Un-unicode output for Windows CMD #120
Conversation
| @@ -23,3 +23,6 @@ def test_colorOutput(self): | |||
| self.assertTrue(issubclass(type(gs.stream), | |||
| colorama.ansitowin32.StreamWrapper)) | |||
|
|
|||
| def test_unicode_override(self): | |||
| pass | |||
There was a problem hiding this comment.
Looking pretty good, except for this poor test here. ;-)
There was a problem hiding this comment.
It is a poor, sad test :(
I couldn't figure out how to get green to run a unittest testcase, and then validate what gets printed out. The actual testcase is nearly trivial...
class UnicodeTest(unittest.TestCase):
def test1(self):
print('\u03BB') # lambda
self.assertTrue(True)
Suggestions on how to go about this?
There was a problem hiding this comment.
The true test is to spawn subprocess of green running that test method, and check the output.
https://github.com/wikimedia/pywikibot-core/blob/master/tests/pwb_tests.py#L71
There all sorts of gotchas, especially when doing this on Windows, and only Python versions, resulting a mess like
https://github.com/wikimedia/pywikibot-core/blob/master/tests/utils.py#L771
However I suspect you'll only need to set PYTHONIOENCODING to utf-8 for the subprocess
|
All of these "on_appveyor" conditions look like handling only the simple case of Appveyor, and dont really test using green on Windows because there is special logic for Appveyor only. e.g. I expect that This is just a general comment, as I realise that this changeset isnt introducing "on_appveyor", but it is making it more "normal" by exposing it as an attribute. |
| # Ironically, AppVeyor doesn't support windows win32 system calls for | ||
| # colors, but it WILL interpret posix ansi escape codes! | ||
| on_windows = platform.system() == 'Windows' | ||
| on_appveyor = os.environ.get('APPVEYOR', False) | ||
| self.on_windows = platform.system() == 'Windows' |
There was a problem hiding this comment.
this shouldnt be an instance variable. it should be a package level constant.
Created a unit test for MinchinWeb's pull request
|
@jayvdb I've reworked the 'flags' inside the module a little to try and make things more explicit. The test failed on Travis because The test failed on Appvoyor on a test where colorama is supposed to strip ANSI codes. But the issue is something gets fed a In any case, the tests do pass on my local machine. |
| @@ -167,6 +171,9 @@ def writeln(self, text=''): | |||
| def write(self, text): | |||
| if type(text) == bytes: | |||
| text = text.decode('utf-8') | |||
| # Compensate for windows' anti-social unicode behavior | |||
| if self._ascii_only_output: | |||
| text = unidecode(text) | |||
There was a problem hiding this comment.
Sorry for the delay. I have been just absolutely swamped.
I suggest just changing line 176 to read:
text = unidecode(unicode(text))
That may be work and we could be done with this and merge it.
There was a problem hiding this comment.
Hmm, wait. Python 3 doesn't have the unicode() function anymore.
Since it's a Python 2-specific fix, I suppose we should make that condition:
if self._ascii_only_output:
if platform.python_version_tuple()[0] == '2': # pragma: no cover
text = unicode(text)
text = unidecode(text)
Explicitely convert to Unicode for Python 2
|
@CleanCut: I've made the changes you suggested |
|
Included in the 2.4.2 release. Which was just released. |
Fixes #119