def test_decompress(self):
"""
Test decompression
"""
data = '-\x00\x00\x00+\x00\x00\x00LZFu\xf1\xc5\xc7\xa7\x03\x00\n\x00' \
'rcpg125B2\n\xf3 hel\t\x00 bw\x05\xb0ld}\n\x80\x0f\xa0'
self.assertEqual(decompress(data),
'{\\rtf1\\ansi\\ansicpg1252\\pard hello world}\r\n')
# test raw decompression
data = '.\x00\x00\x00"\x00\x00\x00MELA \xdf\x12\xce{\\rtf1\\ansi\\an' \
'sicpg1252\\pard test}'
self.assertEqual(decompress(data),
'{\\rtf1\\ansi\\ansicpg1252\\pard test}')
def test_decompress(self):
"""
Test decompression
"""
data = '-\x00\x00\x00+\x00\x00\x00LZFu\xf1\xc5\xc7\xa7\x03\x00\n\x00' \
'rcpg125B2\n\xf3 hel\t\x00 bw\x05\xb0ld}\n\x80\x0f\xa0'
self.assertEqual(
decompress(data),
'{\\rtf1\\ansi\\ansicpg1252\\pard hello world}\r\n')
# test raw decompression
data = '.\x00\x00\x00"\x00\x00\x00MELA \xdf\x12\xce{\\rtf1\\ansi\\an' \
'sicpg1252\\pard test}'
self.assertEqual(
decompress(data),
'{\\rtf1\\ansi\\ansicpg1252\\pard test}')
def email_message(self, policy=email.policy.default):
if self.mfs.get('PidTagCodepage'):
charset = str(Codepage(self.mfs.get('PidTagCodepage')))
email_message = message_from_string(self.mfs.get('PidTagHeader'),
policy=policy)
email_message.clear_content()
if self.mfs.get('PidTagBody'):
email_message.add_alternative(self.mfs.get('PidTagBody'),
charset=charset,
subtype='plain')
if self.mfs.get('PidTagBodyHtml'):
email_message.add_alternative(
self.mfs.get('PidTagBodyHtml').encode('utf-8'),
maintype='text',
subtype='html')
if self.mfs.get('PidTagRtfCompressed'):
email_message.add_alternative(compressed_rtf.decompress(
self.mfs.get('PidTagRtfCompressed')),
maintype='application',
subtype='rtf')
for attachment in self.attachments:
with attachment.open() as fh:
email_message.add_attachment(fh.read(),
maintype='application',
subtype='octet-stream',
filename=attachment.filename)
return email_message
def processFile(filepath, filename, prefix='', _canceler=canceler.FAKE):
messa = extract_msg.Message(filepath, prefix,
Attachment) # Create Message object.
header = getHeader(messa) # Gets the header.
formatted = getFormattedHeader(
header) # The next few lines format the header
headerString = '<p class=MsoNormal>'
for u in headerVals:
headerString = headerString + formatted[u]
headerString = headerString + '</p><br>'
ofilename = filename
filename = mkdir(filename)
os.chdir(filename)
messa.save_attachments(True) # Saves the attatchments
if messa.htmlBody != None: # Has html body?
with open('output.html', 'wb') as o:
o.write(messa.htmlBody)
else:
rtfContents = compressed_rtf.decompress(
messa.compressedRtf
) # Read contents, decompress them, and store them in a variable.
with open('out.rtf', 'wb') as rtfFile:
rtfFile.write(rtfContents)
callNode(ofilename, filepath, filename)
if not debug:
os.remove('out.rtf')
addHeader(headerString)
embedImages()
callToPdf(filename)
if not debug:
os.remove('output.html')
os.chdir('..') # Move back to the parent directory
def load_message_stream(self, entry, is_top_level, doc):
# Load stream data.
props = None
props = self.__parse_properties(entry['__properties_version1.0'],
is_top_level, entry, doc)
# Construct the MIME message....
self._MIME_MSG = None
self._MIME_MSG = email.message.EmailMessage()
# Add the raw headers, if known.
if 'TRANSPORT_MESSAGE_HEADERS' in props:
self.__add_transport_headers(props)
else:
props = self.__add_common_headers(props)
# Add the plain-text body from the BODY field.
if 'BODY' in props:
body = props['BODY']
if isinstance(body, str):
self._MIME_MSG.set_content(body, cte='quoted-printable')
else:
self._MIME_MSG.set_content(body,
maintype="text",
subtype="plain",
cte='8bit')
# Plain-text is not availabe. Use the rich text version.
else:
doc.rtf_attachments += 1
fn = "messagebody_{}.rtf".format(doc.rtf_attachments)
self._MIME_MSG.set_content(
"<no plain text message body --- see attachment {}>".format(
fn),
cte='quoted-printable')
# Decompress the value to Rich Text Format.
from compressed_rtf import compress, decompress
rtf = props['RTF_COMPRESSED']
rtf = decompress(rtf)
# Add RTF file as an attachment.
self._MIME_MSG.add_attachment(rtf,
maintype="text",
subtype="rtf",
filename=fn)
