Oracle 10g和11g r2中的数据透视

Question

I have a table that I need to pivot. I am using oracle 11g R2 on my system and the server machine has oracle 10g running on it.

My table contains data: Origin Table: ppl

ename | edate
emp1 | 01-10-17
emp1 | 01-10-17
emp1 | 01-10-17
emp1 | 02-10-17
emp2 | 01-10-17
emp3 | 01-10-17
emp3 | 01-10-17
emp3 | 02-10-17
emp3 | 02-10-17

Normally in pivot, output should be: output table 1:

ename | 01-10-17 | 02-10-17
emp1  |      3   |      1
emp2  |      1   |      0
emp3  |      2   |      2

But here there is a condition that for each day count has to be >1. So desired output will be: output table 2:

ename | 01-10-17 | 02-10-17
emp3  |      2   |      2

I used pivot in 11g to get required output and the output is partially correct.

select * from
(
  SELECT  ename, edate
  FROM ppl
)
PIVOT
(
  COUNT(edate)
  FOR edate in('01-oct-2017', '02-oct-2017')
)
order by ename;

The above query returns only the pivot in output table 1. My question 1 is where can I put filter in pivot query?

On trying to get the same output in 10g, I found out that pivot command is not supported there. So here is my query for 10g:

select ename,
  count(case when edate ='01-10-2017' then ename end) "01-OCT-2017",
  count(case when edate ='02-10-2017' then ename end) "02-OCT-2017"
from ppl
group by ename
having count(case when edate ='01-10-2017' then ename end) >1
and count(case when edate ='02-10-2017' then ename end) >1
order by ename;

This gives the exact output 2 table as desired. My question 2 is that the dates can range to 30 days even. As I can understand, for that case I need to write "count(case when edate ='01-10-2017' then ename end) "01-OCT-2017"" and "count(case when edate ='01-10-2017' then ename end) >1" 30 times. Will not such a big query will hamper the time of output? Or is there any feasible way to do this better?

Thank you for help..!!

Answer 1

Q1. The where clause may be included prior to the pivot:

select * from
(
  SELECT  ename, edate
  FROM ppl

  WHERE .......
)
PIVOT
(
  COUNT(edate)
  FOR edate in('01-oct-2017', '02-oct-2017')
)
order by ename;

Q2. Yes you can have many case expressions, it will not be a big performance hit.

Or is there any feasible way to do this better? You could implement it through "dynamic SQL" instead, but the query structure will remain the same. Dynamic sql just builds the query SQL for you.