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 _process_multipart_eml(file_path):
"""
Given an EML file, clean it up, parse it, and extract
the contents we want to keep.
:param file_path: The path to the EML file to process
:return: A structured EmailMessage instance
"""
with codecs.open(file_path, 'rb', 'windows-1252') as text_file:
text = unicode(''.join(text_file.readlines()))
if use_full_parser(text):
text = fix_broken_yahoo_headers(text)
parser = Parser()
mime_message = parser.parse(StringIO(text))
return_message = get_nested_payload(mime_message)
return_message.source = "EML File {}".format(file_path)
return return_message
def _process_multipart_eml(file_path):
"""
Given an EML file, clean it up, parse it, and extract
the contents we want to keep.
:param file_path: The path to the EML file to process
:return: A structured EmailMessage instance
"""
with codecs.open(file_path, 'rb', 'windows-1252') as text_file:
text = unicode(''.join(text_file.readlines()))
if use_full_parser(text):
text = fix_broken_yahoo_headers(text)
parser = Parser()
mime_message = parser.parse(StringIO(text))
return_message = get_nested_payload(mime_message)
return_message.source = "EML File {}".format(file_path)
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