def test_bom(self):
# cjk water character in unicode
water_unicode = u'\u6C34'
# BOM + water character encoded
utf16be = '\xfe\xff\x6c\x34'
utf16le = '\xff\xfe\x34\x6c'
utf32be = '\x00\x00\xfe\xff\x00\x00\x6c\x34'
utf32le = '\xff\xfe\x00\x00\x34\x6c\x00\x00'
for string in (utf16be, utf16le, utf32be, utf32le):
bom_encoding, bom = read_bom(string)
decoded = string[len(bom):].decode(bom_encoding)
self.assertEqual(water_unicode, decoded)
# Body without BOM
enc, bom = read_bom("foo")
self.assertEqual(enc, None)
self.assertEqual(bom, None)
# Empty body
enc, bom = read_bom("")
self.assertEqual(enc, None)
self.assertEqual(bom, None)
def test_bom(self):
# cjk water character in unicode
water_unicode = u'\u6C34'
# BOM + water character encoded
utf16be = b'\xfe\xff\x6c\x34'
utf16le = b'\xff\xfe\x34\x6c'
utf32be = b'\x00\x00\xfe\xff\x00\x00\x6c\x34'
utf32le = b'\xff\xfe\x00\x00\x34\x6c\x00\x00'
for string in (utf16be, utf16le, utf32be, utf32le):
bom_encoding, bom = read_bom(string)
decoded = string[len(bom):].decode(bom_encoding)
self.assertEqual(water_unicode, decoded)
# Body without BOM
enc, bom = read_bom("foo")
self.assertEqual(enc, None)
self.assertEqual(bom, None)
# Empty body
enc, bom = read_bom("")
self.assertEqual(enc, None)
self.assertEqual(bom, None)
def test_bom(self):
# cjk water character in unicode
water_unicode = "\u6C34"
# BOM + water character encoded
utf16be = b"\xfe\xff\x6c\x34"
utf16le = b"\xff\xfe\x34\x6c"
utf32be = b"\x00\x00\xfe\xff\x00\x00\x6c\x34"
utf32le = b"\xff\xfe\x00\x00\x34\x6c\x00\x00"
for string in (utf16be, utf16le, utf32be, utf32le):
bom_encoding, bom = read_bom(string)
assert bom_encoding is not None
assert bom is not None
decoded = string[len(bom):].decode(bom_encoding)
self.assertEqual(water_unicode, decoded)
# Body without BOM
enc, bom = read_bom(b"foo")
self.assertEqual(enc, None)
self.assertEqual(bom, None)
# Empty body
enc, bom = read_bom(b"")
self.assertEqual(enc, None)
self.assertEqual(bom, None)