def various_encodings_test():
v = '"=?utf-8?b?6ICD5Y+W5YiG5Lqr?=" <*****@*****.**>'
eq_(u'"考取分享" <*****@*****.**>', encodedword.mime_to_unicode(v))
v = """=?UTF-8?B?0JbQtdC60LA=?= <*****@*****.**>, =?UTF-8?B?0JrQvtC90YbQtdCy0L7QuQ==?= <*****@*****.**>"""
eq_(u"Жека <*****@*****.**>, Концевой <*****@*****.**>", encodedword.mime_to_unicode(v))
v = encodedword.mime_to_unicode("=?utf-8?b?0JrQvtC90YbQtdCy0L7QuQ==?= <*****@*****.**>, Bob <*****@*****.**>, =?utf-8?b?0JLQuNC90YE=?= <*****@*****.**>")
eq_(u"Концевой <*****@*****.**>, Bob <*****@*****.**>, Винс <*****@*****.**>", v)
v = '=?UTF-8?B?0J/RgNC+0LLQtdGA0Y/QtdC8INGA0YPRgdGB0LrQuNC1INGB0LDQsdC2?=\n =?UTF-8?B?0LXQutGC0Ysg0Lgg0Y7QvdC40LrQvtC0IOKYoA==?='
eq_(u'Проверяем русские сабжекты и юникод ☠', encodedword.mime_to_unicode(v))
v = '=?UTF-8?B?0J/RgNC+0LLQtdGA0Y/QtdC8INGA0YPRgdGB0LrQuNC1INGB0LDQsdC2?=\r\n =?UTF-8?B?0LXQutGC0Ysg0Lgg0Y7QvdC40LrQvtC0IOKYoA==?='
eq_(u'Проверяем русские сабжекты и юникод ☠', encodedword.mime_to_unicode(v))
v = '=?utf-8?Q?Evaneos-Concepci=C3=B3n.pdf?='
eq_(u'Evaneos-Concepción.pdf', encodedword.mime_to_unicode(v))
v = u'=?gb2312?Q?Hey_There=D7=B2=D8=B0?='
eq_(u'Hey There撞匕', encodedword.mime_to_unicode(v))
v = u'=?gb18030?Q?Hey_There=D7=B2=D8=B0?='
eq_(u'Hey There撞匕', encodedword.mime_to_unicode(v))
v = parse(u'Тест длинного дисплей нейма <*****@*****.**>')
eq_(v.display_name, encodedword.mime_to_unicode(v.ace_display_name))
def neutral_headings_test():
v = """from mail-iy0-f179.google.com (mail-iy0-f179.google.com
\t[209.85.210.179])
\tby mxa.mailgun.org (Postfix) with ESMTP id 2D0D3F01116
\tfor <*****@*****.**>; Fri, 17 Dec 2010 12:50:07 +0000 (UTC)"""
eq_(
u"from mail-iy0-f179.google.com (mail-iy0-f179.google.com\t[209.85.210.179])\tby mxa.mailgun.org (Postfix) with ESMTP id 2D0D3F01116\tfor <*****@*****.**>; Fri, 17 Dec 2010 12:50:07 +0000 (UTC)",
encodedword.mime_to_unicode(v),
)
v = '''multipart/mixed; boundary="===============7553021138737466228=="'''
eq_(v, encodedword.mime_to_unicode(v))
def happy_mime_to_unicode_test():
v = """ =?utf-8?B?U2ltcGxlIHRleHQuIEhvdyBhcmUgeW91PyDQmtCw0Log0YLRiyDQv9C+0LY=?=\n =?utf-8?B?0LjQstCw0LXRiNGMPw==?="""
eq_(u'Simple text. How are you? Как ты поживаешь?', encodedword.mime_to_unicode(v))
v = ' =?US-ASCII?Q?Foo?= <*****@*****.**>'
eq_(u'Foo <*****@*****.**>', encodedword.mime_to_unicode(v))
v = '''=?UTF-8?Q?=D1=80=D1=83=D1=81=D1=81=D0=BA=D0=B8=D0=B9?=\n =?UTF-8?Q?_=D0=B8?= english112 =?UTF-8?Q?=D1=81=D0=B0=D0=B1=D0=B6?= subject'''
eq_(u'русский и english112 сабж subject', encodedword.mime_to_unicode(v))
