def full_spec(self):
"""
Returns an ASCII-compatable encoding of an email address or raises a
ValueError. Display name and domain parts will be converted to
ASCII-compatable encoding. The transformed address will be ASCII-only
and RFC-2822 compliant.
>>> EmailAddress("Ev K", "*****@*****.**").full_spec()
'Ev K <*****@*****.**>'
>>> EmailAddress("Жека", "*****@*****.**").full_spec()
'=?utf-8?b?0JbQtdC60LA=?= <*****@*****.**>'
"""
if not is_pure_ascii(self.mailbox):
raise ValueError(
'address {} has no ASCII-compatable encoding'.format(
self.address.encode('utf-8')))
if not is_pure_ascii(self.hostname):
try:
ace_hostname = idna.encode(self.hostname)
except idna.IDNAError:
raise ValueError(
'address {} has no ASCII-compatable encoding'.format(
self.address.encode('utf-8')))
else:
ace_hostname = self.hostname
if self.display_name:
ace_display_name = smart_quote(
encode_string(None,
self.display_name,
maxlinelen=MAX_ADDRESS_LENGTH))
return u'{} <{}@{}>'.format(ace_display_name, self.mailbox,
ace_hostname)
return u'{}@{}'.format(self.mailbox, ace_hostname)
def ace_address(self):
if not is_pure_ascii(self.mailbox):
raise ValueError('address {} has no ASCII-compatable encoding'
.format(self.address.encode('utf-8')))
ace_hostname = self.hostname
if not is_pure_ascii(self.hostname):
try:
ace_hostname = idna.encode(self.hostname)
except idna.IDNAError:
raise ValueError('address {} has no ASCII-compatable encoding'
.format(self.address.encode('utf-8')))
return '{}@{}'.format(self.mailbox, ace_hostname)
def requires_non_ascii(self):
"""
Can the address be converted to an ASCII compatible encoding?
"""
if not is_pure_ascii(self.mailbox):
return True
if not is_pure_ascii(self.hostname):
try:
idna.encode(self.hostname)
except idna.IDNAError:
return True
return False
def requires_non_ascii(self):
"""
Can the address be converted to an ASCII compatible encoding?
"""
if not is_pure_ascii(self.mailbox):
return True
if not is_pure_ascii(self.hostname):
try:
idna.encode(self.hostname)
except idna.IDNAError:
return True
return False
def parse_header_value(name, val):
if not is_pure_ascii(val):
val = to_unicode(val)
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if val is not None and not is_pure_ascii(val):
raise DecodingError('Non-ascii content header value')
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
return WithParams(val, params)
return val
def ace_address(self):
if not is_pure_ascii(self.mailbox):
raise ValueError(
'address {} has no ASCII-compatable encoding'.format(
self.address.encode('utf-8')))
ace_hostname = self.hostname
if not is_pure_ascii(self.hostname):
try:
ace_hostname = idna.encode(self.hostname)
except idna.IDNAError:
raise ValueError(
'address {} has no ASCII-compatable encoding'.format(
self.address.encode('utf-8')))
return '{}@{}'.format(self.mailbox, ace_hostname)
def parse_header_value(name, val):
if not is_pure_ascii(val):
val = to_unicode(val)
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if val is not None and not is_pure_ascii(val):
raise DecodingError('Non-ascii content header value')
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
return WithParams(val, params)
return val
def parse_header_value(name, val):
if not is_pure_ascii(val):
if parametrized.is_parametrized(name, val):
# PHOENIX HACK!!!
# FIXME: making the function return a completely different type on
# parameterized headers is wrong.
val = val.decode('utf-8', errors='replace')
val = val.split(";", 1)
if len(val) == 2:
val, params = val
rparams = {}
for p in params.split(";"):
k,v = p.split("=", 1)
rparams[k.strip()] = v.split('"')[1]
return val, rparams
return val[0], {}
return to_unicode(val)
else:
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
else:
return WithParams(val, params)
else:
return val
def address_spec(self, stream):
'''
Extract a single address spec from a stream of input, always
operates in strict mode.