# # Copy over string values of remaining properties as headers
# # so we don't lose any information.
# for k, v in props.items():
# if k == 'RTF_COMPRESSED': continue # not interested, save output
# msg[k] = str(v)
# Add attachments.
for stream in entry:
if stream.name.startswith("__attach_version1.0_#"):
self.__process_attachment(stream, doc)
return self._MIME_MSG
def save_data(in_filename, out_dir, titles, data, meta_data, compressed_rtfs):
"""Save found data to `out_dir`"""
# make out dir if not exists
if not os.path.exists(out_dir):
os.makedirs(out_dir)
# decode compressed RTF
for index, compressed_rtf_data in enumerate(compressed_rtfs):
if not compressed_rtf_data:
continue
filename = '{}_data_{}.rtf'.format(in_filename, index)
print('Saving decompressed RTF data to: {}'.format(filename))
out_data = compressed_rtf.decompress(compressed_rtf_data)
with open(os.path.join(out_dir, filename), 'wb') as out_file:
out_file.write(out_data)
# save attachments and meta files
for index, title in enumerate(titles):
filename = title.decode('utf-8')
print('Saving attachment to: {}'.format(filename))
with open(os.path.join(out_dir, filename), 'wb') as out_file:
out_file.write(data[index])
meta_filename = '{}_meta_{}.raw'.format(title.decode('utf-8'), index)
print('Saving attachment meta file to: {}'.format(meta_filename))
with open(os.path.join(out_dir, meta_filename), 'wb') as out_file:
out_file.write(meta_data[index])
def test_hither_and_thither_long(self):
"""
Test decompression of compressed data larger than 4096
"""
data = '{\\rtf1\\ansi\\ansicpg1252\\pard hello world'
while len(data) < 4096:
data += "testtest"
data += "}"
self.assertEqual(decompress(compress(data, compressed=True)), data)
def rtfbody(self):
if self._rtfbody:
try:
from compressed_rtf import decompress
return decompress(self._rtfbody + b'\x00')
except ImportError:
logger.warning("Returning compressed RTF. Install compressed_rtf to decompress")
return self._rtfbody
else:
return None
def rtfbody(self):
if self._rtfbody:
try:
from compressed_rtf import decompress
return decompress(self._rtfbody + b'\x00')
except ImportError:
logger.warning("Returning compressed RTF. Install compressed_rtf to decompress")
return self._rtfbody
else:
return None
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 test_hither_and_thither(self):
"""
Test decompression of compressed data
"""
data = '{\\rtf1\\ansi\\mac\\deff0\\deftab720'
self.assertEqual(decompress(compress(data, compressed=True)), data)
def _set_properties(self):
property_values = self._message.properties
# setting generally required properties to easily access using MsOxMessage instance.
self.subject = property_values.get("Subject")
header = property_values.get("TransportMessageHeaders")
self.header = parse_email_headers(header, True)
self.header_dict = parse_email_headers(header) or {}
self.created_date = property_values.get("CreationTime")
self.received_date = property_values.get("ReceiptTime")
sent_date = property_values.get("DeliverTime")
if not sent_date:
sent_date = self.header_dict.get("Date")
self.sent_date = sent_date
sender_address = self.header_dict.get("From")
if not sender_address:
sender_address = property_values.get(
"SenderRepresentingSmtpAddress")
self.sender = sender_address
reply_to_address = self.header_dict.get("Reply-To")
if not reply_to_address:
reply_to_address = property_values.get("ReplyRecipientNames")
self.reply_to = reply_to_address
self.message_id = property_values.get("InternetMessageId")
to_address = self.header_dict.get("TO")
if not to_address:
to_address = property_values.get("DisplayTo")
if not to_address:
to_address = property_values.get(
"ReceivedRepresentingSmtpAddress")
self.to = to_address
cc_address = self.header_dict.get("CC")
# if cc_address:
# cc_address = [CONTROL_CHARS.sub(" ", cc_add) for cc_add in cc_address.split(",")]
self.cc = cc_address
bcc_address = self.header_dict.get("BCC")
self.bcc = bcc_address
# prefer HTMl over plain text
if "Html" in property_values:
self.body = property_values.get("Html")
else:
self.body = property_values.get("Body")
if not self.body and "RtfCompressed" in property_values:
try:
import compressed_rtf
except ImportError:
compressed_rtf = None
if compressed_rtf:
compressed_rtf_body = property_values['RtfCompressed']
self.body = compressed_rtf.decompress(compressed_rtf_body)
def test_hither_and_thither(self):
"""
Test decompression of compressed data
"""
data = '{\\rtf1\\ansi\\mac\\deff0\\deftab720'
self.assertEqual(decompress(compress(data, compressed=True)), data)
def load_message_stream(entry: CompoundFileEntity, is_top_level: bool,
doc: CompoundFileReader):