v = '=?iso-8859-1?B?SOlhdnkgTel05WwgVW7uY/hk?=\n\t=?iso-8859-1?Q?=E9?='
eq_(u'Héavy Métål Unîcødé', encodedword.mime_to_unicode(v))
def aol_encodings_test():
v = ''' =?utf-8?Q?=D0=AD=D1=82=D0=BE_=D1=81=D0=BE=D0=BE=D0=B1=D1=89=D0=B5=D0=BD?=
=?utf-8?Q?=D0=B8=D0=B5_=D1=81_=D0=B4=D0=BB=D0=B8=D0=BD=D0=BD=D1=8B=D0=BC?=
=?utf-8?Q?_=D1=81=D0=B0=D0=B1=D0=B6=D0=B5=D0=BA=D1=82=D0=BE=D0=BC_=D1=81?=
=?utf-8?Q?=D0=BF=D0=B5=D1=86=D0=B8=D0=B0=D0=BB=D1=8C=D0=BD=D0=BE_=D1=87?=
=?utf-8?Q?=D1=82=D0=BE=D0=B1=D1=8B_=D0=BF=D1=80=D0=BE=D0=B2=D0=B5=D1=80?=
=?utf-8?Q?=D0=B8=D1=82=D1=8C_=D0=BA=D0=BE=D0=B4=D0=B8=D1=80=D0=BE=D0=B2?=
=?utf-8?Q?=D0=BA=D0=B8?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def __init__(self, raw_display_name=None, raw_addr_spec=None,
_display_name=None, _mailbox=None, _hostname=None):
raw_display_name = _to_parser_input(raw_display_name)
raw_addr_spec = _to_parser_input(raw_addr_spec)
if raw_display_name and raw_addr_spec:
mailbox = addr_spec_parser.parse(raw_addr_spec, lexer.clone())
self._display_name = _to_text(raw_display_name)
self._mailbox = _to_text(mailbox.local_part)
self._hostname = _to_text(mailbox.domain)
elif raw_display_name:
mailbox = mailbox_parser.parse(raw_display_name, lexer.clone())
self._display_name = _to_text(mailbox.display_name)
self._mailbox = _to_text(mailbox.local_part)
self._hostname = _to_text(mailbox.domain)
elif raw_addr_spec:
mailbox = addr_spec_parser.parse(raw_addr_spec, lexer.clone())
self._display_name = u''
self._mailbox = _to_text(mailbox.local_part)
self._hostname = _to_text(mailbox.domain)
elif _mailbox and _hostname:
self._display_name = _display_name or u''
self._mailbox = _mailbox
self._hostname = _hostname
else:
raise SyntaxError('failed to create EmailAddress: bad parameters')
# Convert display name to decoded unicode string.
if (self._display_name.startswith('=?') and
self._display_name.endswith('?=')):
self._display_name = mime_to_unicode(self._display_name)
if (self._display_name.startswith('"') and
self._display_name.endswith('"') and
len(self._display_name) > 2):
self._display_name = smart_unquote(self._display_name)
# Convert hostname to lowercase unicode string.
self._hostname = self._hostname.lower()
if self._hostname.startswith('xn--') or '.xn--' in self._hostname:
self._hostname = idna.decode(self._hostname)
if not is_pure_ascii(self._hostname):
idna.encode(self._hostname)
assert isinstance(self._display_name, six.text_type)
assert isinstance(self._mailbox, six.text_type)
assert isinstance(self._hostname, six.text_type)
def concatenate(parts):
""" Concatenates splitted parts of a parameter in a single parameter,
e.g.
