Encoding errors in non-ASCII locales

Created by: gflohr

Current git checkout with python 2.7.5 on gentoo, running a Gnome session in the locale de_DE.UTF-8:

$ /usr/bin/hamster list

Traceback (most recent call last): File "/usr/bin/hamster", line 391, in getattr(hamster_client, command)(_args) File "/usr/bin/hamster", line 254, in list self._list(start_time, end_time) File "/usr/bin/hamster", line 313, in list print fact_line.format(*headers) UnicodeEncodeError: 'ascii' codec can't encode character u'\xe4' in position 1: ordinal not in range(128) $

The string "Activity" from in /usr/bin/hamster line 313 translates to "Tätigkeit" in German, and 0xe4 is the code point for the "ä". The issue vanishes by calling gettext.install() in src/hamster/lib/i18n.py with the "unicode" argument set to False.

Now without the "list" command:

$ /usr/bin/hamster ... File "/usr/lib64/python2.7/site-packages/hamster/lib/stuff.py", line 91, in format_range title = (u"%(start_B)s %(start_d)s – %(end_B)s %(end_d)s, %(end_Y)s") % dates_dict UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 1: ordinal not in range(128)

After changing the u"" strings into byte strings without prefix, I get this:

...
self.label.set_markup('<b>%s</b>' % stuff.format_range(start_date, end_date).encode("utf-8"))

UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 1: ordinal not in range(128)

Note that 0xc3 is the code point "Ã". So this is obviously a double encoding error for utf-8. The problem vanishes after removing the encode('utf-8') from set_range() in src/hamster/widgets/dates.py:

def set_range(self, start_date, end_date=None):
    end_date = end_date or start_date
    self.start_date, self.end_date = start_date, end_date
    #self.label.set_markup('<b>%s</b>' % stuff.format_range(start_date, end_date).encode("utf-8"))
    self.label.set_markup('<b>%s</b>' % stuff.format_range(start_date, end_date))

After those modifications the window shows up and the application seems to work.

I can send you a real patch if you want but to me it looks like a crude workaround for a more general problem.