def _check_utf8_encoding(self, bstring):
"""
Check whether the given sequence of bytes
is properly encoded in UTF-8.
:param bytes bstring: the byte string to be checked
"""
if not gf.is_bytes(bstring):
self._failed(u"The given string is not a sequence of bytes")
return
if not gf.is_utf8_encoded(bstring):
self._failed(u"The given string is not encoded in UTF-8.")
def test_is_utf8_encoded(self):
tests = [
(u"foo".encode("ascii"), True),
(u"foo".encode("latin-1"), True),
(u"foo".encode("utf-8"), True),
(u"foo".encode("utf-16"), False),
(u"foo".encode("utf-32"), False),
(u"foà".encode("latin-1"), False),
(u"foà".encode("utf-8"), True),
(u"foà".encode("utf-16"), False),
(u"foà".encode("utf-32"), False),
]
for test in tests:
self.assertEqual(gf.is_utf8_encoded(test[0]), test[1])
def test_is_utf8_encoded(self):
tests = [
(u"foo".encode("ascii"), True),
(u"foo".encode("latin-1"), True),
(u"foo".encode("utf-8"), True),
(u"foo".encode("utf-16"), False),
(u"foo".encode("utf-32"), False),
(u"foà".encode("latin-1"), False),
(u"foà".encode("utf-8"), True),
(u"foà".encode("utf-16"), False),
(u"foà".encode("utf-32"), False),
]
for test in tests:
self.assertEqual(gf.is_utf8_encoded(test[0]), test[1])