URL*0="ftp://";
URL*1="cs.utk.edu/pub/moore/bulk-mailer/bulk-mailer.tar"
becomes:
URL="ftp://cs.utk.edu/pub/moore/bulk-mailer/bulk-mailer.tar"
"""
part = parts[0]
if is_old_style(part):
# old-style parameters do not support any continuations
return encodedword.mime_to_unicode(get_value(part))
else:
return "".join(
decode_new_style(p) for p in partition(parts))
def __init__(self, display_name, spec=None, parsed_name=None):
if spec is None:
spec = display_name
display_name = None
assert(spec)
if parsed_name:
self.display_name = smart_unquote(mime_to_unicode(parsed_name))
elif display_name:
self.display_name = display_name
else:
self.display_name = ''
parts = spec.rsplit('@', 1)
self.mailbox = parts[0]
self.hostname = parts[1].lower()
self.address = self.mailbox + "@" + self.hostname
self.addr_type = self.Type.Email
def various_encodings_test():
v = '"=?utf-8?b?6ICD5Y+W5YiG5Lqr?=" <*****@*****.**>'
eq_(u'"考取分享" <*****@*****.**>', encodedword.mime_to_unicode(v))
v = """=?UTF-8?B?0JbQtdC60LA=?= <*****@*****.**>, =?UTF-8?B?0JrQvtC90YbQtdCy0L7QuQ==?= <*****@*****.**>"""
eq_(u"Жека <*****@*****.**>, Концевой <*****@*****.**>", encodedword.mime_to_unicode(v))
v = encodedword.mime_to_unicode(
"=?utf-8?b?0JrQvtC90YbQtdCy0L7QuQ==?= <*****@*****.**>, Bob <*****@*****.**>, =?utf-8?b?0JLQuNC90YE=?= <*****@*****.**>"
)
eq_(u"Концевой <*****@*****.**>, Bob <*****@*****.**>, Винс <*****@*****.**>", v)
v = "=?UTF-8?B?0J/RgNC+0LLQtdGA0Y/QtdC8INGA0YPRgdGB0LrQuNC1INGB0LDQsdC2?=\n =?UTF-8?B?0LXQutGC0Ysg0Lgg0Y7QvdC40LrQvtC0IOKYoA==?="
eq_(u"Проверяем русские сабжекты и юникод ☠", encodedword.mime_to_unicode(v))
v = "=?UTF-8?B?0J/RgNC+0LLQtdGA0Y/QtdC8INGA0YPRgdGB0LrQuNC1INGB0LDQsdC2?=\r\n =?UTF-8?B?0LXQutGC0Ysg0Lgg0Y7QvdC40LrQvtC0IOKYoA==?="
eq_(u"Проверяем русские сабжекты и юникод ☠", encodedword.mime_to_unicode(v))
v = u"=?utf-8?Q?Evaneos-Concepci=C3=B3n.pdf?="
eq_(u"Evaneos-Concepción.pdf", encodedword.mime_to_unicode(v))
def missing_padding_mime_to_unicode_test():
v = """ =?utf-8?B?U2ltcGxlIHRleHQuIEhvdyBhcmUgeW91PyDQmtCw0Log0YLRiyDQv9C+0LY?=\n =?utf-8?B?0LjQstCw0LXRiNGMPw?="""
eq_(u'Simple text. How are you? Как ты поживаешь?', encodedword.mime_to_unicode(v))
def gmail_encodings_test():
v = ''' =?KOI8-R?B?/NTPINPPz8Ldxc7JxSDTIMTMyc7O2c0g08HC1g==?=
=?KOI8-R?B?xcvUz80g09DFw8nBzNjOzyDe1M/C2SDQ0s/XxdLJ1A==?=
=?KOI8-R?B?2CDLz8TJ0s/Xy8k=?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def lying_encodings_mime_to_unicode_test():
v = '''=?US-ASCII?Q?=D1=80=D1=83=D1=81=D1=81=D0=BA=D0=B8=D0=B9?=\n english112 =?UTF-8?Q?=D1=81=D0=B0=D0=B1=D0=B6?= subject'''
eq_(u'русский english112 сабж subject', encodedword.mime_to_unicode(v))
def hotmail_encodings_test():
v = ''' =?koi8-r?B?/NTPINPPz8LdxQ==?= =?koi8-r?B?zsnFINMgxMzJzg==?=
=?koi8-r?B?ztnNINPBwtbFyw==?= =?koi8-r?B?1M/NINPQxcPJwQ==?=
=?koi8-r?B?zNjOzyDe1M/C2Q==?= =?koi8-r?B?INDSz9fF0snU2A==?=
=?koi8-r?B?IMvPxMnSz9fLyQ==?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def test_error_reporting():
eq_("Sasha", encodedword.mime_to_unicode("Sasha"))
def test_error_reporting():
eq_("Sasha", encodedword.mime_to_unicode("Sasha"))
def yahoo_encodings_test():
v = '''
=?utf-8?B?0K3RgtC+INGB0L7QvtCx0YnQtdC90LjQtSDRgSDQtNC70LjQvdC90YvQvCA=?=
