dateutils rrule returns dates that 2 months apart

Question

I am new to Python and also dateutil module. I am passing the following arguments:

disclosure_start_date = resultsDict['fd_disclosure_start_date']
disclosure_end_date = datetime.datetime.now()
disclosure_dates = [dt for dt in rrule(MONTHLY, dtstart=disclosure_start_date, until=disclosure_end_date)]

Here disclosure_start_date = 2012-10-31 00:00:00 which converted to datetime is datetime.datetime(2012, 10, 31, 0, 0)

End date is as of now.

When I use:

disclosure_dates = [dt for dt in rrule(MONTHLY, dtstart=disclosure_start_date, until=disclosure_end_date)]

I get the dates for every other month or 2 months apart. The result is:

>>> list(disclosure_dates)
[datetime.datetime(2012, 10, 31, 0, 0), 
 datetime.datetime(2012, 12, 31, 0, 0), 
 datetime.datetime(2013, 1, 31, 0, 0), 
 datetime.datetime(2013, 3, 31, 0, 0), 
 datetime.datetime(2013, 5, 31, 0, 0), 
 datetime.datetime(2013, 7, 31, 0, 0), 
 datetime.datetime(2013, 8, 31, 0, 0), 
 datetime.datetime(2013, 10, 31, 0, 0), 
 datetime.datetime(2013, 12, 31, 0, 0), 
 datetime.datetime(2014, 1, 31, 0, 0), 
 datetime.datetime(2014, 3, 31, 0, 0), 
 datetime.datetime(2014, 5, 31, 0, 0), 
 datetime.datetime(2014, 7, 31, 0, 0), 
 datetime.datetime(2014, 8, 31, 0, 0), 
 datetime.datetime(2014, 10, 31, 0, 0), 
 datetime.datetime(2014, 12, 31, 0, 0), 
 datetime.datetime(2015, 1, 31, 0, 0), 
 datetime.datetime(2015, 3, 31, 0, 0), 
 datetime.datetime(2015, 5, 31, 0, 0), 
 datetime.datetime(2015, 7, 31, 0, 0), 
 datetime.datetime(2015, 8, 31, 0, 0), 
 datetime.datetime(2015, 10, 31, 0, 0), 
 datetime.datetime(2015, 12, 31, 0, 0), 
 datetime.datetime(2016, 1, 31, 0, 0), 
 datetime.datetime(2016, 3, 31, 0, 0), 
 datetime.datetime(2016, 5, 31, 0, 0)]

I am not sure what I am doing wrong. Can someone please point out the mistake here?

Microphone

Answer 1

The issue you are coming up against comes from the fact that datetime.datetime(2012, 10, 31, 0, 0) is the 31st of the month, and not all months have a 31st. Since the rrule module is an implementation of RFC 2445. Per RFC 3.3.10:

Recurrence rules may generate recurrence instances with an invalid date (e.g., February 30) or nonexistent local time (e.g., 1:30 AM on a day where the local time is moved forward by an hour at 1:00 AM). Such recurrence instances MUST be ignored and MUST NOT be counted as part of the recurrence set.

Since you have a monthly rule that generates the 31st of a month, it will skip all months with 30 or fewer days. You can see this bug report in dateutil about this issue.

If you just want the last day of the month, you should use the bymonthday=-1 argument:

from dateutil.rrule import rrule, MONTHLY
from datetime import datetime

disclosure_start_date = datetime(2012, 10, 31, 0, 0)

rr = rrule(freq=MONTHLY, dtstart=disclosure_start_date, bymonthday=-1)
# >>>rr.between(datetime(2013, 1, 1), datetime(2013, 5, 1))
# [datetime.datetime(2013, 1, 31, 0, 0),
#  datetime.datetime(2013, 2, 28, 0, 0),
#  datetime.datetime(2013, 3, 31, 0, 0),
#  datetime.datetime(2013, 4, 30, 0, 0)]

Unfortunately, I don't think there's an RFC-compliant way to generate a simple RRULE that just falls back to the end of the month if-and-only-if it's necessary (e.g. what do you do with January 30th - you need fallback for February, but you don't want to use bymonthday=-2 because that will give you Feb. 27th, etc).

Alternatively, for a simple monthly rule like this, a better option is probably to just use relativedelta, which does fall back to the end of the month:

from dateutil.relativedelta import relativedelta
from datetime import datetime

def disclosure_dates(dtstart, rd, dtend=None):
    ii = 0
    while True:
        cdate = dtstart + ii*rd
        ii += 1

        yield cdate
        if dtend is not None and cdate >= dtend:
            break


dtstart = datetime(2013, 1, 31, 0, 0)
rd = relativedelta(months=1)
rr = disclosure_dates(dtstart, rd, dtend=datetime(2013, 5, 1))

# >>> list(rr)
# [datetime.datetime(2013, 1, 31, 0, 0),
#  datetime.datetime(2013, 2, 28, 0, 0),
#  datetime.datetime(2013, 3, 31, 0, 0),
#  datetime.datetime(2013, 4, 30, 0, 0),
#  datetime.datetime(2013, 5, 31, 0, 0)]

Note that I specifically used cdate = dtstart + ii * rd, you do not want to just keep a "running tally", as that will pin to the shortest month the tally has seen:

dt_base = datetime(2013, 1, 31)
dt = dt_base
for ii in range(5):
    cdt = dt_base + ii*rd
    print('{} | {}'.format(dt, cdt))
    dt += rd

Result:

2013-01-31 00:00:00 | 2013-01-31 00:00:00
2013-02-28 00:00:00 | 2013-02-28 00:00:00
2013-03-28 00:00:00 | 2013-03-31 00:00:00
2013-04-28 00:00:00 | 2013-04-30 00:00:00
2013-05-28 00:00:00 | 2013-05-31 00:00:00

Kelcie

FYI, there is RFC 7529, which (among other things) extends RRULE by a mode that falls back to the last day of a month if a day doesn't exist. The rule FREQ=MONTHLY;RSCALE=GREGORIAN;SKIP=BACKWARD;BYMONTHDAY=30 would do that. In February this would result in the 28th (or 29th in leap years). Unfortunately there is not much support for this yet. – Stria

@Stria Awesome, didn't know about that. I'll read and implement in dateutil. – Kelcie

@Kelcie is there any progress on this issue in dateutil? – Northward

@IvanVirabyan not done yet but some work towards an implementation exists: github.com/dateutil/dateutil/issues/285 and github.com/dateutil/dateutil/pull/522 – Notify

Close to the end of 2022 now and this problem still exists. Current only alternative is going for a relativedelta solution like the above one. – Catharine

Answer 2

"datetime.datetime(2022, 1, 3, 0, 0)"

Swanhilda

As it’s currently written, your answer is unclear. Please edit to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers in the help center. – Kingofarms