Extracting date field - sql

Question

I need to pull the month field from the first table and I would like the results to look like the second table.
I know to use EXTRACT(MONTH FROM BEGIN_DATE) to get the month but I'm unsure how to show all of the date fields. Any ideas? Using Oracle SQL.

Answer 1

Here is one way to do this. Note that I changed the column name from Month to mth ("month" is an Oracle keyword, it shouldn't be used as a table or column name). I also show the full month description (that is, including the year), so that there is no confusion when you have the same month (September for example) for the same person_id more than once. You can change that to show just the month by adjusting the format model in TO_CHAR, but that doesn't make much sense to me.
The test data is in the WITH clause which is NOT part of the SQL query; remove it, and use the actual table name in the SELECT query (instead of the made-up name I created in the WITH clause).
with
inputs ( person_id, begin_date, end_date ) as (
select 123, date '2017-11-01', date '2199-12-31' from dual union all
select 654, date '2017-09-01', date '2017-10-31' from dual union all
select 789, date '2017-12-01', date '2199-12-31' from dual
)
select person_id,
to_char(add_months(trunc(begin_date, 'mm'), level - 1), 'mm-yyyy') as mth,
case when add_months(trunc(begin_date, 'mm'), level - 1) <= end_date
then 'Y' else 'N' end as yn
from inputs
connect by add_months(trunc(begin_date, 'mm'), level - 1) <= sysdate
and prior person_id = person_id
and prior sys_guid() is not null
order by person_id, level
;
PERSON_ID MTH YN
---------- ------- --
123 11-2017 Y
123 12-2017 Y
123 01-2018 Y
654 09-2017 Y
654 10-2017 Y
654 11-2017 N
654 12-2017 N
654 01-2018 N
789 12-2017 Y
789 01-2018 Y
10 rows selected
One more thing - in the sample data, your begin_date is always the beginning of a month. If this is guaranteed, then you could simplify the code - there would be no need for trunc( ..., 'mm') - but I would still not change it; you don't know if this special property of the column will continue into the future, and if it ever changes, you will be glad the code addresses it already.

Answer 2

I think I finally understood what you want; it is SYSDATE you're running the query on. Based on it, you want to get all months that are between BEGIN_DATE and SYSDATE, for each PERSON_ID.
If that's so, here you go:
SQL> with test (person_id, begin_date) as
2 (select 123, date '2017-11-01' from dual union
3 select 654, date '2017-09-01' from dual
4 )
5 select person_id,
6 to_char(add_months(trunc(begin_date, 'mm'), column_value - 1), 'mm') mon
7 from test,
8 table(cast(multiset
9 (select level lvl from dual
10 connect by level <= round(months_Between (sysdate, begin_date)) + 1)
11 as sys.odcinumberlist))
12 order by 1, 2;
PERSON_ID MO
---------- --
123 01
123 11
123 12
654 01
654 09
654 10
654 11
654 12
8 rows selected.
SQL>

Answer 3

A variation on mathguy's answer:
SELECT
person_id
, TO_CHAR(ADD_MONTHS(begin_date, LEVEL-1), 'mm') m
, TO_CHAR(ADD_MONTHS(begin_date, LEVEL-1), 'yyyy') y
, CASE WHEN ADD_MONTHS(begin_date, LEVEL-1) BETWEEN begin_date and end_date THEN 'Y' ELSE 'N' END AS yn
FROM
person_table
CONNECT BY
LEVEL <= MONTHS_BETWEEN(TRUNC(SYSDATE, 'MON'), TRUNC(begin_date, 'MON')) + 1
AND prior person_id = person_id
AND prior sys_guid() IS NOT NULL
ORDER BY
person_id, y, m;
SQLFiddle for reference.