'''
# sanity check
if stream is None:
raise ParserException('No input provided to parser.')
if isinstance(stream, str) and not is_pure_ascii(stream):
raise ParserException('ASCII string contains non-ASCII chars.')
# to avoid spinning here forever, limit mailbox length
if len(stream) > MAX_ADDRESS_LENGTH:
raise ParserException('Stream length exceeds maximum allowable ' + \
'address length of ' + str(MAX_ADDRESS_LENGTH) + '.')
self.stream = TokenStream(stream)
addr = self._addr_spec()
if addr:
# optional whitespace
self._whitespace()
# if we hit the end of the stream, we have a valid inbox
if self.stream.end_of_stream():
return addr
return None
def __init__(
self, content_type, body, charset=None, disposition=None, filename=None):
self.headers = headers.MimeHeaders()
self.body = body
self.disposition = disposition or ('attachment' if filename else None)
self.filename = filename
self.size = len(body)
if self.filename:
self.filename = path.basename(self.filename)
content_type = adjust_content_type(content_type, body, filename)
if content_type.main == 'text':
# the text should have a charset
if not charset:
charset = "utf-8"
# it should be stored as unicode. period
self.body = charsets.convert_to_unicode(charset, body)
# let's be simple when possible
if charset != 'ascii' and is_pure_ascii(body):
charset = 'ascii'
self.headers['MIME-Version'] = '1.0'
self.headers['Content-Type'] = content_type
if charset:
content_type.params['charset'] = charset
if self.disposition:
self.headers['Content-Disposition'] = WithParams(disposition)
if self.filename:
self.headers['Content-Disposition'].params['filename'] = self.filename
self.headers['Content-Type'].params['name'] = self.filename
def address_spec(self, stream):
'''
Extract a single address spec from a stream of input, always
operates in strict mode.
'''
# sanity check
if stream is None:
raise ParserException('No input provided to parser.')
if isinstance(stream, str) and not is_pure_ascii(stream):
raise ParserException('ASCII string contains non-ASCII chars.')
# to avoid spinning here forever, limit mailbox length
if len(stream) > MAX_ADDRESS_LENGTH:
raise ParserException('Stream length exceeds maximum allowable ' + \
'address length of ' + str(MAX_ADDRESS_LENGTH) + '.')
self.stream = TokenStream(stream)
addr = self._addr_spec()
if addr:
# optional whitespace
self._whitespace()
# if we hit the end of the stream, we have a valid inbox
if self.stream.end_of_stream():
return addr
return None
def __init__(
self, content_type, body, charset=None, disposition=None, filename=None):
self.headers = headers.MimeHeaders()
self.body = body
self.disposition = disposition or ('attachment' if filename else None)
self.filename = filename
self.size = len(body)
if self.filename:
self.filename = path.basename(self.filename)
content_type = adjust_content_type(content_type, body, filename)
if content_type.main == 'text':
# the text should have a charset
if not charset:
charset = "utf-8"
# it should be stored as unicode. period
self.body = charsets.convert_to_unicode(charset, body)
# let's be simple when possible
if charset != 'ascii' and is_pure_ascii(body):
charset = 'ascii'
self.headers['MIME-Version'] = '1.0'
self.headers['Content-Type'] = content_type
if charset:
content_type.params['charset'] = charset
if self.disposition:
self.headers['Content-Disposition'] = WithParams(disposition)
if self.filename:
self.headers['Content-Disposition'].params['filename'] = self.filename
self.headers['Content-Type'].params['name'] = self.filename
def parse_header(header):
""" Accepts a raw header with name, colons and newlines
and returns it's parsed value
"""
name, val = split2(header)
if not is_pure_ascii(name):
raise DecodingError("Non-ascii header name")
return name, parse_header_value(name, encodedword.unfold(val))
def _whitespace_unicode(self):
"""
Grammar: whitespace-unicode -> whitespace-unicode
"""
uwhite = self.stream.get_token(UNI_WHITE)
if uwhite and not is_pure_ascii(uwhite):
return uwhite
return None
def _whitespace_unicode(self):
"""
Grammar: whitespace-unicode -> whitespace-unicode
"""
uwhite = self.stream.get_token(UNI_WHITE)
if uwhite and not is_pure_ascii(uwhite):
return uwhite
return None
def parse_header(header):
""" Accepts a raw header with name, colons and newlines
and returns it's parsed value
"""
name, val = split2(header)
if not is_pure_ascii(name):
raise DecodingError("Non-ascii header name")
return name, parse_header_value(name, encodedword.unfold(val))
def validate_address(addr_spec, metrics=False):
"""
Given an addr-spec, runs the pre-parser, the parser, DNS MX checks,
MX existence checks, and if available, ESP specific grammar for the
local part.
