2 
FROM tablename a, tablename b
WHERE a.col1 = b.col1 AND a.col2 = b.col2 AND a.col3 != b.col3
  AND a.oid < b.oid

Answer 4

Don't have a database handy to test this, but I think it should work...

select
  *
from
  theTable
where
  col1 in
    (
    select
      col1
    from
      theTable
    group by
      col1||col2
    having
      count(col1||col2) > 1
    )

Answer 5

My naive attempt would be

select a.*, b.* from table a, table b where a.col1 = b.col1 and a.col2 = b.col2 and a.col3 != b.col3;

but that would return all the rows twice. I'm not sure how you'd restrict it to just returning them once. Maybe if there was a primary key, you could add "and a.pkey < b.pkey".

Like I said, that's not elegant and there is probably a better way to to do this.

Answer 6

Something like this should work:

SELECT a.COL1, a.COL2, a.COL3
FROM YourTable a
JOIN YourTable b ON b.COL1 = a.COL1 AND b.COL2 = a.COL2 AND b.COL3 <> a.COL3

In general, the JOIN clause should include every column that you're considering to be part of a "duplicate" (COL1 and COL2 in this case), and at least one column (or as many as it takes) to eliminate a row joining to itself (COL3, in this case).

Answer 7

This is pretty similar to the self-join, except it will not have the duplicates.

select COL1,COL2,COL3
from theTable a
where exists (select 'x'
              from theTable b
              where a.col1=b.col1
              and   a.col2=b.col2
              and   a.col3<>b.col3)
order by col1,col2,col3

Answer 8

Here is how you find duplicates. Tested in oracle 10g with your data.

select * from tst where (col1, col2) in (select col1, col2 from tst group by col1, col2 having count(*) > 1)

Answer 9

select COL1,COL2,COL3

from table

group by COL1,COL2,COL3

having count(*)>1

Answer 10

Forget joins -- use an analytic function:

select col1, col2, col3
from
(
select col1, col2, col3, count(*) over (partition by col1, col2) rows_per_col1_col2
from table
)
where rows_per_col1_col2 > 1