plistlib should create non-naïve datetime objects

[wiml]

BPO	9256
Nosy	@ronaldoussoren, @abalkin, @ned-deily, @bitdancer, @hynek
Files	plistdt.py: a simple test case

^{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/ronaldoussoren'
closed_at = None
created_at = <Date 2010-07-14.01:34:02.327>
labels = ['OS-mac', 'type-bug', 'library']
title = 'plistlib should create non-na\xc3\xafve datetime objects'
updated_at = <Date 2014-06-30.00:02:22.179>
user = 'https://bugs.python.org/wiml'

bugs.python.org fields:

activity = <Date 2014-06-30.00:02:22.179>
actor = 'belopolsky'
assignee = 'ronaldoussoren'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)', 'macOS']
creation = <Date 2010-07-14.01:34:02.327>
creator = 'wiml'
dependencies = []
files = ['17993']
hgrepos = []
issue_num = 9256
keywords = []
message_count = 3.0
messages = ['110249', '110250', '162471']
nosy_count = 6.0
nosy_names = ['ronaldoussoren', 'belopolsky', 'ned.deily', 'r.david.murray', 'wiml', 'hynek']
pr_nums = []
priority = 'normal'
resolution = None
stage = 'needs patch'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue9256'
versions = ['Python 3.5']

Linked PRs

• gh-53502: add a new option aware_datetime in plistlib to loads or dumps aware datetime. #113363
• gh-53502: Fixes for tests in gh-113363 #113627
• gh-53502: Fix plistlib.dump() for naive datetime with aware_datetime option #113645

[wiml]

The plistlib module parses <date> elements in the plist into datetime objects. (This is good.) However, it produces naïve datetimes, even though the <date> elements include an explicit timezone specification ('Z' for UTC, since they're in ISO 8601 format).

Now that bpo-5094 has been fixed, this should be a pretty trivial thing to fix— just pass tzinfo=UTC when creating the datetime (since the regex it uses won't even accept a date that doesn't have the 'Z' suffix, this will be correct).

As an aside: RFC 3339 section 4.3 suggests a convention for distinguishing between "timestamps in UTC" and "timestamps which happen to be defined in terms of UTC, but don't really live in any particular timezone". The python datetime module is not currently able to make this distinction AFAIK, but even though the plist date strings don't follow the RFC 3339 convention, they are most accurately described as having an unknown local offset (like POSIX timestamps, they are UTC-referenced but cannot carry local-timezone info).

[abalkin]

Yes, parsing <date> elements ending with 'Z' into aware datetime objects makes perfect sense, but this should be done carefully because doing so will break any code that mixes the result with naive datetimes.

There is a lengthy RFC 3339 discussion in bpo-7584.

[bitdancer]

An alternative is to do what the email package in 3.3 does, and treat naive datetimes as exactly "UTC referenced but with no information as to what local timezone they originated in". Either way, a program using the plistlib is going to have to know about this and handle the distinction itself, since as you say datetime doesn't have a way to make this differentiation (other than naive vs aware). So it needs to be documented clearly if it isn't already.

[ronaldoussoren]

I've added a feature label because changing this is a feature change and likely can only be done by adding a new option to plist parsers.

[ronaldoussoren]

O my, the PR causes builtbot failures (looks like 32 bit time_t)

[hugovk]

Triage: can this be re-closed following the recent fix PRs?