Automated testing for locale

[furkansimsekli]

To be sure that every translation is available in every languages.

• Create a keyword list (a script to extract translation keys from source code)
• Check whether every department name has been translated

[div72]

What you essentially want is GNU gettext but you'll need a custom translation function that automatically deducts the language since you're not using a single language for the entire application.

Something like this (pseudocode):

_LANGUAGES = {l: gettext.translation("hu-announcement-bot", languages=[l]) for l in ["en", "tr", "fr"]}
def _(message: str):
    return _LANGUAGES[inspect.getcurrentframe().f_back.f_locals["language"]].gettext(message)

...

def subscribe(...):
    language = "en"
    print(_("Thanks for subscribing to %(department)s") % {"department": _(department)})

From there you can use xgettext like xgettext -o - test.py to extract the translation strings from the source.

[furkansimsekli]

@div72
Using gettext means much more work at the beginning, the reason why I chose the way I handle translations now was because I wanted to keep things really really simple.

If gettext comes up with a perfect solution for #41, then I might think of using it honestly.

For this issue, writing a script to replicate what xgettext -o test.py should be rather easy (parse keys from decode() function calls (assuming every call is explicit)) and maintain a default language (either english or turkish) to fulfill at least one language and make it a guide for other translations.

[div72]

If gettext comes up with a perfect solution for #41, then I might think of using it honestly.

There's no perfect solution as that issue is kinda irrelevant. You can use % (%s/%(keyword)s style) or str.format({} {0} {keyword} style) with both the current or the gettext backend.