UNION ALL in CTE - sql

I'm trying to get a UNION ALL working with a CTE which I'm using for paging. I need to get all records matching a set of criteria from two tables, then page the results. The first table's CTE looks like this:
;WITH Results_CTE AS (SELECT t1.SomeIntKey1, ROW_NUMBER() OVER (ORDER BY SomeIntKey1) AS RowNum
FROM Table1 t1
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey1
WHERE Postcode LIKE 'CHX 1XX%'
) SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.SomeIntKey1 = d.Key WHERE RowNum >= 0 AND RowNum <= 10 OPTION(RECOMPILE)
The second table's paging SQL (which works fine) is:
;WITH Results_CTE AS (SELECT t2.SomeIntKey2, ROW_NUMBER() OVER (ORDER BY SomeIntKey2) AS RowNum
FROM Table2 t2
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey2
WHERE Postcode LIKE 'CHX 1XX%'
) SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.SomeIntKey2 = d.Key WHERE RowNum >= 0 AND RowNum <= 10 OPTION(RECOMPILE)
For the combined paged data, I've tried something like:
;WITH Results_CTE AS (SELECT t2.SomeIntKey2, ROW_NUMBER() OVER (ORDER BY SomeIntKey2) AS RowNum
FROM Table2 t2
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey2
WHERE Postcode LIKE 'CHX 1XX%'
UNION ALL
SELECT t1.SomeIntKey1
FROM Table1 t1
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey1
WHERE Postcode LIKE 'CHX 1XX%'
) SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.SomeIntKey2 = d.Key WHERE RowNum >= 0 AND RowNum <= 10 OPTION(RECOMPILE)
However, this results in an error:
All queries combined using a UNION, INTERSECT or EXCEPT operator must have an equal number of expressions in their target lists.
I know UNION ALLs can be confusing at the best of times, especially with joins, but I'm essentially getting a list of INT keys from two tables, then joining them to a third which contains the data I need (keys from both tables will be present in the joined column on the Data table.

You need to make sure that both result sets have the same columns:
WITH Results_CTE AS
(
SELECT
t2.SomeIntKey2 as Key,
ROW_NUMBER() OVER (ORDER BY SomeIntKey2) AS RowNum
FROM
Table2 t2
LEFT JOIN CalculatedData d
ON d.Key = t1.SomeIntKey2
WHERE Postcode LIKE 'CHX 1XX%'
UNION ALL
SELECT
t1.SomeIntKey1 as Key,
0 as RowNum
FROM
Table1 t1
LEFT JOIN CalculatedData d
ON d.Key = t1.SomeIntKey1
WHERE Postcode LIKE 'CHX 1XX%'
)
SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.SomeIntKey2 = d.Key
WHERE RowNum >= 0 AND RowNum <= 10 OPTION(RECOMPILE)
Please note that the second part of the UNION ALL now always returns 0 for the RowNum. If you want to have the RowNum column for the result of the UNION ALL you need another sub query:
WITH Results_CTE AS
(
SELECT
s.Key,
ROW_NUMBER() OVER (ORDER BY s.Key) AS RowNum
FROM
(
SELECT
t2.SomeIntKey2 as Key,
FROM
Table2 t2
LEFT JOIN CalculatedData d
ON d.Key = t1.SomeIntKey2
WHERE Postcode LIKE 'CHX 1XX%'
UNION ALL
SELECT
t1.SomeIntKey1 as Key
FROM
Table1 t1
LEFT JOIN CalculatedData d
ON d.Key = t1.SomeIntKey1
WHERE Postcode LIKE 'CHX 1XX%'
)
)
SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.SomeIntKey2 = d.Key
WHERE RowNum >= 0 AND RowNum <= 10 OPTION(RECOMPILE)

