accessing tzinfo for timezones defined via tzwin which don't observe Daylight Saving Time throws ValueError because _dstmonth is 0

[Dick Swanson]

on windows:

from dateutil.tz import gettz
from datetime import datetime

tzinfo = gettz('UTC')
f = datetime.now(tzinfo)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Python27\lib\site-packages\python_dateutil-1.5-py2.7.egg\dateutil\tzwin.py", line 32, in utcoffset
    if self._isdst(dt):
  File "C:\Python27\lib\site-packages\python_dateutil-1.5-py2.7.egg\dateutil\tzwin.py", line 67, in _isdst
    self._dstweeknumber)
  File "C:\Python27\lib\site-packages\python_dateutil-1.5-py2.7.egg\dateutil\tzwin.py", line 166, in picknthweekday
    first = datetime.datetime(year, month, 1, hour, minute)
ValueError: month must be in 1..12

Patch:
--- tzwin.py~ 2012-06-05 09:08:36.212393800 -0400
+++ tzwin.py 2012-06-07 10:24:33.878270600 -0400
@@ -62,6 +62,8 @@
         return self._display

     def _isdst(self, dt):
+ if self._dstmonth == 0:
+ return False
         dston = picknthweekday(dt.year, self._dstmonth, self._dstdayofweek,
                                self._dsthour, self._dstminute,
                                self._dstweeknumber)

[ccanepa (ccanepacc)]

Update 2014

This defect is still present in python-dateutil 2.2 and in lp:dateutil 135

It will manifest when testing under py 3.X , will not show up in py 2.X due to another bug (will report as separate issue)

The rationale for the fix is solid: as seen in the docs of win32 API for TIME_ZONE_INFORMATION structure [1], dstmonth == 0 is a flag for 'this timezone don't have Daylight Saving Time'

We ( https://github.com/getnikola/nikola ) are affected for this defect.

I will post a merge request, with test, if that helps to land the patch.

[1] http://msdn.microsoft.com/en-us/library/windows/desktop/ms725481(v=vs.85).aspx

[ccanepa (ccanepacc)]

Merge request at
https://code.launchpad.net/~ccanepacc/dateutil/fix-isdst-in-tzwin/+merge/224424