_vendor/webencodings/tests.py

16 from . import (lookup, LABELS, decode, encode, iter_decode, iter_encode,
30     assert lookup('utf-8').name == 'utf-8'
31     assert lookup('Utf-8').name == 'utf-8'
32     assert lookup('UTF-8').name == 'utf-8'
33     assert lookup('utf8').name == 'utf-8'
34     assert lookup('utf8').name == 'utf-8'
35     assert lookup('utf8 ').name == 'utf-8'
36     assert lookup(' \r\nutf8\t').name == 'utf-8'
37     assert lookup('u8') is None  # Python label.
38     assert lookup('utf-8 ') is None  # Non-ASCII white space.
40     assert lookup('US-ASCII').name == 'windows-1252'
41     assert lookup('iso-8859-1').name == 'windows-1252'
42     assert lookup('latin1').name == 'windows-1252'
43     assert lookup('LATIN1').name == 'windows-1252'
44     assert lookup('latin-1') is None
45     assert lookup('LATİN1') is None  # ASCII-only case insensitivity.
50         assert decode(b'', label) == ('', lookup(label))
64         assert lookup(name).name == name
77     assert decode(b'\x80', 'latin1') == ('€', lookup('latin1'))
78     assert decode(b'\x80', lookup('latin1')) == ('€', lookup('latin1'))
79     assert decode(b'\xc3\xa9', 'utf8') == ('é', lookup('utf8'))
80     assert decode(b'\xc3\xa9', UTF8) == ('é', lookup('utf8'))
81     assert decode(b'\xc3\xa9', 'ascii') == ('Ã©', lookup('ascii'))
82     assert decode(b'\xEF\xBB\xBF\xc3\xa9', 'ascii') == ('é', lookup('utf8'))  # UTF-8 with BOM
84     assert decode(b'\xFE\xFF\x00\xe9', 'ascii') == ('é', lookup('utf-16be'))  # UTF-16-BE with BOM
85     assert decode(b'\xFF\xFE\xe9\x00', 'ascii') == ('é', lookup('utf-16le'))  # UTF-16-LE with BOM
86     assert decode(b'\xFE\xFF\xe9\x00', 'ascii') == ('\ue900', lookup('utf-16be'))
87     assert decode(b'\xFF\xFE\x00\xe9', 'ascii') == ('\ue900', lookup('utf-16le'))
89     assert decode(b'\x00\xe9', 'UTF-16BE') == ('é', lookup('utf-16be'))
90     assert decode(b'\xe9\x00', 'UTF-16LE') == ('é', lookup('utf-16le'))
91     assert decode(b'\xe9\x00', 'UTF-16') == ('é', lookup('utf-16le'))
93     assert decode(b'\xe9\x00', 'UTF-16BE') == ('\ue900', lookup('utf-16be'))
94     assert decode(b'\x00\xe9', 'UTF-16LE') == ('\ue900', lookup('utf-16le'))
95     assert decode(b'\x00\xe9', 'UTF-16') == ('\ue900', lookup('utf-16le'))
152     assert decode(encoded, 'x-user-defined') == (decoded, lookup('x-user-defined'))