def test_python_crash(self):
import random
from io import BytesIO
random.seed(42)
buf = BytesIO()
for i in range(150000):
buf.write(bytes([random.randint(0, 255)]))
to_unicode(buf.getvalue(), 'utf-16-le')
to_unicode(buf.getvalue(), 'utf-16-be')
to_unicode(buf.getvalue(), 'utf-32-le')
to_unicode(buf.getvalue(), 'utf-32-be')
def test_python_crash(self):
import random
from io import BytesIO
random.seed(42)
buf = BytesIO()
for i in range(150000):
buf.write(bytes([random.randint(0, 255)]))
to_unicode(buf.getvalue(), 'utf-16-le')
to_unicode(buf.getvalue(), 'utf-16-be')
to_unicode(buf.getvalue(), 'utf-32-le')
to_unicode(buf.getvalue(), 'utf-32-be')
def test_invalid_utf8(self):
self.assertEqual(to_unicode(b'\xc2\xc2\xa3', 'utf-8'), u'\ufffd\xa3')
def test_utf8(self):
self.assertEqual(to_unicode(b'\xc2\xa3', 'utf-8'), u'\xa3')
def bytes_to_unicode(bytes, encoding):
"""
将给定的bytes，通过encoding转换成unicode
"""
return to_unicode(bytes, encoding)
def body_as_unicode(self):
"""Return body as unicode"""
if self._cached_ubody is None:
self._cached_ubody = to_unicode(self.body, self.encoding)
return self._cached_ubody
def test_invalid_utf8(self):
self.assertEqual(to_unicode('\xc2\xc2\xa3', 'utf-8'), u'\ufffd\xa3')
def test_utf8(self):
self.assertEqual(to_unicode('\xc2\xa3', 'utf-8'), u'\xa3')
def body_as_unicode(self):
"""Return body as unicode"""
if self._cached_ubody is None:
self._cached_ubody = to_unicode(self.body, self.encoding)
return self._cached_ubody
def test_invalid_utf8(self):
self.assertEqual(to_unicode(b"\xc2\xc2\xa3", "utf-8"), "\ufffd\xa3")
def test_utf8(self):
self.assertEqual(to_unicode(b"\xc2\xa3", "utf-8"), "\xa3")
