Check for multiple distinct occurrences by group - sql

Question

Input Data
Some_Table
ST_Field1 ST_Field2
Apple A
Apple A
Apple D
Orange D
Orange E
Orange Z
Pear D
Pear G
Pear C
Reference_Table
RT_Field1 RT_Field2
1 A
1 B
1 C
2 D
2 E
2 F
3 G
Expected Result:
ST_Field1 ST_Field2
Orange D
Orange E
CREATE TABLE SOME_TABLE
( ST_Field1 VARCHAR(100),
ST_Field2 VARCHAR(100)
);
INSERT INTO SOME_TABLE VALUES ('Apple','A');
INSERT INTO SOME_TABLE VALUES ('Apple','A');
INSERT INTO SOME_TABLE VALUES ('Apple','D');
INSERT INTO SOME_TABLE VALUES ('Orange','D');
INSERT INTO SOME_TABLE VALUES ('Orange','E');
INSERT INTO SOME_TABLE VALUES ('Orange','Z');
INSERT INTO SOME_TABLE VALUES ('Pear','D');
INSERT INTO SOME_TABLE VALUES ('Pear','G');
INSERT INTO SOME_TABLE VALUES ('Pear','C');
CREATE TABLE REFERENCE_TABLE
( RT_Field1 INTEGER,
RT_Field2 VARCHAR(100)
);
INSERT INTO REFERENCE_TABLE VALUES (1,'A');
INSERT INTO REFERENCE_TABLE VALUES (1,'B');
INSERT INTO REFERENCE_TABLE VALUES (1,'C');
INSERT INTO REFERENCE_TABLE VALUES (2,'D');
INSERT INTO REFERENCE_TABLE VALUES (2,'E');
INSERT INTO REFERENCE_TABLE VALUES (2,'F');
INSERT INTO REFERENCE_TABLE VALUES (3,'G');
It can be assumed that RT_Field2 is unique.
I'm looking to get the records from Some_Table which have multiple distinct values from RT_Field2, group by RT_Field1, by ST_Field1.
So from the reference table {A,B,C} are a grouping. I want to see if for a given ST_Field1 I see either {A,B};{B,c},{A,C}. I don't, I see A and C present, but across Apple and Pear.
The only success is Orange, where I'm looking for {D,E},{D,F}, or {E,F} and find D and E both for Orange.
I have:
WITH DUP_VALUES_RTF2 AS
( SELECT *
FROM (SELECT DST.ST_Field1,
DST.ST_Field2,
COUNT(1) OVER (PARTITION BY RT.RT_Field1) cnt_RTF1
FROM (SELECT DISTINCT
ST_Field1,
ST_Field2
FROM Some_Table
) DST
INNER
JOIN REFERENCE_TABLE RT
ON DST.ST_Field2 = RT.RT_Field2
) TMP
WHERE cnt_RTF1 > 1
)
SELECT *
FROM SOME_TABLE ST
WHERE EXISTS
( SELECT 1
FROM DUP_VALUES_RTF2 DVR
WHERE ST.ST_Field1 = DVR.ST_Field1
AND ST.ST_Field2 = DVR.ST_Field2
);
Which doesn't even really come close because it doesn't handle the grouping at all correctly and is really ugly. Maybe I'm just going brain dead after 5pm.

Answer 1

If I understand this correctly, you want to match on st_field1 and rt_field1, looking for duplicates in rt_field2. You can use window functions for this:
select s.*
from (select s.*, rt.rt_field1, rt.rt_field2,
min(rt.rt_field2) over (partition by s.st_field1, r.rt_field1) as min_rt2,
max(rt.rt_field2) over (partition by s.st_field1, r.rt_field1) as max_rt2
from sometable s join
reference_table r
on s.st_field2 = r.rt_field2
) s
where min_rt2 <> max_rt2;

Answer 2

you can try something like below
; with distinctSet as
(select distinct s.*,RT_Field1 from SOME_TABLE s join REFERENCE_TABLE r on s.ST_Field2=r.RT_Field2
)
,
uniqueSet as
(
select RT_Field1,ST_Field1
from distinctSet
group by RT_Field1,ST_Field1
having count(1)>1
),
resultSet as
(
select
distinct
s.*
from SOME_TABLE s
join REFERENCE_TABLE r
on s.ST_Field2=r.RT_Field2
join uniqueSet u
on u.RT_Field1=r.RT_Field1
and u.ST_Field1=s.ST_Field1
)
select * from resultSet