Your second SELECT in the your UNION ALL doesn't return the same number of columns, hence the problem.
Try:
;WITH Results_CTE AS (
SELECT IntKey, ROW_NUMBER() OVER (ORDER BY IntKey) AS RowNum
FROM
(
SELECT t2.SomeIntKey2 AS IntKey
FROM Table2 t2
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey2
WHERE Postcode LIKE 'CHX 1XX%'
UNION ALL
SELECT t1.SomeIntKey1 AS IntKey
FROM Table1 t1
LEFT JOIN CalculatedData d ON d.Key = t1.SomeIntKey1
WHERE Postcode LIKE 'CHX 1XX%'
) t
)
SELECT * FROM Results_CTE a INNER JOIN CalclatedData d ON a.IntKey = d.Key

Related

Join to only work if a record present in SQL

I have 2 tables - table1 and table2.
In table2, there are multiple records matching criteria for table1 based on:
c_type, h_level, loop, e_id
I want records from RIGHT table ONLY if there is EXACT 1 match. If not, element_nm should be NULL, so I have the exact same records in the output as left table.
SELECT a.*,
b.element_nm
FROM table1 a
LEFT JOIN table2 b ON
a.c_type = b.c_type
AND a.h_level = b.h_level
AND a.loop = b.loop
AND a.e_id = b.e_id
ORDER BY a.file_name,
a.line_num asc;
As this is about one value only, you can use a Sub query in the select clause. Otherwise you'd use a subquery in a LEFT OUTER JOIN or use OUTER APPLY.
SELECT
t1.*,
(
SELECT MIN(t2.element_nm)
FROM table2 t2
WHERE t2.c_type = t1.c_type
AND t2.h_level = t1.h_level
AND t2.loop = t1.loop
AND t2.e_id = t1.e_id
HAVING COUNT(*) = 1
) AS element_nm
FROM table1 t1
ORDER BY t1.file_name, t1.line_num;
Thorsten's answer works when you want only one column from the second table. But if you want multiple columns, it is a bit cumbersome.
Alternatively:
SELECT a.*, b.*
FROM table1 a LEFT JOIN
(SELECT b.*,
COUNT(*) OVER (PARTITION BY b.c_type, b.h_level, b.loop, b.e_id) as cnt
FROM b
) b
ON a.c_type = b.c_type AND
a.h_level = b.h_level AND
a.loop = b.loop AND
a.e_id = b.e_id AND
b.cnt = 1
ORDER BY a.file_name, a.line_num asc;
you should use Row_Number , like this :
WITH cte AS (
SELECT ROW_NUMBER() OVER(PARTITION BY a.c_type ,a.h_level,a.loop ,a.e_id ) rnum
,a.c_type ,a.h_level,a.loop ,a.e_id FROM table1 a
LEFT JOIN table2 b ON a.c_type = b.c_type AND a.h_level = b.h_level AND a.loop = b.loop AND a.e_id = b.e_id
)
,cte2 AS (SELECT * FROM cte WHERE rnum = 2)
SELECT a.*,
CASE WHEN cte2.element_nm IS NULL then b.element_nm ELSE NULL END element_nm
FROM table1 a
LEFT JOIN table2 b ON a.c_type = b.c_type AND a.h_level = b.h_level AND a.loop = b.loop AND a.e_id = b.e_id
LEFT JOIN cte2 ON a.c_type = cte2.c_type AND a.h_level = cte2.h_level AND a.loop = cte2.loop AND a.e_id = cte2.e_id

Count of matching IDs with in C as query

I'm looking to add a column to display a count of all records where the drgpackid matches.
Essentially I want one line in the example provided and a count of how many records have that ID and meet the conditions of the query.
with C as (
select t1.*
from DrgPack t1 join
DrgPack t2
on t1.DrgID = t2.DrgID and t1.CentralMaintFieldMask <> t2.CentralMaintFieldMask
)
select *
from rxworkflowpack
where drgpackid in (select ID from c where CentralMaintFieldMask = 0)
There are a thousand ways to do this, like adding another CTE with the counts and joining to it
with C as (
select t1.*
from DrgPack t1 join
DrgPack t2
on t1.DrgID = t2.DrgID and t1.CentralMaintFieldMask <> t2.CentralMaintFieldMask
),
D as (
select drgpackid, count(*) from rxworkflowpack group by drgpackid)
select *
from rxworkflowpack left join D on rxworkflowpack.drgpackid = d.drgpackid
where drgpackid in (select ID from c where CentralMaintFieldMask = 0)
You can use window function like this:
with C as (
select t1.*
from DrgPack t1 join
DrgPack t2
on t1.DrgID = t2.DrgID and t1.CentralMaintFieldMask <> t2.CentralMaintFieldMask
)
select DISTINCT *, COUNT(*) OVER (PARTITION BY drgpackid) AS CountRecords from rxworkflowpack
where drgpackid in (select ID from c where CentralMaintFieldMask = 0)
You should use < to not double count
select t1.drgpackid, count(*) as cnt
from DrgPack t1
join DrgPack t2
on t1.DrgID = t2.DrgID
and t1.CentralMaintFieldMask < t2.CentralMaintFieldMask
join rxworkflowpack
on rx.ID = t1.drgpackid
and rx.CentralMaintFieldMask = 0
group by t1.drgpackid

