def parsemail(mail, logger='none'):
log = logging.getLogger(logger)
if len(mail) == 0:
raise MailParserException('Empty mail')
data = { 'type': 'mail' }
# parse mail
try:
message = email.parser.Parser().parsestr(mail)
except UnicodeEncodeError:
message = email.parser.Parser().parsestr(mail.encode('latin_1'))
# test defects and try to save defect mails
if len(message.defects) != 0:
raise MailParserException("Parser signaled defect:\n %s" % (str(message.defects)))
# encoded word is not decoded here, because it only should appear in the
# display name that is discarded by the last map function
# parse from and sender addresses
addresses = itertools.chain(*(message.get_all(field) for field in ('from', 'sender') if message.has_key(field)))
data['from'] = map(lambda adrs: adrs[1], set(email.utils.getaddresses(addresses)))
log.info("From: %s"%(' '.join(data['from'])))
# parse recipient addresses
addresses = itertools.chain(*(message.get_all(field) for field in ('to', 'cc') if message.has_key(field)))
data['to'] = map(lambda adrs: adrs[1], set(email.utils.getaddresses(addresses)))
log.info("To: %s"%(' '.join(data['to'])))
# parse date and convert it to standard format in UTC
if message.get('Date', None):
try:
# guesses format and parses 10-tuple
parsedtime = email.utils.parsedate_tz(message.get('Date'))
# seconds since epoch
utc_timestamp = calendar.timegm(parsedtime[0:9])-parsedtime[9]
# formatted
data['date'] = time.strftime("%Y-%m-%d %H:%M:%S", time.gmtime(utc_timestamp))
log.info("Date: %s", data['date'])
except Exception as e:
raise MailParserException("Could not convert %s to YYYY-MM-DD hh:mm:ss\n %s" % (message.get('Date'), str(e)))
# format current UTC time
data['upload_date'] = time.strftime("%Y-%m-%d %H:%M:%S", time.gmtime(time.time()))
log.info("upload-date: %s", data['upload_date'])
# add labels
data['labels'] = []
if message.get('Status', None):
if not 'R' in message.get('Status'):
data['labels'].append('unread')
else: data['labels'].append('unread')
if config.autolabels:
labeller = labels.Labeller(path=config.autolabels)
labeller.check(data)
log.info("Labels: %s", ' '.join(data['labels']))
return data
def get_receivers(self):
"""
List of all receiver email-addresses (without the name parts).
This includes addresses from the TO, CC and BCC headers.
"""
message = self.message
to_addrs = (message.get_all('to', []) +
message.get_all('cc', []) +
message.get_all('bcc', []))
to_addrs = [str(addr) for addr in to_addrs]
to_addrs = email.utils.getaddresses(to_addrs)
# addresses only without name parts
return [addr[1] for addr in to_addrs]
def extract_sender(
message: Union[email.message.EmailMessage, email.message.Message]
) -> Optional[str]:
"""
Extract the sender from the message object given.
"""
resent_dates = message.get_all("Resent-Date")
if resent_dates is not None and len(resent_dates) > 1:
raise ValueError("Message has more than one 'Resent-' header block")
elif resent_dates:
sender_header_name = "Resent-Sender"
from_header_name = "Resent-From"
else:
sender_header_name = "Sender"
from_header_name = "From"
# Prefer the sender field per RFC 2822:3.6.2.
if sender_header_name in message:
sender_header = message[sender_header_name]
else:
sender_header = message[from_header_name]
if sender_header is None:
return None
return extract_addresses(sender_header)[0]
def extract_recipients(message: email.message.Message) -> List[str]:
"""
Extract the recipients from the message object given.