=?utf-8?B?0YHQsNCx0LbQtdC60YLQvtC8INGB0L/QtdGG0LjQsNC70YzQvdC+INGH0YI=?=
=?utf-8?B?0L7QsdGLINC/0YDQvtCy0LXRgNC40YLRjCDQutC+0LTQuNGA0L7QstC60Lg=?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def lying_encodings_mime_to_unicode_test():
v = '''=?US-ASCII?Q?=D1=80=D1=83=D1=81=D1=81=D0=BA=D0=B8=D0=B9?=\n english112 =?UTF-8?Q?=D1=81=D0=B0=D0=B1=D0=B6?= subject'''
eq_(u'русский english112 сабж subject',
encodedword.mime_to_unicode(v))
def hotmail_encodings_test():
v = ''' =?koi8-r?B?/NTPINPPz8LdxQ==?= =?koi8-r?B?zsnFINMgxMzJzg==?=
=?koi8-r?B?ztnNINPBwtbFyw==?= =?koi8-r?B?1M/NINPQxcPJwQ==?=
=?koi8-r?B?zNjOzyDe1M/C2Q==?= =?koi8-r?B?INDSz9fF0snU2A==?=
=?koi8-r?B?IMvPxMnSz9fLyQ==?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def missing_padding_mime_to_unicode_test():
v = """ =?utf-8?B?U2ltcGxlIHRleHQuIEhvdyBhcmUgeW91PyDQmtCw0Log0YLRiyDQv9C+0LY?=\n =?utf-8?B?0LjQstCw0LXRiNGMPw?="""
eq_(u'Simple text. How are you? Как ты поживаешь?',
encodedword.mime_to_unicode(v))
def outlook_encodings_test():
v = '''=?koi8-r?B?/NTPINPPz8Ldxc7JxSDTIMTMyc7O2c0g08HC1sXL1M/NINPQxcPJwQ==?=
=?koi8-r?B?zNjOzyDe1M/C2SDQ0s/XxdLJ1Nggy8/EydLP18vJ?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def outlook_encodings_test():
v = '''=?koi8-r?B?/NTPINPPz8Ldxc7JxSDTIMTMyc7O2c0g08HC1sXL1M/NINPQxcPJwQ==?=
=?koi8-r?B?zNjOzyDe1M/C2SDQ0s/XxdLJ1Nggy8/EydLP18vJ?='''
eq_(
u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки",
encodedword.mime_to_unicode(v))
def test_header_to_unicode():
assert_equal(u'Eugueny ώ Kontsevoy', mime_to_unicode("=?UTF-8?Q?Eugueny_=CF=8E_Kontsevoy?=") )
assert_equal(u'hello', mime_to_unicode("hello"))
assert_equal(None, mime_to_unicode(None))
def test_header_to_unicode():
assert_equal(u'Eugueny ώ Kontsevoy',
mime_to_unicode("=?UTF-8?Q?Eugueny_=CF=8E_Kontsevoy?="))
assert_equal(u'hello', mime_to_unicode("hello"))
assert_equal(None, mime_to_unicode(None))
def gmail_encodings_test():
v = ''' =?KOI8-R?B?/NTPINPPz8Ldxc7JxSDTIMTMyc7O2c0g08HC1g==?=
=?KOI8-R?B?xcvUz80g09DFw8nBzNjOzyDe1M/C2SDQ0s/XxdLJ1A==?=
=?KOI8-R?B?2CDLz8TJ0s/Xy8k=?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def __init__(self, raw_display_name=None, raw_addr_spec=None, display_name=None, mailbox=None, hostname=None):
if isinstance(raw_display_name, unicode):
raw_display_name = raw_display_name.encode('utf-8')
if isinstance(raw_addr_spec, unicode):
raw_addr_spec = raw_addr_spec.encode('utf-8')
if raw_display_name and raw_addr_spec:
parser = addr_spec_parser
mailbox = parser.parse(raw_addr_spec.strip(), lexer=lexer.clone())
self._display_name = raw_display_name
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif raw_display_name:
parser = mailbox_parser
mailbox = parser.parse(raw_display_name.strip(), lexer=lexer.clone())
self._display_name = mailbox.display_name
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif raw_addr_spec:
parser = addr_spec_parser
mailbox = parser.parse(raw_addr_spec.strip(), lexer=lexer.clone())
self._display_name = ''
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif mailbox and hostname:
self._display_name = display_name or ''
self._mailbox = mailbox
self._hostname = hostname
else:
raise SyntaxError('failed to create EmailAddress: bad parameters')
