Python dateutil.parser throws “ValueError: day is out of range for month”

Question

I have a the following code that runs fine with input format like {Year}/{Month} except when it comes to 1994/02

Here is the sample code

>>> import dateutil.parser as dtp
>>> dtp.parse('1994/01')
datetime.datetime(1994, 1, 29, 0, 0)
>>> dtp.parse('1994/03')
datetime.datetime(1994, 3, 29, 0, 0)
>>> dtp.parse('1994/02')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/antony/.virtualenvs/comp-invest/lib/python2.7/site-packages/dateutil/parser.py", line 720, in parse
    return DEFAULTPARSER.parse(timestr, **kwargs)
  File "/Users/antony/.virtualenvs/comp-invest/lib/python2.7/site-packages/dateutil/parser.py", line 317, in parse
    ret = default.replace(**repl)
ValueError: day is out of range for month

As you can see, the code works fine with 1994/01 and 1994/03, but fails with 1994/02 Does this have anything to do with leap year? But more important, how do I get around this problem and make my code work again?

Thanks

Answer 1

dtp.parse is filling in the missing day with the current date's day. You ran the code on 2013/01/29 and day 29 does not exist in February (i.e. 1994/02/29).

Use this instead:

dtp.parse('1994/01'+'/01')

It will give consistent results (first day of month) regardless of when the code is executed.

Answer 2

This was a bug in dateutil that has since been fixed. Version 2.5.0 and higher will no longer have this issue.

If you must use an earlier version, I think that the "correct" way to handle things is to specify the default parameter:

from dateutil.parser import parse
from datetime import datetime, date

# First of the current month, at midnight.
default_date = datetime.combine(date.today(), datetime.min.time()).replace(day=1)
dt = parse('1994/01', default=default_date)

This will default to the 1st of the month rather than the current day.