parser doesn't seem to properly parse exif formatted datetime

[jMyles]

Hey there. Thanks for pendulum.

It seems that pendulum.parse() does the wrong thing for strings gleaned by PIL's ExifTags interface.

For example, on a Canon S-series camera, the string for the key 'DateTime" is:
'2016:12:26 15:45:28'

By pendulum parses this into this Pendulum object:
2017-03-16T11:21:27+00:00

[sdispater]

Well, the thing is the string is not really standard (the : separator is mostly reserved for time information) so pendulum will fallback on the dateutil parser which produces the error:

>>> from dateutil.parser import parse

>>> parse('2016:12:26 15:45:28')
datetime.datetime(2017, 3, 16, 15, 45, 28)

To make it work you have to use from_format():

>>> import pendulum

>>> pendulum.from_format('2016:12:26 15:45:28', '%Y:%m:%d %H:%M:%S')
# <Pendulum [2016-12-26T15:45:28+00:00]>

[jMyles]

Roger. I agree it's an odd format, but to the degree that parse() is generally applicable for datetime representations to be found in the wild: do you think that Exif data from Canon cameras is common enough to warrant special logic?

If so, I can work on a pull request.

[sethmlarson]

I'm +1 on this having dealt with EXIF data in the past.

[emattiza]

+1 on the issue as well. implemented a unit test for the single case presented above, and a slight modification to the COMMON regex used in the parser for YYYY:MM:DD separators. from_format does look to be the appropriate parse option though fwiw. PR this?

[sdispater]

Version 1.2.0 has just been released with this feature included.