def test_xml_encoding(self):
"""A character encoding is found via the meta tag."""
encoding = get_html_media_encoding(
b"""
<?xml version="1.0" encoding="ascii"?>
<html>
</html>
""",
"text/html",
)
self.assertEqual(encoding, "ascii")
def test_meta_charset_underscores(self):
"""A character encoding contains underscore."""
encoding = get_html_media_encoding(
b"""
<html>
<head><meta charset="Shift_JIS">
</head>
</html>
""",
"text/html",
)
self.assertEqual(encoding, "Shift_JIS")
def test_meta_xml_encoding(self):
"""Meta tags take precedence over XML encoding."""
encoding = get_html_media_encoding(
b"""
<?xml version="1.0" encoding="ascii"?>
<html>
<head><meta charset="UTF-16">
</head>
</html>
""",
"text/html",
)
self.assertEqual(encoding, "UTF-16")
def test_content_type(self):
"""A character encoding is found via the Content-Type header."""
# Test a few variations of the header.
headers = (
'text/html; charset="ascii";',
"text/html;charset=ascii;",
'text/html; charset="ascii"',
"text/html; charset=ascii",
'text/html; charset="ascii;',
'text/html; charset=ascii";',
)
for header in headers:
encoding = get_html_media_encoding(b"", header)
self.assertEqual(encoding, "ascii")
def test_meta_charset(self):
"""A character encoding is found via the meta tag."""
encoding = get_html_media_encoding(
b"""
<html>
<head><meta charset="ascii">
</head>
</html>
""",
"text/html",
)
self.assertEqual(encoding, "ascii")
# A less well-formed version.
encoding = get_html_media_encoding(
b"""
<html>
<head>< meta charset = ascii>
</head>
</html>
""",
"text/html",
)
self.assertEqual(encoding, "ascii")
def test_fallback(self):
"""A character encoding cannot be found in the body or header."""
encoding = get_html_media_encoding(b"", "text/html")
self.assertEqual(encoding, "utf-8")