# Load stream data.
props = parse_properties(entry["__properties_version1.0"], is_top_level,
entry, doc)
# Construct the MIME message....
msg = email.message.EmailMessage()
# Add the raw headers, if known.
if "TRANSPORT_MESSAGE_HEADERS" in props:
# Get the string holding all of the headers.
headers = props["TRANSPORT_MESSAGE_HEADERS"]
if isinstance(headers, bytes):
headers = headers.decode("utf-8")
# Remove content-type header because the body we can get this
# way is just the plain-text portion of the email and whatever
# Content-Type header was in the original is not valid for
# reconstructing it this way.
headers = re.sub("Content-Type: .*(\n\\s.*)*\n", "", headers, re.I)
# Parse them.
headers = email.parser.HeaderParser(
policy=email.policy.default).parsestr(headers)
# Copy them into the message object.
for header, value in headers.items():
msg[header] = value
else:
# Construct common headers from metadata.
if "MESSAGE_DELIVERY_TIME" in props:
msg["Date"] = formatdate(
props["MESSAGE_DELIVERY_TIME"].timestamp())
del props["MESSAGE_DELIVERY_TIME"]
if "SENDER_NAME" in props:
if "SENT_REPRESENTING_NAME" in props:
if props["SENT_REPRESENTING_NAME"]:
if props["SENDER_NAME"] != props["SENT_REPRESENTING_NAME"]:
props["SENDER_NAME"] += (
" (" + props["SENT_REPRESENTING_NAME"] + ")")
del props["SENT_REPRESENTING_NAME"]
if props["SENDER_NAME"]:
msg["From"] = formataddr((props["SENDER_NAME"], ""))
del props["SENDER_NAME"]
if "DISPLAY_TO" in props:
if props["DISPLAY_TO"]:
msg["To"] = props["DISPLAY_TO"]
del props["DISPLAY_TO"]
if "DISPLAY_CC" in props:
if props["DISPLAY_CC"]:
msg["CC"] = props["DISPLAY_CC"]
del props["DISPLAY_CC"]
if "DISPLAY_BCC" in props:
if props["DISPLAY_BCC"]:
msg["BCC"] = props["DISPLAY_BCC"]
del props["DISPLAY_BCC"]
if "SUBJECT" in props:
if props["SUBJECT"]:
msg["Subject"] = props["SUBJECT"]
del props["SUBJECT"]
# Add the plain-text body from the BODY field.
if "BODY" in props:
body = props["BODY"]
if isinstance(body, str):
msg.set_content(body, cte="quoted-printable")
else:
msg.set_content(body, maintype="text", subtype="plain", cte="8bit")
# Plain-text is not availabe. Use the rich text version.
else:
doc.rtf_attachments += 1
fn = f"messagebody_{doc.rtf_attachments}.rtf"
msg.set_content(
f"<no plain text message body --- see attachment {fn}>",
cte="quoted-printable",
)
# Decompress the value to Rich Text Format.
rtf = props["RTF_COMPRESSED"]
rtf = compressed_rtf.decompress(rtf)
# Add RTF file as an attachment.
msg.add_attachment(rtf, maintype="text", subtype="rtf", filename=fn)
# # Copy over string values of remaining properties as headers
# # so we don't lose any information.
# for k, v in props.items():
# if k == 'RTF_COMPRESSED': continue # not interested, save output
# msg[k] = str(v)
# Add attachments.
for stream in entry:
if stream.name.startswith("__attach_version1.0_#"):
process_attachment(msg, stream, doc)
return msg
def rtfBody(self):
return decompress(self.compressedRtf)
def _set_properties(self):
