find holidays for persons work in some branches [closed] - sql

Question

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 3 years ago.
Improve this question
I have 3 table of data: some people work in a branch and others work in 2 or more branch.
We define 2 case holiday: type 1 all branch is Closed and type 0 is some branch is Closed (not all).
Now I want to find person how is in holiday if all person-branch is Closed
Holiday
ID Date AllBranch
1 2019/02/01 1
2 2019/02/05 0
3 2019/02/06 0
BranchHoliday
ID HolidayID BranchID
1 2 2
2 2 3
3 3 2
PersonBranch
ID PersonID BranchID
1 10 2
2 11 2
3 11 3
4 12 2
5 12 4
The result I want is :
PersonID Hdate
10 2019/02/01
11 2019/02/01
12 2019/02/01
10 2019/02/05
11 2019/02/05
10 2019/02/06
In date (2019/02/01) AllBranch is 1 so all person is in holiday
In date (2019/02/05) AllBranch is 0 so:
PersonID-10 work only in BranchId 2 is define in BranchHoliday table so is in holiday
PersonID-11 work in 2 branch(2,3) and all 2,3 is define closed in BranchHoliday table so is in holiday
PersonID-12 work in 2 branch(2,4) and BranchId 4 is NOT define closed in BranchHoliday so is NOT in holiday
In date (2019/02/06) AllBranch is 0 so:
PersonID-10 work only in BranchId 2 and branch 2 is define in BranchHoliday table so is in holiday
PersonID-11 work in 2 branch(2,3) and only branch 3 is NOT define closed in BranchHoliday table so is in NOT holiday
PersonID-12 work in 2 branch(2,4) and branch 3 is NOT define closed in BranchHoliday table so is NOT in holiday
I need sql query and I am using SQL Server 2016

Answer 1

I don't know why you accepted Esteban's answer, given that your expected result is this. 6 rows only.
PersonID Hdate
10 2019/02/01
11 2019/02/01
12 2019/02/01
10 2019/02/05
11 2019/02/05
10 2019/02/06
Yet Esteban's query produces 7 rows. On February 6 holiday, it wrongly includes person #11 to allowed to have a holiday on February 6. It contradicts your problem definition:
In date (2019/02/06) AllBranch is 0 so:
* ...
* PersonID-11 work in 2 branch(2,3) and only branch 3 is NOT define closed
in BranchHoliday table so is in NOT holiday
Presumably, Person's BranchHoliday(s) should be checked(joined) against the Holiday's table, right? But Esteban's query is not doing that.
Esteban's query produces incorrect output:
Live test: http://sqlfiddle.com/#!18/86728/2
| PersonID | Date |
|----------|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
| 11 | 2019-02-06 |
And given these different data:
Holiday
ID Date AllBranch
1 2019/02/01 1
2 2019/02/05 0
3 2019/02/06 0
4 2019/04/02 0
BranchHoliday
ID HolidayID BranchID
1 2 2
2 2 3
3 3 2
4 4 2
5 4 4
Esteban's query produces incorrect output again:
Live test: http://sqlfiddle.com/#!18/10348/1
| PersonID | Date |
|----------|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 12 | 2019-02-05 |
| 10 | 2019-02-06 |
| 11 | 2019-02-06 |
| 12 | 2019-02-06 |
| 10 | 2019-04-02 |
| 11 | 2019-04-02 |
| 12 | 2019-04-02 |
It should produce this:
Live test: http://sqlfiddle.com/#!17/7b6be/1
| id | date |
|----|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
| 10 | 2019-04-02 |
| 12 | 2019-04-02 |
And given these data. Note that only persons on branch #5 are allowed to have a holiday on April 2.
Holiday
ID Date AllBranch
1 2019/02/01 1
2 2019/02/05 0
3 2019/02/06 0
4 2019/04/02 0
BranchHoliday
ID HolidayID BranchID
1 2 2
2 2 3
3 3 2
4 4 5
Esteban's query produces incorrect output again, as only the persons on branch #5 are allowed to have a holiday on April 2. Yet his query includes persons that should not be allowed to have a holiday on April 2. Persons #10 and #11 are not included on branch #5, they should not be on April 2.
Live test: http://sqlfiddle.com/#!18/fd1d3/1
| PersonID | Date |
|----------|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
| 11 | 2019-02-06 |
| 10 | 2019-04-02 |
| 11 | 2019-04-02 |
It should produce this. No employees are allowed to have a holiday on April 2, even person #10 is not allowed to have a holiday on April 2, as person #10 is on branch #2 only. Only persons on branch #5 are allowed to have a holiday.
Live test: http://sqlfiddle.com/#!17/299f73/1
| id | date |
|----|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
This is the correct query (Postgres version):
Live test: http://sqlfiddle.com/#!17/6cf97/2
select h.date, p.id
from holiday h
cross join person p
join personbranch pb
on p.id = pb.personid
left join branchholiday bh
on h.id = bh.holidayid and pb.branchid = bh.branchid
group by h.date, p.id
having
every(h.allbranch)
or
-- choose only every person whose branch ids are all present in branchholiday
-- filtered by holidayid from holiday.id
every(bh.branchid is not null)
order by h.date, p.id
Output matches your problem definition's expected output:
| id | date |
|----|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
SQL Server version:
Live test: http://sqlfiddle.com/#!18/41a54/3
select p.id, h.date
from holiday h
cross join person p
join personbranch pb
on p.id = pb.personid
left join branchholiday bh
on h.id = bh.holidayid and pb.branchid = bh.branchid
group by h.date, p.id
having
count(case when h.allbranch = 1 then 1 end) = count(*)
or
-- choose only every person whose branch ids are all present in branchholiday
-- filtered by holidayid from holiday.id
count(case when bh.branchid is not null then 1 end) = count(*)
order by h.date, p.id
Output matches your problem definition's expected output:
| id | date |
|----|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |
This is another SQL Server's every idiom. You can change the count(condition) = count(*) to min approach:
Live test: http://sqlfiddle.com/#!18/41a54/4
having
min(case when h.allbranch = 1 then 1 else 0 end) = 1
or
-- choose only every person whose branch ids are all present in branchholiday
-- filtered by holidayid from holiday.id
min(case when bh.branchid is not null then 1 else 0 end) = 1
Output matches your problem definition's expected output:
| id | date |
|----|------------|
| 10 | 2019-02-01 |
| 11 | 2019-02-01 |
| 12 | 2019-02-01 |
| 10 | 2019-02-05 |
| 11 | 2019-02-05 |
| 10 | 2019-02-06 |

Answer 2

One solution would be to break it into two parts:
First selecting the persons for the AllBranch holidays, in a second turn selecting the persons for the specific holidays, where all branches they're working in, have a defined holiday:
SELECT AllPers.PersonID, H.Date
FROM Holiday H
INNER JOIN (
SELECT DISTINCT PersonID
FROM PersonBranch
) AllPers ON H.AllBranch = 1
UNION
SELECT Sub.PersonID, H.Date
FROM Holiday H
INNER JOIN (
SELECT PB.PersonID
, SUM(IIF(BH.BranchID IS NULL, 1, 0)) AS AnyBranchOpen
FROM PersonBranch PB
LEFT JOIN BranchHoliday BH ON PB.BranchID = BH.BranchID
GROUP BY PB.PersonID
) Sub ON AnyBranchOpen = 0
WHERE H.AllBranch = 0