How to display locale sensitive time format without seconds in python

Question

I can output a locale sensitive time format using strftime('%X'), but this always includes seconds. How might I display this time format without seconds?

>>> import locale
>>> import datetime
>>> locale.setlocale(locale.LC_ALL, 'en_IE.utf-8')
'en_IE.utf-8'
>>> print datetime.datetime.now().strftime('%X')
12:22:43
>>> locale.setlocale(locale.LC_ALL, 'zh_TW.utf-8')
'zh_TW.utf-8'
>>> print datetime.datetime.now().strftime('%X')
12時22分58秒

The only way I can think of doing this is attempting to parse the output of locale.nl_langinfo(locale.T_FMT) and strip out the seconds bit, but that brings it's own trickery.

>>> print locale.nl_langinfo(locale.T_FMT)
%H時%M分%S秒
>>> locale.setlocale(locale.LC_ALL, 'en_IE.utf-8')
'en_IE.utf-8'
>>> print locale.nl_langinfo(locale.T_FMT)
%T

---

Solution:

(Based on pixelbeat's answer.)

# -*- coding: utf-8 -*-
import locale
def locale_time(t, show_seconds=False):
    if show_seconds:
        return t.strftime('%X')
    replacement_fmts = [
        (u'.%S', u''),
        (u':%S', u''),
        (u',%S', u''),
        (u':%OS', ''),
        (u'ཀསར་ཆ%S', u''),
        (u' %S초', u''),
        (u'%S秒', u''),
        (u'%r', '%I:%M %p'),
        (u'%t', '%H:%M'),
        (u'%T', '%H:%M')
    ]
    enc=locale.getpreferredencoding(do_setlocale=False)
    t_fmt = locale.nl_langinfo(locale.T_FMT).decode(enc)
    for fmt in replacement_fmts:
        new_t_fmt = t_fmt.replace(*fmt)
        if new_t_fmt != t_fmt:
            return t.strftime(new_t_fmt.encode(enc))
    return t.strftime(t_fmt.encode(enc)

Usage:

>>> locale.setlocale(locale.LC_ALL, 'en_IE.utf-8')
'en_IE.utf-8'
>>> print locale_time(t)
15:47
>>> locale.setlocale(locale.LC_ALL, 'zh_TW.utf-8')
'zh_TW.utf-8'
>>> print locale_time(t)
15時47分

Answer 1

I would suggest special casing the returned T_FMT as there aren't that many to consider really:

$ for l in $(locale -a | grep utf8); do locale | cut -d= -f1 | LANG=$l xargs locale -kc | grep ^t_fmt=; done | sort -u

t_fmt="%H:%M:%S"
t_fmt="%H.%M.%S"
t_fmt="%H시 %M분 %S초"
t_fmt="ཆུ་ཚོད%Hཀསར་མ%Mཀསར་ཆ%S"
t_fmt="%H时%M分%S秒"
t_fmt="%H时%M分%S秒 %Z"
t_fmt="%H時%M分%S秒"
t_fmt="%I.%M.%S %p"
t_fmt="%I:%M:%S  %Z"
t_fmt="%I:%M:%S %Z"
t_fmt="%I.%M.%S. %Z"
t_fmt="%I時%M分%S秒 %Z"
t_fmt="kl. %H.%M %z"
t_fmt="%k,%M,%S"
t_fmt="%k:%M:%S"
t_fmt="%l:%M:%S"
t_fmt="%OH:%OM:%OS"
t_fmt="%OI:%OM:%OS %p"
t_fmt="%p%I.%M.%S %Z"
t_fmt="%r"
t_fmt="%t"
t_fmt="%T"
t_fmt="%Z %I:%M:%S "

Answer 2

This is bad solution. What happens with some new different locale?

Use following:

t.strftime(gettext('%H:%M'))

Now each translator for each languages would provide suitable format for each string, for en_US it would be '%I:M %p', for zh_TW: %H時%M分

This is how usually problems of missing resource in standard localization tools are solved.