# Convert display name to decoded unicode string.
if (self._display_name.startswith('=?') and
self._display_name.endswith('?=')):
self._display_name = mime_to_unicode(self._display_name)
if (self._display_name.startswith('"') and
self._display_name.endswith('"') and
len(self._display_name) > 2):
self._display_name = smart_unquote(self._display_name)
if isinstance(self._display_name, str):
self._display_name = self._display_name.decode('utf-8')
# Convert localpart to unicode string.
if isinstance(self._mailbox, str):
self._mailbox = self._mailbox.decode('utf-8')
# Convert hostname to lowercase unicode string.
self._hostname = self._hostname.lower()
if self._hostname.startswith('xn--') or '.xn--' in self._hostname:
self._hostname = idna.decode(self._hostname)
if isinstance(self._hostname, str):
self._hostname = self._hostname.decode('utf-8')
if not is_pure_ascii(self._hostname):
idna.encode(self._hostname)
def yahoo_encodings_test():
v = '''
=?utf-8?B?0K3RgtC+INGB0L7QvtCx0YnQtdC90LjQtSDRgSDQtNC70LjQvdC90YvQvCA=?=
=?utf-8?B?0YHQsNCx0LbQtdC60YLQvtC8INGB0L/QtdGG0LjQsNC70YzQvdC+INGH0YI=?=
=?utf-8?B?0L7QsdGLINC/0YDQvtCy0LXRgNC40YLRjCDQutC+0LTQuNGA0L7QstC60Lg=?='''
eq_(u"Это сообщение с длинным сабжектом специально чтобы проверить кодировки", encodedword.mime_to_unicode(v))
def lying_encodings_mime_to_unicode_test():
v = """=?US-ASCII?Q?=D1=80=D1=83=D1=81=D1=81=D0=BA=D0=B8=D0=B9?=\n =?unknonw?Q?_=D0=B8?= english112 =?UTF-8?Q?=D1=81=D0=B0=D0=B1=D0=B6?= subject"""
eq_(u"русский и english112 сабж subject", encodedword.mime_to_unicode(v))
def display_name(self):
if self._display_name is None:
return u''
return mime_to_unicode(self._display_name)
def __init__(self,
raw_display_name=None,
raw_addr_spec=None,
display_name=None,
mailbox=None,
hostname=None):
if isinstance(raw_display_name, unicode):
raw_display_name = raw_display_name.encode('utf-8')
if isinstance(raw_addr_spec, unicode):
raw_addr_spec = raw_addr_spec.encode('utf-8')
if raw_display_name and raw_addr_spec:
parser = addr_spec_parser
mailbox = parser.parse(raw_addr_spec.strip(), lexer=lexer.clone())
self._display_name = raw_display_name
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif raw_display_name:
parser = mailbox_parser
mailbox = parser.parse(raw_display_name.strip(),
lexer=lexer.clone())
self._display_name = mailbox.display_name
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif raw_addr_spec:
parser = addr_spec_parser
mailbox = parser.parse(raw_addr_spec.strip(), lexer=lexer.clone())
self._display_name = ''
self._mailbox = mailbox.local_part
self._hostname = mailbox.domain
elif mailbox and hostname:
self._display_name = display_name or ''
self._mailbox = mailbox
self._hostname = hostname
else:
raise SyntaxError('failed to create EmailAddress: bad parameters')
# Convert display name to decoded unicode string.
if (self._display_name.startswith('=?')
and self._display_name.endswith('?=')):
self._display_name = mime_to_unicode(self._display_name)
if (self._display_name.startswith('"')
and self._display_name.endswith('"')
and len(self._display_name) > 2):
self._display_name = smart_unquote(self._display_name)
if isinstance(self._display_name, str):
self._display_name = self._display_name.decode('utf-8')
# Convert localpart to unicode string.
if isinstance(self._mailbox, str):
self._mailbox = self._mailbox.decode('utf-8')
# Convert hostname to lowercase unicode string.
self._hostname = self._hostname.lower()
if self._hostname.startswith('xn--') or '.xn--' in self._hostname:
self._hostname = idna.decode(self._hostname)
if isinstance(self._hostname, str):
self._hostname = self._hostname.decode('utf-8')
if not is_pure_ascii(self._hostname):
idna.encode(self._hostname)