def _msg_to_eml(self, msg_bytes: bytes) -> EmailMessage:
"""Converts a msg into an eml."""
msg_obj = openMsg(msg_bytes)
# msg obj stores the original raw header here
message, body, attachments = self._extract_msg_objects(msg_obj)
eml = self._build_eml(message, body, attachments)
return eml
def main():
args = parse_arguments()
set_logging(args.verbose, args.debug)
msg_path = args.msg_path
with extract_msg.openMsg(msg_path) as msg:
attachments = None
try:
attachments = get_attachments(msg)
except KeyError as _e:
log.debug("Msg does not have attachments embedded. Likely you used a low quality eml -> msg converter for testing and it provided somewhat broken msg files. Or at least that's when this pops off the most for me.")
if attachments is None:
log.debug("No attachments found in msg.")
else:
log.debug("{0} attachments found in msg.".format(len(attachments)))
raw_rtf = msg.rtfBody
if args.extract_raw:
if args.outfile:
with open(args.outfile, 'wb') as fp:
fp.write(raw_rtf)
else:
print(raw_rtf.decode())
else:
rtf_obj = DeEncapsulator(raw_rtf.decode())
rtf_obj.deencapsulate()
if rtf_obj.content_type == 'html':
print(rtf_obj.html)
else:
print(rtf_obj.text)
def parse_msg(msg=None) -> List[Dict]:
try:
messages = []
if isinstance(msg, str):
msg = extract_msg.openMsg(msg)
attachments = []
for attachment in msg.attachments:
if isinstance(attachment.data, extract_msg.message.Message):
messages += parse_msg(attachment.data)
elif attachment.cid is not None:
attachments.append({
"filename": attachment.longFilename,
"size": len(attachment.data)
})
parsed_message = {
"to":
list({address
for address in parse_email_addresses(msg.to)}),
"from":
parse_email_address(msg.sender),
"recipients":
list({parse_email_address(rec.email)
for rec in msg.recipients}),
"emails_in_body":
list({address
for address in parse_email_addresses(msg.body)}),
"subject":
msg.subject if msg.subject is not None else "",
"body":
remove_email_address(msg.body),
"date":
parse_datetime(msg.date),
"messageID":
msg.messageId if msg.messageId is not None else "",
"inReplyTo":
msg.inReplyTo if msg.inReplyTo is not None else "",
"attachments":
attachments
}
messages.append(parsed_message)
return messages
except Exception as e:
logger.error(f"Failed to parse .msg file: {msg}. Exception: {e}")
return []
def run(config: config_loader.Config) -> None:
if config.cli_args.file and config.cli_args.file.exists():
file_name = config.cli_args.file.name
with open(config.cli_args.file, mode="rb") as rtf_file:
rp = Rtf_Parser(rtf_file=rtf_file)
rp.parse_file()
elif config.cli_args.msg:
file_name = config.cli_args.msg.name
msg = em.openMsg(f"{config.cli_args.msg}")
for attachment in msg.attachments:
with open(config.html / f"{attachment.longFilename}", mode="wb") as att_file:
att_file.write(attachment.data)
decompressed_rtf = cr.decompress(msg.compressedRtf)
with open((config.email_rtf / config.cli_args.msg.name).with_suffix(".rtf"), mode="wb") as email_rtf:
email_rtf.write(decompressed_rtf)
with io.BytesIO(decompressed_rtf) as rtf_file:
rp = Rtf_Parser(rtf_file=rtf_file)
rp.parse_file()
if config.cli_args.de_encapsulate_html:
de_encapsulate(rp, (config.html / file_name).with_suffix(".html"))
def get_text_with_msg(self) -> str:
return extract_msg.openMsg(self.path)
contact = ''
contactName = ''
contactEmail = ''
contactPhone = ''
contactMob = ''
contactAddress = ''
contactFax = ''
emailbody= ' '
signature=''
address=''
temp = ''
addressNo=0
#reads .msg file
msg = extract_msg.openMsg('test.msg')
contact = msg.sender
#split name and email address
splitContact = contact.split('<')
#removes any whitespace to match with text in bosy
contactName = re.search("(^(\w+)(\-|\s)(\w+))",splitContact[0])
contactName = contactName[0].strip()
contactEmail = splitContact[1].replace('>','')
emailBody,temp,signature = msg.body.partition(contactName)
# Find Telephone number , grabs first number found . Wont grab second telephone number e.g. FAX
contactPhone = re.search("(\+\d{2}\s\d{1}\s\d{4}\s\d{4})|(\d{2}\s\d{4}\s\d{4})|(\d{3}[-\.\s]??\d{3}[-\.\s]??\d{4}|\(\d{3}\)\s*\d{3}[-\.\s]??\d{4}|\d{3}[-\.\s]??\d{4})",signature)
contactPhone =(contactPhone[0])
print(contactPhone)