"""
recipients = [] # type: List[str]
resent_dates = message.get_all("Resent-Date")
if resent_dates is not None and len(resent_dates) > 1:
raise ValueError("Message has more than one 'Resent-' header block")
elif resent_dates:
recipient_headers = ("Resent-To", "Resent-Cc", "Resent-Bcc")
else:
recipient_headers = ("To", "Cc", "Bcc")
for header in recipient_headers:
for recipient in message.get_all(header, failobj=[]):
recipients.extend(extract_addresses(recipient))
return recipients
def get_mail_addresses(message, header_name):
"""
retrieve all email addresses from one message header
@type message: email.message.Message
@param message: the email message
@type header_name: str
@param header_name: the name of the header, can be 'from', 'to', 'cc' or
any other header containing one or more email addresses
@rtype: list
@returns: a list of the addresses in the form of tuples
C{[(u'Name', '*****@*****.**'), ...]}
>>> import email
>>> import email.mime.text
>>> msg=email.mime.text.MIMEText('The text.', 'plain', 'us-ascii')
>>> msg['From']=email.email.utils.formataddr(('Me', '*****@*****.**'))
>>> msg['To']=email.email.utils.formataddr(('A', '*****@*****.**'))+', '+email.email.utils.formataddr(('B', '*****@*****.**'))
>>> print msg.as_string(unixfrom=False)
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
From: Me <*****@*****.**>
To: A <*****@*****.**>, B <*****@*****.**>
<BLANKLINE>
The text.
>>> get_mail_addresses(msg, 'from')
[(u'Me', '*****@*****.**')]
>>> get_mail_addresses(msg, 'to')
[(u'A', '*****@*****.**'), (u'B', '*****@*****.**')]
"""
addrs = email.utils.getaddresses(
[_friendly_header(h) for h in message.get_all(header_name, [])])
for i, (addr_name, addr) in enumerate(addrs):
if not addr_name and addr:
# only one string! Is it the address or the address name ?
# use the same for both and see later
addr_name = addr
if is_usascii(addr):
# address must be ascii only and must match address regex
if not email_address_re.match(addr):
addr = ''
else:
addr = ''
addrs[i] = (decode_mail_header(addr_name), addr)
return addrs
def get_mail_addresses(message, header_name):
"""
retrieve all email addresses from one message header
@type message: email.message.Message
@param message: the email message
@type header_name: str
@param header_name: the name of the header, can be 'from', 'to', 'cc' or
any other header containing one or more email addresses
@rtype: list
@returns: a list of the addresses in the form of tuples
C{[(u'Name', '*****@*****.**'), ...]}
>>> import email
>>> import email.mime.text
>>> msg=email.mime.text.MIMEText('The text.', 'plain', 'us-ascii')
>>> msg['From']=email.email.utils.formataddr(('Me', '*****@*****.**'))
>>> msg['To']=email.email.utils.formataddr(('A', '*****@*****.**'))+', '+email.email.utils.formataddr(('B', '*****@*****.**'))
>>> print msg.as_string(unixfrom=False)
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
From: Me <*****@*****.**>
To: A <*****@*****.**>, B <*****@*****.**>
<BLANKLINE>
The text.
>>> get_mail_addresses(msg, 'from')
[(u'Me', '*****@*****.**')]
>>> get_mail_addresses(msg, 'to')
[(u'A', '*****@*****.**'), (u'B', '*****@*****.**')]
"""
addrs = email.utils.getaddresses(
[_friendly_header(h) for h in message.get_all(header_name, [])])
for i, (addr_name, addr) in enumerate(addrs):
if not addr_name and addr:
# only one string! Is it the address or the address name ?
# use the same for both and see later
addr_name = addr
if is_usascii(addr):
# address must be ascii only and must match address regex
if not email_address_re.match(addr):
addr = ''
else:
addr = ''
addrs[i] = (decode_mail_header(addr_name), addr)
return addrs
def find_emailgateway_recipient(message: message.Message) -> str:
# We can't use Delivered-To; if there is a X-Gm-Original-To
# it is more accurate, so try to find the most-accurate
# recipient list in descending priority order
recipient_headers = ["X-Gm-Original-To", "Delivered-To",
"Resent-To", "Resent-CC", "To", "CC"]
pattern_parts = [re.escape(part) for part in settings.EMAIL_GATEWAY_PATTERN.split('%s')]
match_email_re = re.compile(".*?".join(pattern_parts))
header_addresses = [str(addr)
for recipient_header in recipient_headers
for addr in message.get_all(recipient_header, [])]
for addr_tuple in getaddresses(header_addresses):
if match_email_re.match(addr_tuple[1]):
return addr_tuple[1]
raise ZulipEmailForwardError("Missing recipient in mirror email")
def find_emailgateway_recipient(message: message.Message) -> str:
# We can't use Delivered-To; if there is a X-Gm-Original-To
# it is more accurate, so try to find the most-accurate
# recipient list in descending priority order
recipient_headers = ["X-Gm-Original-To", "Delivered-To",
"Resent-To", "Resent-CC", "To", "CC"]
pattern_parts = [re.escape(part) for part in settings.EMAIL_GATEWAY_PATTERN.split('%s')]
match_email_re = re.compile(".*?".join(pattern_parts))
header_addresses = [str(addr)
for recipient_header in recipient_headers
for addr in message.get_all(recipient_header, [])]
for addr_tuple in getaddresses(header_addresses):
if match_email_re.match(addr_tuple[1]):
return addr_tuple[1]
raise ZulipEmailForwardError("Missing recipient in mirror email")
def find_emailgateway_recipient(message: message.Message) -> str:
# We can't use Delivered-To; if there is a X-Gm-Original-To
# it is more accurate, so try to find the most-accurate
# recipient list in descending priority order
recipient_headers = ["X-Gm-Original-To", "Delivered-To", "To"]
recipients = [] # type: List[Union[str, Header]]
for recipient_header in recipient_headers:
r = message.get_all(recipient_header, None)
if r:
recipients = r
break
pattern_parts = [re.escape(part) for part in settings.EMAIL_GATEWAY_PATTERN.split('%s')]
match_email_re = re.compile(".*?".join(pattern_parts))
for recipient_email in [str(recipient) for recipient in recipients]:
if match_email_re.match(recipient_email):
return recipient_email
raise ZulipEmailForwardError("Missing recipient in mirror email")
def find_emailgateway_recipient(message: message.Message) -> str:
# We can't use Delivered-To; if there is a X-Gm-Original-To
# it is more accurate, so try to find the most-accurate
# recipient list in descending priority order
recipient_headers = ["X-Gm-Original-To", "Delivered-To", "To"]
recipients = [] # type: List[Union[str, Header]]
for recipient_header in recipient_headers:
r = message.get_all(recipient_header, None)
if r:
recipients = r
break
pattern_parts = [re.escape(part) for part in settings.EMAIL_GATEWAY_PATTERN.split('%s')]
match_email_re = re.compile(".*?".join(pattern_parts))
for recipient_email in [str(recipient) for recipient in recipients]:
if match_email_re.match(recipient_email):
return recipient_email
raise ZulipEmailForwardError("Missing recipient in mirror email")
def test_dkim_and_feedback_loop(self):
privkey = self._get_dkim_privkey()
mailing = factories.MailingFactory(dkim={'selector': 'mail', 'domain': 'unittest.cloud-mailing.net', 'privkey':privkey},
feedback_loop={'dkim': {'selector': 'mail', 'domain': 'unittest.cloud-mailing.net', 'privkey':privkey},
'sender_id': 'CloudMailing'})
recipient = factories.RecipientFactory(mailing=mailing)
message_str = self._customize(recipient)
self.assertNotIn(b"\r\n", message_str)
parser = email.parser.Parser()
message = parser.parsestr(message_str, headersonly=False)
assert (isinstance(message, email.message.Message))
self.assertTrue('Feedback-ID' in message)
self.assertEqual(2, len(message.get_all('DKIM-Signature')))
d = dkim.DKIM(message_str)
self.assertTrue(d.verify(0, dnsfunc=self._get_txt))
self.assertTrue(d.verify(1, dnsfunc=self._get_txt))