In the case of a valid address returns an EmailAddress object, otherwise
returns None. If requested, will also return the parsing time metrics.
Examples:
>>> address.validate_address('*****@*****.**')
None
>>> address.validate_address('*****@*****.**')
None
>>> address.validate_address('*****@*****.**')
[email protected]
"""
mtimes = {'parsing': 0, 'mx_lookup': 0,
'dns_lookup': 0, 'mx_conn':0 , 'custom_grammar':0}
# sanity check
if addr_spec is None:
return None, mtimes
if not is_pure_ascii(addr_spec):
return None, mtimes
# preparse address into its parts and perform any ESP specific pre-parsing
addr_parts = flanker.addresslib.validate.preparse_address(addr_spec)
if addr_parts is None:
return None, mtimes
# run parser against address
bstart = time.time()
paddr = parse('@'.join(addr_parts), addr_spec_only=True)
mtimes['parsing'] = time.time() - bstart
if paddr is None:
return None, mtimes
# lookup if this domain has a mail exchanger
exchanger, mx_metrics = \
flanker.addresslib.validate.mail_exchanger_lookup(addr_parts[-1], metrics=True)
mtimes['mx_lookup'] = mx_metrics['mx_lookup']
mtimes['dns_lookup'] = mx_metrics['dns_lookup']
mtimes['mx_conn'] = mx_metrics['mx_conn']
if exchanger is None:
return None, mtimes
# lookup custom local-part grammar if it exists
bstart = time.time()
plugin = flanker.addresslib.validate.plugin_for_esp(exchanger)
mtimes['custom_grammar'] = time.time() - bstart
if plugin and plugin.validate(addr_parts[0]) is False:
return None, mtimes
return paddr, mtimes
def validate_address(addr_spec, metrics=False):
'''
Given an addr-spec, runs the pre-parser, the parser, DNS MX checks,
MX existence checks, and if available, ESP specific grammar for the
local part.
In the case of a valid address returns an EmailAddress object, otherwise
returns None. If requested, will also return the parsing time metrics.
