Support February 29 month end date for leap year

[qwertytam]

Some companies use February for their end month for the fiscal year e.g. CarMax and their 10-K with last day as Feb 29, 2016 https://www.sec.gov/Archives/edgar/data/0001170010/000117001016000117/kmx0229201610-k.htm

Currently _check_day use 2001 a non-leap year to validate the day of the month:

def _check_day(month, day):
    """Check if day is a valid day of month.
    :param month: The month to test
    :param day: The day to test
    :return: The day
    :rtype: int
    :raises TypeError: If month or day is not an int or int-like string
    :raises ValueError: If month or day is out of range
    """
    month = _check_month(month)
    day = _check_int(day)

    # Find the last day of the month
    # Use a non-leap year
    max_day = calendar.monthrange(2001, month)[1]

Resolution would be to use year if provided e.g.

def _check_day(day, month, year=None):
    """Check if day is a valid day of month.
    :param day: The day to test
    :param month: The month to test
    :param day: Optional; the year to test; defaults to 2001 a non-leap year
    :return: The day
    :rtype: int
    :raises TypeError: If month or day is not an int or int-like string
    :raises ValueError: If month or day is out of range
    """
    year = year if year else 2001
    month = _check_month(month)
    day = _check_int(day)

    # Find the last day of the month
    max_day = calendar.monthrange(year, month)[1]

Note, in this example, changed order of args (for OCD purposes :) with year last as optional

[adamjstewart]

Interesting, I didn't realize that someone might want to use a leap day/month in this way.

I wonder if we should just change 2001 to be a leap year. That would keep the code simple but also allow for this behavior. Thoughts?

[adamjstewart]

Also, the logic you pointed out is only used to define the start of a fiscal calendar, not the end. Isn't is already possible to have a fiscal calendar that ends on Fed 29th?

[nicmendoza]

If I understand the thread correctly, yes it does seem you can already have a fiscal year that ends on Feb 29th.

>>> import fiscalyear
>>> fiscalyear.setup_fiscal_calendar(start_month=3)
>>> year = fiscalyear.FiscalYear(2016)
>>> year.end
FiscalDateTime(2016, 2, 29, 23, 59, 59)

Additional thought: We could consider adding an explicit error case for anyone attempting to call it with those parameters (i.e. _check_day(2, 29)), just for the sake of clarity/completeness. Setting Feb 29th as the start date for a fiscal year seems meaningless (unless we find evidence that there are businesses which use Feb 29th as the start date in leap years and Feb 28th/March 1st in other years?). It might make sense to clarify in the comments that the _check_day function checks whether the day/month arguments are a valid start date for a fiscal year rather than "a valid day of month." And/or perhaps give it a more meaningful name.

[qwertytam]

Where I initially came across this is the data I'm working with gives fiscal year end not start. From looking through the 10-k/q's etc filed with the SEC, all of them (that I've looked through) give the date for the period ending. Perhaps this issue is better approached by being able to initialize a fiscal calendar with a period ending year/month/day? Just a thought given how the SEC financial reports are organized/reported.

[adamjstewart]

Yes, we could definitely modify setup_fiscal_calendar to accept either start_* or end_* (but not both). If you want to submit a PR to make this change I'm happy to review. I'm currently swamped with a paper deadline at the end of the month, so I may not be able to get to this anytime soon.