Oracle PIVOT - unused columns affect rollup - oracle

Question

I've got a dataset that consists of several key fields, a status, and a count. PIVOT does a nice job of giving me one row per key field combo, with one column per status. Sample below:
with dataquery (field1, field2, field3, my_status, statcount) as (
select 'AAA', 'BBB', 'CCC', 'XX', 11 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 22 from dual
union
select 'AAA', 'BBB', 'CCC', 'ZZ', 33 from dual
),
pivotquery as (
SELECT field1,
field2,
field3,
xx_stat_count AS xx,
yy_stat_count AS yy,
zz_stat_count AS zz
FROM dataquery
PIVOT (
SUM (statcount) AS stat_count
FOR (my_status)
IN ('XX' AS XX,
'YY' AS yy,
'ZZ' AS zz
)
)
)
select * from pivotquery;
This gives the expected results:
FIELD1 FIELD2 FIELD3 XX YY ZZ
AAA BBB CCC 11 22 33
If I have extra columns in the data, however, it does not ignore them and roll things up as I had expected. It seems to be trying to use the unused columns as part of the grouping logic, rather than ignoring them. If I change the input to this:
with dataquery (field1, field2, field3, my_status, statcount, cnt2, cnt3, cnt4) as (
select 'AAA', 'BBB', 'CCC', 'XX', 11, 12, 13, 14 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 22, 22, 23, 24 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'ZZ', 33, 32, 33, 34 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 122, 32, 33, 34 as cnt4 from dual
)
I get this:
FIELD1 FIELD2 FIELD3 XX YY ZZ
AAA BBB CCC 22
AAA BBB CCC 122 33
AAA BBB CCC 11
Instead of the expected
FIELD1 FIELD2 FIELD3 XX YY ZZ
AAA BBB CCC 11 144 33
If I change the input data further, so there is no repetition of those other fields, it shows up as 4 lines:
with dataquery (field1, field2, field3, my_status, statcount, cnt2, cnt3, cnt4) as (
select 'AAA', 'BBB', 'CCC', 'XX', 11, 12, 13, 14 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 22, 22, 23, 24 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'ZZ', 33, 32, 33, 34 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 122, 132, 33, 34 as cnt4 from dual
)
gives
FIELD1 FIELD2 FIELD3 XX YY ZZ
AAA BBB CCC 22
AAA BBB CCC 33
AAA BBB CCC 122
AAA BBB CCC 11
Is this expected behavior?
I can work around it, either by adding a subquery beforehand that selects just field1, field2, field3, mystatus and statcount to be processed by the PIVOT, or doing a select / group by afterward that gets rolls things up.

Answer 1

Is this expected behavior?
Yes, from the SELECT documentation:
The pivot_clause computes the aggregation functions specified at the beginning of the clause. Aggregation functions must specify a GROUP BY clause to return multiple values, yet the pivot_clause does not contain an explicit GROUP BY clause. Instead, the pivot_clause performs an implicit GROUP BY. The implicit grouping is based on all the columns not referred to in the pivot_clause, along with the set of values specified in the pivot_in_clause.). If you specify more than one aggregation function, then you must provide aliases for at least all but one of the aggregation functions.
To solve it, use a sub-query to only SELECT the required columns so that the other columns are not used implicitly as the GROUP BY clause:
with dataquery (field1, field2, field3, my_status, statcount, cnt2, cnt3, cnt4) as (
select 'AAA', 'BBB', 'CCC', 'XX', 11, 12, 13, 14 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 22, 22, 23, 24 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'ZZ', 33, 32, 33, 34 as cnt4 from dual
union
select 'AAA', 'BBB', 'CCC', 'YY', 122, 132, 33, 34 as cnt4 from dual
)
SELECT field1,
field2,
field3,
xx_stat_count AS xx,
yy_stat_count AS yy,
zz_stat_count AS zz
FROM (
SELECT field1,
field2,
field3,
my_status,
statcount
FROM dataquery
)
PIVOT (
SUM (statcount) AS stat_count
FOR (my_status) IN (
'XX' AS XX,
'YY' AS yy,
'ZZ' AS zz
)
);
Outputs:
FIELD1
FIELD2
FIELD3
XX
YY
ZZ
AAA
BBB
CCC
11
144
33
db<>fiddle here