Oracle 18c - Complex sql - sql

Question

I have a table with following columns:
Emp_ID Number
Emp_flag Varchar2(1)
Date_1 Date
Date_2 Date
create_date Date
No PK on this table , there are many records with duplicates of Emp_id..
What I need to know, is when a new Date_1 is entered (so Null to a date, or from Date 1 to Date 2) on what date that happened.
I can’t just look at a single record to compare Date_1 with create_date because there are many times in the many records for a given Emp_ID when the Date_1 is simply “copied” to the new record. A Date_1 may have been originally entered on 02/15/2019 with a value of 02/01/2019. Now let’s say Date_2 gets added on 02/12/2020. So the table looks like this:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
123 Y Null Null 1/18/2018
123 Y 02/1/2019 Null 02/15/2019
123 Y 02/1/2019 02/12/2021 02/12/2020
I need a SQL query that would tell me that Emp_ID 123 had a Date_1 of 02/1/2019 entered on 02/15/2019 and NOT pick up any other record.
Expected output:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
123 Y 02/1/2019 Null 02/15/2019
Example 2 (notice date_1 is different):
Emp_ID Emp_flag Date_1 Date_2 Create_Date
456 Y Null Null 1/18/2018
456 Y 10/1/2019 Null 02/15/2019
456 Y 11/2/2019 02/12/2021 02/12/2020
Expected output:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
456 Y 10/1/2019 Null 02/15/2019
456 Y 11/2/2019 02/12/2021 02/12/2020
Example 3:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
456 Y Null Null 1/18/2018
456 Y 10/1/2019 Null 02/15/2019
456 Y 10/1/2019 Null 02/15/2019
456 Y 11/2/2019 02/12/2021 02/12/2020
Expected output:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
456 Y 10/1/2019 Null 02/15/2019
456 Y 11/2/2019 02/12/2021 02/12/2020
Example 4:
Emp_ID Emp_flag Date_1 Date_2 Create_Date
456 Y 10/1/2019 Null 02/15/2019
456 Y 10/1/2019 Null 02/16/2019
Expected output: No records.

Answer 1

You can use the Lag function to check whether the previous value of date_1 existed or not.
SELECT x.emp_id,
x.date_1,
x.create_date AS first_date_with_date_1
FROM (
SELECT t.emp_id,
t.create_date,
t.date_1,
LAG(t.date_1) OVER (PARTITION BY t.emp_id ORDER BY t.create_date) AS last_date_1
FROM your_table t
) x
WHERE x.date_1 IS NOT NULL
AND x.last_date_1 IS NULL

Answer 2

Test for all cases:
with t(emp_id, emp_flag, date_1, date_2, create_date) as (
select 101, 'Y', null, null, date '2018-01-18' from dual union all
select 101, 'Y', date '2019-02-01', null, date '2019-02-15' from dual union all
select 101, 'Y', date '2019-02-01', date '2021-02-12', date '2019-02-16' from dual union all
select 102, 'Y', null, null, date '2018-01-18' from dual union all
select 102, 'Y', date '2019-02-10', null, date '2019-02-15' from dual union all
select 102, 'Y', date '2019-02-11', date '2021-02-12', date '2019-02-16' from dual union all
select 103, 'Y', null, null, date '2018-01-18' from dual union all
select 103, 'Y', date '2019-02-10', null, date '2019-02-15' from dual union all
select 103, 'Y', date '2019-02-10', null, date '2019-02-15' from dual union all
select 103, 'Y', date '2019-02-11', date '2021-02-21', date '2020-12-02' from dual )
select emp_id, emp_flag, date_1, date_2, create_date
from (
select emp_ID, emp_flag, date_1, date_2, create_date,
lag(date_1) over (partition by emp_id order by create_date) prev_dt1
from t )
where date_1 <> nvl(prev_dt1, date_1 - 1);
Result:
EMP_ID EMP_FLAG DATE_1 DATE_2 CREATE_DATE
---------- -------- ----------- ----------- -----------
101 Y 2019-02-01 2019-02-15
102 Y 2019-02-10 2019-02-15
102 Y 2019-02-11 2021-02-12 2019-02-16
103 Y 2019-02-10 2019-02-15
103 Y 2019-02-11 2021-02-21 2020-12-02
Edit:
when there are more than one records with no change in Date_1. It
should not return a record for that Emp_id
In this case date_1 is set in first row (id 104). If you want hide rows in such case use:
with t(emp_id, emp_flag, date_1, date_2, create_date) as (
select 104, 'Y', date '2019-02-10', null, date '2019-02-15' from dual union all
select 104, 'Y', date '2019-02-10', null, date '2019-02-16' from dual union all
select 105, 'Y', date '2019-02-10', null, date '2019-02-15' from dual union all
select 105, 'Y', null, null, date '2019-02-16' from dual )
select emp_id, emp_flag, date_1, date_2, create_date
from (
select emp_ID, emp_flag, date_1, date_2, create_date,
lag(date_1) over (partition by emp_id order by create_date) prev_dt1,
row_number() over (partition by emp_id order by create_date) rn
from t )
where (date_1 is not null and prev_dt1 is null and rn > 1)
or date_1 <> prev_dt1
or date_1 is null and prev_dt1 is not null;
I also added case when previous date was set and now it is null (id 105). If it is not possible or you don't want it then remove last row.

Answer 3

You can use the lag function instead of lead here:
with tableA as
(
select 456 as Emp_ID,'Y' as Emp_flag,CAST(NUll as date) as Date_1,CAST(NULL as date) as Date_2,CAST('18Jan2018' as date) as Create_date from dual union
select 456,'Y',CAST('01Oct2019' as date),Null,CAST('15Feb2019' as date) from dual union
select 456,'Y',CAST('02Nov2019' as date),CAST('12Feb2021' as date),CAST('12Feb2020' as date) from dual)
select x.Emp_ID,x.Emp_flag,x.Date_1,x.Date_2,x.Create_date
from
(select a.*
,lag(a.date_1) Over (partition by a.Emp_ID order by a.create_date) as lag_date
from tableA a) x
where x.date_1 is not null and x.date_1<>COALESCE(x.lag_date,CAST('01Jan2100' as date))
This will give out the values only when there is a change in date_1. Since NULL comparisons won't work, I have replace them with 1/1/2100. Hope this helps.
Edit:
I checked for a sample like you mentioned and it does seem to be working. If it's not working, kindly share the expected and the result you are getting:
with tableA as
(
select 456 as Emp_ID,'Y' as Emp_flag,CAST(NUll as date) as Date_1,CAST(NULL as date) as Date_2,CAST('18Jan2018' as date) as Create_date from dual union
select 456,'Y',CAST('01Oct2019' as date),Null,CAST('15Feb2019' as date) from dual union
select 456,'Y',CAST('01Oct2019' as date),CAST('12Feb2021' as date),CAST('12Feb2020' as date) from dual)
select x.Emp_ID,x.Emp_flag,x.Date_1,x.Date_2,x.Create_date
from
(select a.*
,lag(a.date_1) Over (partition by a.Emp_ID order by a.create_date) as lag_date
from tableA a) x
where x.date_1 is not null and x.date_1<>COALESCE(x.lag_date,CAST('01Jan2100' as date))