Examples:
>>> address.validate_address('*****@*****.**')
None
>>> address.validate_address('*****@*****.**')
None
>>> address.validate_address('*****@*****.**')
[email protected]
'''
mtimes = {'parsing': 0, 'mx_lookup': 0,
'dns_lookup': 0, 'mx_conn':0 , 'custom_grammar':0}
# sanity check
if addr_spec is None:
return None, mtimes
if not is_pure_ascii(addr_spec):
return None, mtimes
# preparse address into its parts and perform any ESP specific pre-parsing
addr_parts = flanker.addresslib.validate.preparse_address(addr_spec)
if addr_parts is None:
return None, mtimes
# run parser against address
bstart = time.time()
paddr = parse('@'.join(addr_parts), addr_spec_only=True)
mtimes['parsing'] = time.time() - bstart
if paddr is None:
return None, mtimes
# lookup if this domain has a mail exchanger
exchanger, mx_metrics = \
flanker.addresslib.validate.mail_exchanger_lookup(addr_parts[-1], metrics=True)
mtimes['mx_lookup'] = mx_metrics['mx_lookup']
mtimes['dns_lookup'] = mx_metrics['dns_lookup']
mtimes['mx_conn'] = mx_metrics['mx_conn']
if exchanger is None:
return None, mtimes
# lookup custom local-part grammar if it exists
bstart = time.time()
plugin = flanker.addresslib.validate.plugin_for_esp(exchanger)
mtimes['custom_grammar'] = time.time() - bstart
if plugin and plugin.validate(addr_parts[0]) is False:
return None, mtimes
return paddr, mtimes
def parse_header_value(name, val):
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
else:
return WithParams(val, params)
else:
return val if is_pure_ascii(val) else to_unicode(val)
def parse_header_value(name, val):
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
else:
return WithParams(val, params)
else:
return val if is_pure_ascii(val) else to_unicode(val)
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 __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 parse_header_value(name, val):
if not is_pure_ascii(val):
if parametrized.is_parametrized(name, val):
raise DecodingError("Unsupported value in content- header")
return to_unicode(val)
else:
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
else:
return WithParams(val, params)
else:
return val
def parse_header_value(name, val):
if not is_pure_ascii(val):
if parametrized.is_parametrized(name, val):
raise DecodingError("Unsupported value in content- header")
return to_unicode(val)
else:
if parametrized.is_parametrized(name, val):
val, params = parametrized.decode(val)
if name == 'Content-Type':
main, sub = parametrized.fix_content_type(val)
return ContentType(main, sub, params)
else:
return WithParams(val, params)
else:
return val
def address_list(self, stream):
'''
Extract a mailbox and/or url list from a stream of input, operates in
strict and relaxed modes.
'''
# sanity check
if not stream:
raise ParserException('No input provided to parser.')
if isinstance(stream, str) and not is_pure_ascii(stream):
raise ParserException('ASCII string contains non-ASCII chars.')
# to avoid spinning here forever, limit address list length
if len(stream) > MAX_ADDRESS_LIST_LENGTH:
raise ParserException('Stream length exceeds maximum allowable ' + \
'address list length of ' + str(MAX_ADDRESS_LIST_LENGTH) + '.')
# set stream
self.stream = TokenStream(stream)
if self.strict is True:
return self._address_list_strict()
return self._address_list_relaxed()
def address_list(self, stream):
'''
Extract a mailbox and/or url list from a stream of input, operates in
strict and relaxed modes.
'''
# sanity check
if not stream:
raise ParserException('No input provided to parser.')
if isinstance(stream, str) and not is_pure_ascii(stream):
raise ParserException('ASCII string contains non-ASCII chars.')
# to avoid spinning here forever, limit address list length
if len(stream) > MAX_ADDRESS_LIST_LENGTH:
raise ParserException('Stream length exceeds maximum allowable ' + \
'address list length of ' + str(MAX_ADDRESS_LIST_LENGTH) + '.')
# set stream
self.stream = TokenStream(stream)
if self.strict is True:
return self._address_list_strict()
return self._address_list_relaxed()
def contains_non_ascii(self):
"""
Does the address contain any non-ASCII characters?
"""
return not is_pure_ascii(self.address)
def contains_non_ascii(self):
"""
Does the address contain any non-ASCII characters?
"""
return not is_pure_ascii(self.address)
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 ace_address(self):
if not is_pure_ascii(self._mailbox):
raise ValueError('Address {} has no ASCII-compatable encoding'
.format(self.address))
return _to_str('{}@{}'.format(self._mailbox, self.ace_hostname))
def ace_address(self):
if not is_pure_ascii(self._mailbox):
raise ValueError(
'Address {} has no ASCII-compatable encoding'.format(
self.address))
return _to_str('{}@{}'.format(self._mailbox, self.ace_hostname))
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)