Getting Count() in CTE results in a table scan

I have the following query:
;WITH CTE
AS (
SELECT ROW_NUMBER() OVER (
ORDER BY RTT.DAYSTOWAIT DESC
) AS ROW
,COUNT(*) OVER () AS ROWCNT
--...
-- ADDITIONAL COLUMNS
--...
FROM TABLE1 T1
INNER JOIN TABLE2 T2 ON T1.OID = T2.OID
INNER JOIN ORG ON T1.ORGOID = ORG.OID
INNER JOIN EPI E ON E.OID = T1.EOID
AND E.PKEY = T1.PKEY
AND E.STATUS = 'A'
INNER JOIN PATHWAY ON OID = EPI.PATHWAYOID
AND PATHWAY.PARTKEY = T1.PARTKEY
AND PATPATHWAY.STATUS = 'A'
)
SELECT CTE.ROW
,CTE.ROWCNT
FROM CTE
When Selecting Cte with Cte.RowCount it leads to a Table scan,
but when I remove the "RowCount" column from select, it works fine (seeks).
Is there a more efficient way to get the count()?

Alternative for union in Oracle

Is there any alternate way to fetch the following data without using union?
select A.name,A.age,B.Address,C.phoneNo from table1 A,Table2 B,Table3 C where a.pkId = b.FkId and b.pkId = c.FkId
union
select A.name,A.age,B.Address,C.phoneNo from table4 A,Table5 B,Table3 C where a.pkId = b.FkId and b.pkId = c.FkId
I am using this in Hibernate and unfortunately hibernate doesnt support Union. I was just wondering if there is any other way to achieve it else ill have to write it in a procedure and save the data in temp table and fire a sql to read data from that temp table
There is an alternative for union, but it is not pretty:
select distinct coalesce(x1.name, x2.name) as name,
coalesce(x1.age, x2.age) as age,
coalesce(x1.Address, x2.Address) as age,
coalesce(x1.phoneNo, x2.phoneNo) as age,
from (select A.name, A.age, B.Address, C.phoneNo
from table1 A join
Table2 B
on a.pkId = b.FkId join
Table3 C
on b.pkId = c.FkId
) x1 full outer join
(select A.name, A.age, B.Address, C.phoneNo
from table4 A join
Table5 B
on a.pkId = b.FkId join
Table3 C
on b.pkId = c.FkId
) x2
on 1 = 0; -- always false
I can't imagine why you would want to express a union like this. I would highly recommend, though, that you start using proper, explicit join syntax.
is this working ?
SELECT
CASE DISTINCT_FLG WHEN 1 THEN nameA ELSE nameB END name,
CASE DISTINCT_FLG WHEN 1 THEN ageA ELSE ageB END age,
CASE DISTINCT_FLG WHEN 1 THEN AddressA ELSE AddressB END Address,
CASE DISTINCT_FLG WHEN 1 THEN phoneNoA ELSE phoneNoB END phoneNo
FROM (
SELECT
T1.name AS nameA, T1.age AS ageA, T2.Address AS AddressA, T3.phoneNo AS phoneNoA,
T4.name AS nameB, T4.age AS ageB, T5.Address AS AddressB, T3.phoneNo AS phoneNoB,
ROW_NUMBER() OVER(PARTITION BY T1.name, T1.age, T2.Address, T4.name, T4.age, T5.Address, T3.phoneNo ORDER BY NULL) AS DISTINCT_FLG
FROM
table1 T1,
table2 T2,
table4 T4,
table5 T5,
table3 T3
WHERE
T1.pkId = T2.FkId AND
T4.pkId = T5.FkId AND
(
T2.pkId = T3.FkId OR
T5.pkId = T3.FkId
)
) WHERE DISTINCT_FLG IN (1, 2)
Those two UNION parts have Table3 C in common, so we can join the rest to it. To emulate UNION records from the source table can be replicated through unconditional cross join of an auxiliary table with the required number or rows:
Select Distinct
CASE R.r WHEN 1 THEN A1.name ELSE A2.name END As name ,
CASE R.r WHEN 1 THEN A1.age ELSE A2.age END As age ,
CASE R.r WHEN 1 THEN B1.Address ELSE B2.Address END As Address,
C.phoneNo
From Table3 C, --< Start at common Table3
(Select Rownum r From USER_TABLES Where Rownum < 3) R --< Two rows to replicate Table3 C
-- Any table with more than one row will do
-- USER_TABLES should have enough rows in this particular case
Left Join Table2 B1 On R.r = 1 AND B1.pkId = C.FkId --< Left Join branch one
Left Join table1 A1 On R.r = 1 AND A1.pkId = B1.FkId
Left Join Table5 B2 On R.r = 2 AND B2.pkId = C.FkId --< Left Join branch two
Left Join table4 A2 On R.r = 2 AND A2.pkId = B2.FkId
Where (R.r = 1 AND A1.pkId Is NOT NULL) --/ Make sure we have values
OR (R.r = 2 AND A2.pkId Is NOT NULL) --\ for the branch
But really, consider a view.

