查询以查找两个表和两列之间的一对多关系（不包括一对一的关系）

Question

New to this forum, and new to SQL. I'm trying to clean up some data and pull out the errors. I have two tables, which both share the columns FILENAME, TYPE, SEC, I would like to pull out any records from both tables where there is a one to many relationship between SEC and TYPE, anything with a SEC and TYPE where only a one to one relationship can be ignored and is considered valid.

For Example, I have in table1.

FILENAME TYPE SEC

a----------------x----1

b----------------x----2

c----------------y----1

d----------------y----3

in table2 i would have something similar,

FILENAME TYPE SEC

e----------------x----1 

f----------------x----2

g----------------z----1

h----------------y----3

so i would like a query that would find

FILENAME TYPE SEC

a----------------x----1

c----------------y----1

e----------------x----1

g----------------z----1

My database is very large and any help would be much appreciated!

Thanks.

Answer 1

select t1.sec, t1.type, count(*) from table1 as t1 
join table2 as t2 on t1.sec = t2.sec and t1.type = t2.type
group by t1.sec, t1.type
having count(*) > 1

EDIT: That's not quite what you asked, that will show which ones have multiple records but they will be grouped.

Answer 2

You need to find the values of sec that have more than one type . The following query does this and it should work on any database:

select t.*
from ((select t1.*
       from t1
      ) union all
      (select t2.*
       from t2
      )
     ) t join
     (select sec
      from ((select t1.*
             from t1
            ) union all
            (select t2.*
             from t2
            )
           ) t
      group by sec
      having count(distinct type) > 1
     ) s
     on t.sec = s.sec

There may be more efficient ways of doing the query, depending on the database.

Answer 3

select 'table1' which, t.*
from (
    select sec, min(type) typea, max(type) typeb
    from table1
    group by sec
    having min(type) <> max(type)
) x
join table1 t on t.sec = x.sec
UNION ALL
select 'table2', t.*
from (
    select sec, min(type) typea, max(type) typeb
    from table1
    group by sec
    having min(type) <> max(type)
) x
join table2 t on t.sec = x.sec

The 2nd one just duplicates the first, combined using UNION ALL