translation.get_language() should remove sublanguage code if it is not in settings.LANGUAGES

[mrts]

When translation.get_language_from_request() processes the Accept-Language HTTP header it makes sure to remove the sublanguage code if it is not in settings.LANGUAGES, so that translation.get_language() subsequently returns only the "main" language code.

However, the behaviour is different in shell:

$ grep -2 LANGUAGES settings.py

ugettext = lambda s: s
LANGUAGES = (
    ('et', ugettext('Estonian')),
    ('en', ugettext('English')),

$ grep LANGUAGE_CODE settings.py
LANGUAGE_CODE = 'et'

$ locale
LANG=en_US.UTF-8

$ ./manage.py shell --plain
Python 2.6.2 (release26-maint, Apr 19 2009, 01:58:18) 
[GCC 4.3.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> from django.utils.translation import get_language
>>> get_language()
'en-us'

I.e. the shell locale is used unchanged.

That breaks code that assumes translation.get_language_from_request() behaviour, e.g. auto-translated model fields (properties that return getattr(self, fieldname + langcode) where langcode comes from get_language()).

This is a minor inconsistency easily fixable with translation.activate() in the shell, so feel free to wontfix this. However, if #9340 gets fixed, perhaps this could be something that can be looked into while at it.

[Malcolm Tredinnick]

Dupe of #9340.