def do_walk(self):
"""
do_walk is the main function of the module.
:return:
"""
# If this is a TOMES_TOOL Struct use the folder_map
if self.from_tomes:
self.account_directory = os.path.join(self.data_dir,
self.account_name)
# Did someone make a mistake? Check to make sure folder_map is there
if os.path.exists(
os.path.join(self.account_directory, "folder_map.tsv")):
self._build_folder_map(
os.path.join(self.account_directory, "folder_map.tsv"))
else:
CommonMethods.set_from_tomes(False)
print("Scanning data structure for emails.")
for root, dirs, files in os.walk(self.account_directory):
for f in files:
if root not in self.message_pack:
self.message_pack[root] = []
if f.endswith("eml"):
self.message_pack[root].append(f)
self.process_folders()
def _arg_parse(self):
parser = argparse.ArgumentParser(description='Convert mbox into XML.')
parser.add_argument('--account',
'-a',
dest='account_name',
required=True,
help='email account name')
parser.add_argument(
'--directory',
'-d',
dest='account_directory',
help='directory to hold all files for this account')
parser.add_argument(
'--chunk',
'-c',
dest='chunk',
default=self.NO_CHUNK,
help=
'An approximate number of messages to put in one output XML file. '
'NOTE: This this will be approximate because '
'DarcMailCLI maintains the integrity of the Folder. A chunk test is made after a '
'folder is processed. If the total number of processed messages is below the chunk '
'size, the next folder will be added to the current XML file. If it exceeds '
'the chunk limit the a new file will be opened. Each file will have the '
'starting LocalID as part of the filename. default = no limit or all messages '
'in one file.')
parser.add_argument(
'--no_subdirectories',
'-n',
dest='no_subdirectories',
action='store_true',
help='do NOT make subdirectories to hold external content'
'(default = make subdirectories)')
parser.add_argument(
'--data-dir',
'-dd',
dest='data_dir',
type=str,
default='attachments',
help='path to store the account attachments. DEFAULT: "attachments"'
)
parser.add_argument('--from-eml',
'-fe',
dest='from_eml',
help='The destination is a series of emls',
action='store_true')
parser.add_argument(
'--stitch',
'-st',
dest='stitch',
action='store_true',
help=
'Stitch can only be used in conjunction with the --chunk switch. The purpose is to '
'reduce memory load. If you find that DarcMailCLI is crashing on your email accounts'
'due to memory errors, but you still want to have a single EAXS file, then chunk the '
'process, which clears memory faster, and stitch will rebuild a single file at the end'
'of the process')
parser.add_argument(
'--tomes_tool',
'-tt',
dest='tomes_tool',
action='store_true',
help=
'Indicates that the source tree is built from the Dockerized pst_extractor.'
)
args = parser.parse_args()
argdict = vars(args)
for k, v in argdict.items():
print("{}:{}".format(k, v))
CommonMethods.set_base_path(CommonMethods.get_process_paths())
self.account_name = argdict['account_name'].strip()
self.account_directory = os.path.join(CommonMethods.get_base_path(),
'mboxes')
if argdict['account_directory']:
self.account_directory = os.path.normpath(
os.path.abspath(argdict['account_directory'].strip()))
# Initialize common features and common attributes
CommonMethods.set_store_rtf_body(False)
CommonMethods.init_hash_dict()
CommonMethods.set_dedupe()
self.eaxs = os.path.join(CommonMethods.get_base_path(), 'eaxs')
self.mboxes = os.path.join(CommonMethods.get_base_path(), 'eaxs')
self.emls = os.path.join(CommonMethods.get_base_path(), 'emls')
self.psts = os.path.join(CommonMethods.get_base_path(), 'pst')
if argdict['from_eml']:
self.eml_struct = True
CommonMethods.set_from_tomes(False)
if argdict['tomes_tool']:
CommonMethods.set_from_tomes(True)
#TODO: Remove this maybe
if 'max_internal' in argdict.keys():
self.max_internal = int(argdict['max_internal'])
if argdict['no_subdirectories']:
self.levels = self.NO_LEVELS
if argdict['data_dir']:
self._data_dir(argdict['data_dir'])
else:
self._data_dir()
if argdict['chunk']:
self.chunksize = argdict['chunk']
CommonMethods.set_chunk_size(1000)
CommonMethods.set_stitch(argdict['stitch'])