strptime() can produce invalid date with negative year day

[serhiy-storchaka]

BPO	23718
Nosy	@malemburg, @abalkin, @serhiy-storchaka
Files	julian_date.patch julian_date_v1.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/serhiy-storchaka'
closed_at = <Date 2016-03-12.08:55:19.765>
created_at = <Date 2015-03-20.11:30:32.291>
labels = ['type-bug', 'library']
title = 'strptime() can produce invalid date with negative year day'
updated_at = <Date 2016-03-12.08:55:19.764>
user = 'https://github.com/serhiy-storchaka'

bugs.python.org fields:

activity = <Date 2016-03-12.08:55:19.764>
actor = 'serhiy.storchaka'
assignee = 'serhiy.storchaka'
closed = True
closed_date = <Date 2016-03-12.08:55:19.765>
closer = 'serhiy.storchaka'
components = ['Library (Lib)']
creation = <Date 2015-03-20.11:30:32.291>
creator = 'serhiy.storchaka'
dependencies = []
files = ['41877', '41936']
hgrepos = []
issue_num = 23718
keywords = ['patch']
message_count = 10.0
messages = ['238648', '259752', '259956', '260168', '260351', '260353', '260360', '261238', '261642', '261643']
nosy_count = 6.0
nosy_names = ['lemburg', 'belopolsky', 'python-dev', 'serhiy.storchaka', 'beng94', 'iaslan']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue23718'
versions = ['Python 2.7', 'Python 3.5', 'Python 3.6']

[serhiy-storchaka]

strptime() can produce invalid time with negative year day when parse year-week-weekday set. Such time is rejected by strftime(), so strptime/strftime roundtrip doesn't work.

>>> t = time.strptime('2015 0 0', '%Y %U %w')
>>> t
time.struct_time(tm_year=2014, tm_mon=12, tm_mday=28, tm_hour=0, tm_min=0, tm_sec=0, tm_wday=6, tm_yday=-3, tm_isdst=-1)
>>> time.strftime('%Y %U %w', t)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: day of year out of range

[beng94]

It looks interesting, let me try to solve this. At first it seems odd that _calc_julian_from_U_or_W returns -2, I guess something is wrong around there.

[beng94]

I made a patch, that solves this issue.

The problem was that there wasn't any Sunday (%w = 0) on the 0th week in 2015. 2015 started on Thursday, therefore the first Sunday was on 2014.12.28.

julian variable is used to set the tm_yday, which was a minus value, as the given date was not in the same year.

[serhiy-storchaka]

Could you please add tests?

[beng94]

Actually there are already test cases, but they test for the wrong behaviour. The very same example is tested there, but the test gives the expected result, so tm_yday = -3. My implementation returns 362, which looks more reasonable. So currently with my patch the test fails.

See the tests here: https://github.com/python/cpython/blob/master/Lib/test/test_strptime.py#L523

[serhiy-storchaka]

Could you provide a patch? See also my comment on Rietveld (the "review" link).

[beng94]

I've added a new patch, it uses an other way to calculate the number of days in a given year. I updated the tests, so now it doesn't fail, I also added some extra test cases to test leap years.

[serhiy-storchaka]

LGTM. Alexander, what would you say?

[python-dev]

New changeset f03da87a79fa by Serhiy Storchaka in branch '3.5':
Issue bpo-23718: Fixed parsing time in week 0 before Jan 1. Original patch by
https://hg.python.org/cpython/rev/f03da87a79fa

New changeset 4fb167ec3108 by Serhiy Storchaka in branch '2.7':
Issue bpo-23718: Fixed parsing time in week 0 before Jan 1. Original patch by
https://hg.python.org/cpython/rev/4fb167ec3108

New changeset a7093386efaf by Serhiy Storchaka in branch 'default':
Issue bpo-23718: Fixed parsing time in week 0 before Jan 1. Original patch by
https://hg.python.org/cpython/rev/a7093386efaf

[serhiy-storchaka]

Thank you Tamás for your contribution.