[issue15873] datetime: add ability to parse RFC 3339 dates and times

[Anders Hovmöller]

Anders Hovmöller added the comment:

The tests attached to this ticket seem pretty bare. Issues that I can spot directly:

- only tests for datetimes, not times or dates
- only tests for zulu and "-8:00” timezones
- no tests for invalid input (parsing a valid date as a datetime for example)
- only tests for YYYY-MM-DDTHH:MM:SSZ, but ISO8601 supports:
    - Naive times
    - Timezone information (specified as offsets or as Z for 0 offset)
    - Year
    - Year-month
    - Year-month-date
    - Year-week
    - Year-week-weekday
    - Year-ordinal day
    - Hour
    - Hour-minute
    - Hour-minute
    - Hour-minute-second
    - Hour-minute-second-microsecond
    - All combinations of the three "families" above!
(the above list is a copy paste from my project that implements all ISO8601 that fits into native python: https://github.com/boxed/iso8601 <https://github.com/boxed/iso8601>)

This is a more reasonable test suite: https://github.com/boxed/iso8601/blob/master/iso8601.py#L166 <https://github.com/boxed/iso8601/blob/master/iso8601.py#L166> although it lacks the tests for bogus inputs.

> On 2016-07-16, at 03:41, Alexander Belopolsky <report at bugs.python.org> wrote:
> 
> 
> Alexander Belopolsky added the comment:
> 
> I would very much like to see this ready before the feature cut-off for Python 3.6.  Could someone post a summary on python-ideas to get a show of hands on some of the remaining wrinkles?
> 
> I would not worry about a C implementation at this point.  We can put python implementation in _strptime.py and call it from C as we do for the strptime method.
> 
> ----------
> 
> _______________________________________
> Python tracker <report at bugs.python.org>
> <http://bugs.python.org/issue15873>
> _______________________________________

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue15873>
_______________________________________