[MERGE] Add get_username() call to the UIFactory.

[Vincent Ladeuil]

>>>>> "Jelmer" == Jelmer Vernooij <jelmer at samba.org> writes:

    Jelmer> Vincent Ladeuil wrote:
    >>>>>>> "Jelmer" == Jelmer Vernooij <jelmer at samba.org> writes:> 
    Jelmer> TestUIFactory provides a custom implementation of
    Jelmer> get_non_echoed_password(); making it provide a mock
    Jelmer> implementation of get_username() that doesn't raise
    Jelmer> NotATerminal seems kind of pointless as we would only
    Jelmer> be testing the mock implementation, not the actual
    Jelmer> one.
    >> 
    >> That's a different issue.
    >> 
    >> If you mean NotATerminal is tested nowhere, you've found yet
    >> another hole in our test suite, feel free to provide more tests
    >> but I don't require them at that point (a FIXME: will do :)

So I sent a patch regarding NotATerminal, you may want to merge
from it once it lands.

    Jelmer> Attached is an updated version that no longer uses
    Jelmer> NotATerminal and tests utf8, as we discussed on IRC.

Could you refactor a bit instead of blindly copy/pasting there ?

<snip/>

    Jelmer> +
    Jelmer> +    def test_text_ui_getusername_utf8(self):
    Jelmer> +        ui = TestUIFactory(stdin=u'someuser\u1234'.encode('utf8'),
    Jelmer> +                           stdout=StringIOWrapper())
    Jelmer> +        ui.stdin.encoding = "utf8"
    Jelmer> +        ui.stdout.encoding = ui.stdin.encoding
    Jelmer> +        pb = ui.nested_progress_bar()
    Jelmer> +        try:
    Jelmer> +            # there is no output from the base factory
    Jelmer> +            password = ui.get_username(u'Hello\u1234 %(host)s',
    Jelmer> +                host=u'some\u1234')
    Jelmer> +            self.assertEquals(u"someuser\u1234", password.decode('utf8'))

If you don't refactor, at least rename the variables:
s/password/username/...

I also think encoding username/password to utf-8 may need to be
discussed as AFAICS we aren't very clear about what is really
needed here and what we really do in the code base.

I won't block on that as using utf8 is compatible with ascii and
until we encounter the problem (as in someone file a bug) we may
as well continue as we do now: leave upper levels handle the
encoding.

   Vincent