def _process_single_node(self, node):
"""
Extract the contents of a single XML dump node
:param node: The XML node corresponding to a message
:return: An EmailMessage instance containing the message contents
"""
text = unicode(node.find('text').text)
text = unicode.lstrip(text, u'>') # remove leading char that got into the text somehow
if use_full_parser(text):
text = fix_broken_hotmail_headers(text)
parser = Parser()
mime_message = parser.parse(StringIO(text))
return_message = get_nested_payload(mime_message)
else:
return_message = EmailMessage()
subject_node = node.find('subject')
from_node = node.find('from')
to_node = node.find('to')
date_node = node.find('receivedat')
subject = unicode(subject_node.text, 'utf-8') if not subject_node is None else ''
sender = clean_sender('{} <{}>'.format(from_node.find('name').text, from_node.find('email').text))
recipient = clean_recipient('{} <{}>'.format(to_node.find('name').text, to_node.find('email').text))
date_string = '{} {}'.format(date_node.find('date').text, date_node.find('time').text)
return_message.append_body(unicode(text))
return_message.subject = subject
return_message.sender = sender
return_message.recipient = recipient
return_message.date = parse(date_string)
return_message.date = normalize_to_utc(return_message.date, self._timezone)
return_message.source = "XML File {} node {}".format(self._process_path, node.attrib)
return return_message
def get_nested_payload(mime_message):
"""
Returns a single message object from a list of text content and attachments in a MIME message,
after filtering out unwanted content. Also handles nested content like forwarded messages.
:param mime_message: The MIME message to traverse looking for content
:return: A list of plain-text email bodies and a list of base-64 attachments (if any)
"""
return_message = EmailMessage()
return_message.subject = mime_message.get('Subject')
return_message.sender = clean_sender(mime_message.get('From'))
return_message.recipient = clean_recipient(mime_message.get('To'))
return_message.date = parse(mime_message.get('Date'))
for sub_message in mime_message.walk():
content_type = sub_message.get_content_type()
disposition = sub_message.get('Content-Disposition')
if content_type == 'text/plain' and disposition is None:
x = unicode(sub_message.get_payload())
return_message.append_body(x)
elif content_type in _ignored_content_types and disposition is None:
pass # throw away contents we don't want
else:
return_message.add_attachment(sub_message.get_payload(),
content_type=content_type,
filename=disposition)
return return_message
def _process_single_node(self, node):
"""
Extract the contents of a single XML dump node
:param node: The XML node corresponding to a message
:return: An EmailMessage instance containing the message contents
"""
text = unicode(node.find('text').text)
text = unicode.lstrip(
text, u'>') # remove leading char that got into the text somehow
if use_full_parser(text):
text = fix_broken_hotmail_headers(text)
parser = Parser()
mime_message = parser.parse(StringIO(text))
return_message = get_nested_payload(mime_message)
else:
return_message = EmailMessage()
subject_node = node.find('subject')
from_node = node.find('from')
to_node = node.find('to')
date_node = node.find('receivedat')
subject = unicode(subject_node.text,
'utf-8') if not subject_node is None else ''
sender = clean_sender('{} <{}>'.format(
from_node.find('name').text,
from_node.find('email').text))
recipient = clean_recipient('{} <{}>'.format(
to_node.find('name').text,
to_node.find('email').text))
date_string = '{} {}'.format(
date_node.find('date').text,
date_node.find('time').text)
return_message.append_body(unicode(text))
return_message.subject = subject
return_message.sender = sender
return_message.recipient = recipient
return_message.date = parse(date_string)
return_message.date = normalize_to_utc(return_message.date,
self._timezone)
return_message.source = "XML File {} node {}".format(
self._process_path, node.attrib)
return return_message
def get_nested_payload(mime_message):
"""
Returns a single message object from a list of text content and attachments in a MIME message,
after filtering out unwanted content. Also handles nested content like forwarded messages.
:param mime_message: The MIME message to traverse looking for content
:return: A list of plain-text email bodies and a list of base-64 attachments (if any)
"""
return_message = EmailMessage()
return_message.subject = mime_message.get('Subject')
return_message.sender = clean_sender(mime_message.get('From'))
return_message.recipient = clean_recipient(mime_message.get('To'))
return_message.date = parse(mime_message.get('Date'))
for sub_message in mime_message.walk():
content_type = sub_message.get_content_type()
disposition = sub_message.get('Content-Disposition')
if content_type == 'text/plain' and disposition is None:
x = unicode(sub_message.get_payload())
return_message.append_body(x)
elif content_type in _ignored_content_types and disposition is None:
pass # throw away contents we don't want
else:
return_message.add_attachment(sub_message.get_payload(), content_type=content_type, filename=disposition)
return return_message