Oracle SQL 除了确定的天数外，计算到最接近的每月 1 日

Question

I have been trying to create an Oracle SQL statement that can work based on first of the month conditions. Dates are murdering me because their formatting is very specific.

Essentially, if a date isn't the first of the month I need to count the days until the next first of the month.

So lets say if I had a wait period of 54 days and 54 days didn't land on a first of the month condition, I need to add the remaining days until the next first of the month. So it'd be 54 days PLUS the remaining days until the next first day of the month. I'm using the below statement to calculate days remaining for me.

current_date and sysdate I've been using interchangably.

ROUND((ENROLLDT-CURRENT_DATE + WAITPRD),1) AS WAIT_DAYS)

Enrolldt is a date someone "enrolls" for a service - sysdate/current_date so we can get the amount of time passed by already + waitprd [which is "activation wait period after enrollment" that is an integer value]. So for example, someone can enroll on 5/5 with a 90 day wait period which would be August 3rd but if they have a first of the month condition it would be September 1st. So if we were to do today 5/7, it'd be 88 days + the new first of the month calculation from there.

I've also been playing with ROUND((((ADD_MONTHS(TRUNC(SYSDATE, 'MM'), +1)))-SYSDATE),1) That's to calculate the remaining days from now, to the next first of the month but something isn't clicking when I try to combine these together because date formatting is new to me (obviously).

Any help or direction is greatly appreciated.

Answer 1

Rather than trying to work out a variable number of days to adjust by, you can manipulate the end of the wait period using trunc() and add_months() (or interval addition):

add_months(trunc(enrolldt + waitprd - 1, 'MM'), 1)

So with your example enrol date of 2021-05-05 and wait period of 90 days:

• enrolldt + waitprd = 2021-08-03
• enrolldt + waitprd - 1 = 2021-08-02
• trunc(enrolldt + waitprd - 1, 'MM') = 2021-08-01
• add_months(trunc(enrolldt + waitprd - 1, 'MM'), 1) = 2021-09-01

The -1 is there to handle when the wait period ends on the first of the month. For example, if the enrol date was 2021-04-08 and wait period was 54:

• enrolldt + waitprd = 2021-06-01
• enrolldt + waitprd - 1 = 2021-05-31
• trunc(enrolldt + waitprd - 1, 'MM') = 2021-05-01
• add_months(trunc(enrolldt + waitprd - 1, 'MM'), 1) = 2021-06-01

Then, if I'm following that part, your wait calculation is just:

add_months(trunc(enrolldt + waitprd - 1, 'MM'), 1) - trunc(sysdate) 

giving 177 and 25 respectively for those two examples when run today.

db<>fiddle demo with some sample values covering month end/start.

You could use a case expression to decide whether to adjust based on whether it's already on the first, but I think it's simpler to just always do it.

---

As @mathguy pointed out, if you aren't already filtering them out somehow then some past enrol date/wait period calculations could give you a negative result, and you can use greatest() to avoid negative numbers, replacing them with zero:

greatest(0, add_months(trunc(enrolldt + waitprd - 1, 'MM'), 1) - trunc(sysdate))

db<>fiddle with an example.