_get_unix_timezone incorrect timezone parsing on Amazon Linux AMI release 2018.03

[Dmitrii-I]

The following happens on Amazon Linux AMI release 2018.03 with pendulum 2.0.0 (but also with 1.5.1):

Python 3.4.8 (default, Apr 25 2018, 23:50:36)
[GCC 4.8.5 20150623 (Red Hat 4.8.5-11)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from pendulum.tz.local_timezone import _get_unix_timezone
>>> _get_unix_timezone()
Traceback (most recent call last):
  File "/home/dmitrii/.local/lib/python3.4/site-packages/pendulum/tz/zoneinfo/reader.py", line 49, in read_for
    file_path = pytzdata.tz_path(timezone)
  File "/usr/local/lib/python3.4/site-packages/pytzdata/__init__.py", line 74, in tz_path
    raise TimezoneNotFound('Timezone {} not found at {}'.format(name, filepath))
pytzdata.exceptions.TimezoneNotFound: Timezone /usr/share/zoneinfo/Europe/Zurich not found at /usr/local/lib/python3.4/site-packages/pytzdata/zoneinfo/usr/share/zoneinfo/Europe/Zurich

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/dmitrii/.local/lib/python3.4/site-packages/pendulum/tz/local_timezone.py", line 206, in _get_unix_timezone
    return Timezone(etctz.replace(' ', '_'))
  File "/home/dmitrii/.local/lib/python3.4/site-packages/pendulum/tz/timezone.py", line 30, in __init__
    tz = read(name, extend=extended)
  File "/home/dmitrii/.local/lib/python3.4/site-packages/pendulum/tz/zoneinfo/__init__.py", line 9, in read
    return Reader(extend=extend).read_for(name)
  File "/home/dmitrii/.local/lib/python3.4/site-packages/pendulum/tz/zoneinfo/reader.py", line 51, in read_for
    raise InvalidTimezone(timezone)
pendulum.tz.zoneinfo.exceptions.InvalidTimezone: Invalid timezone "/usr/share/zoneinfo/Europe/Zurich"

The issue seems to be that _get_unix_timezone is unable to parse the string 'ZONE="/usr/share/zoneinfo/Europe/Zurich"\n which is returned when the zonefile is read.

I think the issue is here that you expect there Europe/Zurich.

[Dmitrii-I]

A workaround is to add etctz = etctz.replace('/usr/share/zoneinfo/', '') before return Timezone(etctz.replace(' ', '_')) on line 203 in local_timezone.py

[austinbutler]

Same problem with an Ubuntu 18.04 container on Amazon Linux ECS-optimized. Worked around it for now by setting environment variable TZ to "UTC" in the container.

[sdispater]

It should be fixed in the latest release 2.0.1

[Dmitrii-I]

thanks

[rkissos]

There is still an issue in 2.0.1:

  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/__init__.py", line 59, in local_timezone
    return get_local_timezone()
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/local_timezone.py", line 30, in get_local_timezone
    tz = _get_system_timezone()
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/local_timezone.py", line 58, in _get_system_timezone
    return _get_unix_timezone()
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/local_timezone.py", line 222, in _get_unix_timezone
    return Timezone(tzpath)
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/timezone.py", line 30, in __init__
    tz = read(name, extend=extended)
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/zoneinfo/__init__.py", line 9, in read
    return Reader(extend=extend).read_for(name)
  File "/opt/conda/envs/dev/lib/python3.6/site-packages/pendulum/tz/zoneinfo/reader.py", line 50, in read_for
    raise InvalidTimezone(timezone)
pendulum.tz.zoneinfo.exceptions.InvalidTimezone: Invalid timezone "zoneinfo/UTC"

[Dmitrii-I]

@rkissos

OK I am re-opening this issue. Could you help out and specify what Linux distribution you are using (e.g. cat /etc/issue).

[rkissos]

centos 7.4.1708 (Core)

[lithammer]

Same here (CentOS 7.4.1708). The issue is this part:

https://github.com/sdispater/pendulum/blob/10a76643c8f100a81368f3dba05dc74404ff9cdd/pendulum/tz/local_timezone.py#L219-L226

In the first iteration the variable tzpath has the value 'zoneinfo/UTC'. This is wrong, and causes an error. Unfortunately you get a InvalidTimezone exception instead of one of the handled exceptions (i.e. ValueError, IOError, and OSError).

So you never get to parts.pop(0), which would've removed the 'zoneinfo' part from parts, and the next iteration would've been successful.

Something like this could be a potential fix, or look over what gets raised from the Timezone(tzpath) call:

         while parts:
             tzpath = '/'.join(parts)
             try:
                 return Timezone(tzpath)
-            except (ValueError, IOError, OSError):
+            except (ValueError, IOError, OSError, InvalidTimezone):
                 pass

             parts.pop(0)

[sdispater]

This should be fixed in the latest release: 2.0.2

[lithammer]

@sdispater are you thinking of 6ee68b9? Because I think you need to do a similar change to around pendulum/tz/local_timezone.py#L224. At least that's where my bug/error happens.

[lithammer]

Some pdb debug output (v2.0.2).

-> while parts:
(Pdb) l
216  	    tzpath = os.path.join(_root, 'etc/localtime')
217  	    if os.path.exists(tzpath) and os.path.islink(tzpath):
218  	        tzpath = os.path.realpath(tzpath)
219  	        parts = tzpath.split('/')[-2:]
220  	        import pdb; pdb.set_trace()
221  ->	        while parts:
222  	            tzpath = '/'.join(parts)
223  	            try:
224  	                return Timezone(tzpath)
225  	            except (ValueError, IOError, OSError):
226  	                pass
(Pdb) parts
['zoneinfo', 'UTC']
(Pdb) tzpath
'/usr/share/zoneinfo/UTC'
(Pdb) Timezone(tzpath)
*** InvalidTimezone: Invalid timezone "/usr/share/zoneinfo/UTC"

[sdispater]

@Renstrom Yes, sorry about that. I didn't apply the fix to this case.

I will change it.