Structuring SQL - sql

Question

I have the below requirement to generate a report.
TASKTYPE.TaskTypeName,TASKWIP.DMTaskState_key FROM MercuryProd.TEAMSPACE.F_DMCaseWIP WIP,
MercuryProd.TEAMSPACE.F_DMTaskWIP TASKWIP,
MercuryProd.TEAMSPACE.D_DMDataField_BM_ExternalCaseIdentifier EXTID,
MercuryProd.TEAMSPACE.D_DMTaskType TASKTYPE
WHERE WIP.DMCase_key=TASKWIP.DMCase_key
AND EXTID.BM_ExternalCaseIdentifier_key=WIP.VMAE_BM_ExternalCaseIdentifier_key
AND TASKTYPE.DMTaskType_key=TASKWIP.DMTaskType_key
AND EXTID.BM_ExternalCaseIdentifier='BMAX5C62970'
--AND TASKTYPE.DMTaskType_key=9 AND TASKWIP.DMTaskState_key=2
--AND TASKTYPE.DMTaskType_key=10 AND TASKWIP.DMTaskState_key=0
If you look at the last two lines of sql, that's critical. I need all records satisfying both condition. A case type can have multiple corresponding child records in the taskwip table. I need to filter only those cases where within the child records both criteria meets. That's task 9 with state 2 and task 10 with state 0. What I have given here is an example data for one record. There will be multiple records similarly, like for another case key, multiple child record where task 9 with state 3 not 2, and task 10 with state 2 not 0. The report should not show this record.
I am happy if you can develop a query in any of the DB language whether its slq server, Oracle, mysql. I am interested more on the logic than the language format.
Because as seen in the result set, for this case key, there is a tasktype 10 with state 0 and task type 9 with state 2.

Answer 1

The specification isn't clear; I'm guessing, and this is just a guess, that we want to return rows ONLY if BOTH of a couple specific rows exist.
One option is to use correlated subqueries in an EXISTS predicate.
for example, something like this:
TASKTYPE.TaskTypeName
, TASKWIP.DMTaskState_key
FROM MercuryProd.TEAMSPACE.F_DMCaseWIP WIP
JOIN MercuryProd.TEAMSPACE.F_DMTaskWIP TASKWIP
ON TASKWIP.DMCase_key = WIP.DMCase_key
JOIN MercuryProd.TEAMSPACE.D_DMDataField_BM_ExternalCaseIdentifier EXTID
ON EXTID.BM_ExternalCaseIdentifier_key = WIP.VMAE_BM_ExternalCaseIdentifier_key
JOIN MercuryProd.TEAMSPACE.D_DMTaskType TASKTYPE
ON TASKTYPE.DMTaskType_key = TASKWIP.DMTaskType_key
WHERE EXTID.BM_ExternalCaseIdentifier = 'BMAX5C62970'
AND EXISTS ( SELECT 1
FROM MercuryProd.TEAMSPACE.D_DMTaskType tt92
WHERE tt92.DMTaskType_key = 9
AND TASKWIP.DMTaskState_key = 2
)
AND EXISTS ( SELECT 1
FROM MercuryProd.TEAMSPACE.D_DMTaskType tt10
WHERE tt10.DMTaskType_key = 10
AND TASKWIP.DMTaskState_key = 0
)
Note that it doesn't matter what value the subqueries return, the EXISTS is just checking if at least one row is return.
Note that this doesn't restrict which rows from TASKTYPE are returned. If we want to limit the return to just specific matching rows, we can add to the ON clause of the TASKTYPE join, or to the WHERE clause ...
AND ( ( TASKTYPE.DMTaskType_key = 9 AND TASKWIP.DMTaskState_key = 2 )
OR ( TASKTYPE.DMTaskType_key = 10 AND TASKWIP.DMTaskState_key = 0 )
)
There are other query patterns we could use; we could do a single EXISTS like this:
AND EXISTS ( SELECT 1
FROM MercuryProd.TEAMSPACE.D_DMTaskType ttx
WHERE ( ttx.DMTaskType_key = 9 AND TASKWIP.DMTaskState_key = 2 )
OR ( ttx.DMTaskType_key = 10 AND TASKWIP.DMTaskState_key = 0 )
HAVING COUNT(DISTINCT ttx.DMTaskType_key) = 2
)
EDIT
The first pattern demonstrated isn't sufficient. That requires both TASKTYPE rows to be related to the same TASKWIP row, and that can't happen because each TASKTYPE row require a different value from the TASKWIP row.
We would need to do the join in the correlated subqueries.
Something along these lines:
AND EXISTS ( SELECT 1
FROM MercuryProd.TEAMSPACE.F_DMTaskWIP tw92
JOIN MercuryProd.TEAMSPACE.D_DMTaskType tt92
ON tt92.DMTaskType_key = tw92.DMTaskType_key
AND tt92.DMTaskType_key = 9
WHERE tw92.DMTaskState_key = 2
AND tw92.DMCase_key = WIP.DMCase_key
)
AND EXISTS ( SELECT 1
FROM MercuryProd.TEAMSPACE.F_DMTaskWIP tw10
JOIN MercuryProd.TEAMSPACE.D_DMTaskType tt10
ON tt10.DMTaskType_key = tw10.DMTaskType_key
AND tt10.DMTaskType_key = 10
WHERE tw10.DMTaskState_key = 0
AND tw10.DMCase_key = WIP.DMCase_key
)

Answer 2

For oracle, you can use the listagg like below
SELECT DM_Case_Key,listagg(TaskTypeName,',') within group (order by DMTaskType_key)
over (partition by DM_Case_Key) as Tasks
FROM your_data