当按其他列分组时，如何在特定列中具有非唯一值的 select 行？

Question

I have a table tbl like this:

| id | grp | pid | oid |
| -- | --- | --- | --- |
| 1  | 1   | 1   | 1   |
| 2  | 2   | 2   | 1   |
| 3  | 3   | 1   | 1   |
| 4  | 3   | 2   | 1   |
| 5  | 4   | 1   | 1   |
| 6  | 1   | 1   | 2   |
| 7  | 2   | 2   | 2   |
| 8  | 3   | 1   | 2   |
| 9  | 4   | 1   | 2   |

I am trying to write a PostgreSQL query which selects the rows where for a given GRP in a given OID, PID has a distinct count greater than 1. For the above, since PID has two distinct values (1 and 2) for GRP 3 in OID 1, it should return:

| id | grp | pid | oid |
| -- | --- | --- | --- |
| 3  | 3   | 1   | 1   |
| 4  | 3   | 2   | 1   |

I have a solution for this using Python + Pandas, though this is less than ideal:

import pandas as pd
rows = pd.read_sql("SELECT * FROM tbl", db.engine)
output = pd.DataFrame(columns = rows.columns)
oids = rows['oid'].tolist()
for oid in oids:
   oid_rows = rows[rows['oid'] == oid]
   grps = oid_rows['grp'].tolist()
   for grp in grps:
       grp_rows = oid_rows[oid_rows['grp'] == grp]
       if len(grp_rows) > 1:
           output = pd.concat([output, grp_rows],axis=0)
print(output)

I'd prefer to do this purely in SQL, essentially a query along the lines of:

SELECT * FROM tbl HAVING COUNT(pid) > 1 IN
    (SELECT * FROM tbl GROUP BY grp, oid)

How do I write this query?

Answer 1

You can use exists :

select t.*
from tbl t
where exists (select 1
              from tbl t2
              where t2.grp = t.grp and t2.oid = t.oid and
                    t2.id <> t.id
             );

You can also use window functions -- although this may be less efficient:

select t.*
from (select t.*, count(*) over (partition by grp, oid) as cnt
      from tbl t
     ) t
where cnt >= 2;