HTTPResponse.content is not unicode in context of test cases.

[Will Pimblett]

Came across the following when adding from __future__ import unicode_literals to my test.py-s

When doing the following for an existing view with an expected output:

def test_page_not_found(self):
        url = '/help/404/'
        c = self.get_client()
        response = c.get(url)
        self.assertIn("Page Not Found", response.content)

We get:

======================================================================
ERROR: test_page_not_found (xsd_help.tests.HelpViewTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/builds/wjdp/xsacdb/src/xsd_help/tests.py", line 23, in test_page_not_found
    self.assertTrue("Page Not Found" in response.content)
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 21370: ordinal not in range(128)

Workaround is to replace response.content with unicode(response.content, encoding=settings.DEFAULT_CHARSET). According to what I could find in the docs the content should be encoded with uft-8, I think this issue is because request.content is a bytestring without an encoding attached.

I'm raising this either in case this in not expected or if it is suggesting a more descriptive error is generated and/or better documentation for those who come across this - there was a *lot* of headscratching needed to find the root of these test failures.

Another suggestion: could we have another property of HTTPResponse that gives us a unicode string rather than a bytestring for test cases?

See ​http://stackoverflow.com/questions/36004324/function-assertin-causes-the-unicodedecodeerror/37456287 for another similar use case.

[Simon Charette]

The SimpleTestCase class already provides the ​assertContains method for this use case.

I'm afraid there's not much we can do to make the error message more descriptive as that's standard Python stuff.

[Simon Charette]

I also noticed that the ​Porting to Python 3 guide mentions why assertContains should be used instead.

Is this the reason you added the from __future__ import unicode_literals import in the first place?

[Will Pimblett]

Ah, must have glossed over that, assertContains seems to fix all of this. unicode_literals was unrelated I think, just fixing up use of literals.

[Will Pimblett]

Feel free to mark as invalid.