from_format('12/2/2015 12:00:00 AM EST', 'M/D/YYYY h:mm:ss A zz')

>>> pendulum.from_format('12/2/2015 12:00:00 AM EST', 'M/D/YYYY h:mm:ss A zz')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/nackjicholson/Library/Caches/pypoetry/virtualenvs/source-data-service-py3.6/lib/python3.6/site-packages/pendulum/__init__.py", line 256, in from_format
    parts = _formatter.parse(string, fmt, now(), locale=locale)
  File "/Users/nackjicholson/Library/Caches/pypoetry/virtualenvs/source-data-service-py3.6/lib/python3.6/site-packages/pendulum/formatting/formatter.py", line 400, in parse
    lambda m: self._replace_tokens(m.group(0), locale), escaped_fmt
  File "/Users/nackjicholson/Library/Caches/pypoetry/virtualenvs/source-data-service-py3.6/lib/python3.6/site-packages/pendulum/formatting/formatter.py", line 400, in <lambda>
    lambda m: self._replace_tokens(m.group(0), locale), escaped_fmt
  File "/Users/nackjicholson/Library/Caches/pypoetry/virtualenvs/source-data-service-py3.6/lib/python3.6/site-packages/pendulum/formatting/formatter.py", line 651, in _replace_tokens
    raise ValueError("Unsupported token: {}".format(token))
ValueError: Unsupported token: zz

This is from the master branch. The zz is in the documentation, I feel like it should work.

The from_format functionality is very important to my usage of pendulum. I’d like to find some time to dig in and see if I can help make it better but I could use some guidance. Do you have any idea why it has been unreliable? Is there some other approach that you might want to take in order to improve it?

[altendky]

Some abbreviations are present in the regular pendulum.timezones list (in other words, z).

>>> tuple(tz for tz in pendulum.timezones if '/' not in tz and tz[:2].isupper())
('CET', 'CST6CDT', 'EET', 'EST', 'EST5EDT', 'GB', 'GB-Eire', 'GMT', 'GMT+0', 'GMT-0', 'GMT0', 'HST', 'MET', 'MST', 'MST7MDT', 'NZ', 'NZ-CHAT', 'PRC', 'PST8PDT', 'ROC', 'ROK', 'UCT', 'UTC', 'W-SU', 'WET')

But that list seems lacking enough for the US, let alone with a less US-centric view, to be not particularly useful.

I’m not sure there’s a general solution to this as I don’t believe the abbreviations are well defined enough to be unique. Still, they certainly are used and the ambiguity could be overcome by allowing the caller to pass a mapping of unique abbreviations to unique names. {'CDT': 'America/Chicago'} (hopefully that’s a good example but…) Optionally, regional default lists could be provided for easy use.

[altendky]

I agree that it is accurate for this issue to be in the open state. I don’t have authority to make the fix happen and I have lots of other open source work I am more responsible for that needs my time. Sorry. :[

(nitpick, this issue is three years old, not two)