def process_folders(self):
for path, files in self.message_pack.items():
self.current_relpath = self.get_rel_path(path)
for f in files:
if CommonMethods.get_chunksize(
) != 0 and CommonMethods.get_chunksize() == self.chunks:
# Render the folder and reopen
self._fldr_render_reopen(path)
self.chunks = 0
self.cur_fn = f
self.message_generator(os.path.join(path, f))
self._fldr_render(path)
self.account.close_account()
if CommonMethods.get_stitch():
self.account.stitch_account()
def message_generator(self, path):
"""
This is the main method that extracts email messages from an mbox.
:type path: str
:param path:
:return:
"""
b_mark = None
buff = []
with open(path, 'rb') as fh:
# Open the mbox found at path
while True:
line = CommonMethods.sanitize(fh.readline())
if len(line) == 0:
# Clunky ass way to find end of file, but whatevs. write the final message and clear
# buffer.
self._transform_buffer(buff, path)
buff = []
break
if re.search(b'^From((\s(\"|.+).+\@)|(\s(\".+\")\s))', line):
# Per RFC
if b_mark is None:
# Found the beginning of a message
# set the beginning bit, and put everything else, until the next 'From ' block,
# into a buffer.
b_mark = 1
else:
# Process the buffered message into an email.message.Message object
b_mark = None
if CommonMethods.get_chunksize(
) != 0 and CommonMethods.get_chunksize(
) == self.chunks:
# Render the folder and reopen
self._fldr_render_reopen(path)
self.chunks = 0
self._transform_buffer(buff, path, fh.tell())
buff = []
buff.append(line)
def __init__(self, root_level, xml_dir, account_name):
"""Constructor for DirectoryWalker"""
self.mbx = None # type: mailbox.mbox
self.root = root_level
self.folders = {}
self.messages = []
self.current_relpath = None # type: str
self.xml_dir = xml_dir
self.account = Account(account_name, xml_dir)
self.logger = logging.getLogger("MboxWalker")
self.total_messages_processed = 0 # type: int
self.chunks = CommonMethods.get_chunksize() # type: int
self.tracking_pos = 0 # type: int
self.messages_in_folder = 0 # type: int
self.messages_no_start_fldr = 0 # type: int
self.message_no_end_flder = 0 # type: int
self.new_account = True
self.mboxes = [] # type: list
self.new_folder = False
self.mesg_begin = re.compile('^From((\s(\"|.+).+\@)|(\s(\".+\")\s))')
self.json_folders = []
if CommonMethods.get_store_json():
self.json_write = CommonMethods.get_json_directory()
def start_account(self):
if CommonMethods.get_chunksize() != 0:
self._start_account_chunks()
return
self.current_eaxs_file = os.path.join(self.xml_loc, self.xml_name)
self._write_file()