property_values = self._message.properties
# setting generally required properties to easily access using MsOxMessage instance.
self.subject = property_values.get("Subject")
header = property_values.get("TransportMessageHeaders")
self.header = parse_email_headers(header, True)
self.header_dict = parse_email_headers(header) or {}
self.created_date = property_values.get("CreationTime")
self.received_date = property_values.get("ReceiptTime")
sent_date = property_values.get("DeliverTime")
if not sent_date:
sent_date = self.header_dict.get("Date")
self.sent_date = sent_date
sender_address = self.header_dict.get("From")
if not sender_address:
sender_address = property_values.get("SenderRepresentingSmtpAddress")
self.sender = sender_address
reply_to_address = self.header_dict.get("Reply-To")
if not reply_to_address:
reply_to_address = property_values.get("ReplyRecipientNames")
self.reply_to = reply_to_address
self.message_id = property_values.get("InternetMessageId")
to_address = self.header_dict.get("TO")
if not to_address:
to_address = property_values.get("DisplayTo")
if not to_address:
to_address = property_values.get("ReceivedRepresentingSmtpAddress")
self.to = to_address
cc_address = self.header_dict.get("CC")
# if cc_address:
# cc_address = [CONTROL_CHARS.sub(" ", cc_add) for cc_add in cc_address.split(",")]
self.cc = cc_address
bcc_address = self.header_dict.get("BCC")
self.bcc = bcc_address
# prefer HTMl over plain text
if "Html" in property_values:
self.body = property_values.get("Html")
else:
self.body = property_values.get("Body")
# Trying to decode body if is bytes obj. This is not the way to go. Quick-fix only.
# See IMAP specs. Use charset-normalizer, cchardet or chardet as last resort.
if isinstance(self.body, bytes):
self.body = self.body.decode("utf-8", "ignore")
if not self.body and "RtfCompressed" in property_values:
try:
import compressed_rtf
except ImportError:
compressed_rtf = None
if compressed_rtf:
compressed_rtf_body = property_values["RtfCompressed"]
self.body = compressed_rtf.decompress(compressed_rtf_body)
def load_message_stream(entry, is_top_level, doc):
# Load stream data.
props = parse_properties(entry['__properties_version1.0'], is_top_level,
entry, doc)
# Construct the MIME message....
msg = email.message.EmailMessage()
# Add the raw headers, if known.
if 'TRANSPORT_MESSAGE_HEADERS' in props:
# Get the string holding all of the headers.
headers = props['TRANSPORT_MESSAGE_HEADERS']
if isinstance(headers, bytes):
headers = headers.decode("utf-8")
# Remove content-type header because the body we can get this
# way is just the plain-text portion of the email and whatever
# Content-Type header was in the original is not valid for
# reconstructing it this way.
headers = re.sub("Content-Type: .*(\n\s.*)*\n", "", headers, re.I)
# Parse them.
headers = email.parser.HeaderParser(policy=email.policy.default)\
.parsestr(headers)
# Copy them into the message object.
for header, value in headers.items():
msg[header] = value
else:
# Construct common headers from metadata.
msg['Date'] = formatdate(props['MESSAGE_DELIVERY_TIME'].timestamp())
del props['MESSAGE_DELIVERY_TIME']
if props['SENDER_NAME'] != props['SENT_REPRESENTING_NAME']:
props[
'SENDER_NAME'] += " (" + props['SENT_REPRESENTING_NAME'] + ")"
del props['SENT_REPRESENTING_NAME']
msg['From'] = formataddr((props['SENDER_NAME'], ""))
del props['SENDER_NAME']
msg['To'] = props['DISPLAY_TO']
del props['DISPLAY_TO']
msg['CC'] = props['DISPLAY_CC']
del props['DISPLAY_CC']
msg['BCC'] = props['DISPLAY_BCC']
del props['DISPLAY_BCC']
msg['Subject'] = props['SUBJECT']
del props['SUBJECT']
# Add the plain-text body from the BODY field.
if 'BODY' in props:
body = props['BODY']
if isinstance(body, str):
msg.set_content(body, cte='quoted-printable')
else:
msg.set_content(body, maintype="text", subtype="plain", cte='8bit')
# Plain-text is not availabe. Use the rich text version.
else:
doc.rtf_attachments += 1
fn = "messagebody_{}.rtf".format(doc.rtf_attachments)
msg.set_content(
"<no plain text message body --- see attachment {}>".format(fn),
cte='quoted-printable')
# Decompress the value to Rich Text Format.
import compressed_rtf
rtf = props['RTF_COMPRESSED']
rtf = compressed_rtf.decompress(rtf)
# Add RTF file as an attachment.
msg.add_attachment(rtf, maintype="text", subtype="rtf", filename=fn)
# # Copy over string values of remaining properties as headers
# # so we don't lose any information.
# for k, v in props.items():
# if k == 'RTF_COMPRESSED': continue # not interested, save output
# msg[k] = str(v)
# Add attachments.
for stream in entry:
if stream.name.startswith("__attach_version1.0_#"):
process_attachment(msg, stream, doc)
return msg
def rtfBody(self):
"""
Returns the decompressed Rtf body from the message.
"""
return compressed_rtf.decompress(self.compressedRtf)