Select group with maximum value of aggregate function

I have this query
select dca_sif, maxupp
from (select d.sifra dca_sif , sum(col1) as maxup
from tableD d join tablev v on d.id=v.d_id
join tableT t on v.id=t.v_id
group by d.sif
order by maxup desc
)
where rownum=1;
This returns first value, but If I have more groups with same maxup, how can I return all of them?
If I delete rownum=1, it returns all groups and coresponding maxup.
I want to return max(sum(col1)) but this cause error. Database is oracle.
You can try somthing like this:-
SELECT dca_sif, MAX(maxup)
FROM (SELECT d.sifra dca_sif , SUM(col1) AS maxup
FROM tableD d JOIN tablev v ON d.id=v.d_id
JOIN tableT t ON v.id=t.v_id
GROUP BY d.sif
ORDER BY maxup DESC
)
WHERE ROWNUM=1
GROUP BY dca_sif;
This might be helpful to you.
You can combine your select with the same select, but without the rownum limitation. Than join on the max id. There are many tables in your statement, so it is difficult to test for me, but this should work:
SELECT count(qry2.dcasif) || ' groups', sum( qry2.maxup2)
FROM ( SELECT dca_sif, MAX (maxup) AS maxup1
FROM ( SELECT d.sifra dca_sif, SUM (col1) AS maxup
FROM tabled d
JOIN tablev v ON d.id = v.d_id
JOIN tablet t ON v.id = t.v_id
GROUP BY d.sif
ORDER BY maxup DESC)
WHERE ROWNUM = 1
GROUP BY dca_sif) qry1
, ( SELECT dca_sif, MAX (maxup) AS maxup2
FROM ( SELECT d.sifra dca_sif, SUM (col1) AS maxup
FROM tabled d
JOIN tablev v ON d.id = v.d_id
JOIN tablet t ON v.id = t.v_id
GROUP BY d.sif
ORDER BY maxup DESC)
GROUP BY dca_sif) qry2
WHERE qry1.maxup1 = qry2.maxup2
SELECT dca_sif, maxup
FROM
(SELECT a.*,rank() over(order by maxup desc)as Rank
FROM (SELECT d.sifra dca_sif , SUM(col1) AS maxup
FROM tableD d JOIN tablev v ON d.id=v.d_id
JOIN tableT t ON v.id=t.v_id
GROUP BY d.sif
)a)
WHERE Rank=1