paperless-ngx/src/documents/management/commands/document_consumer.py

99 lines
3 KiB
Python
Raw Normal View History

2015-12-26 13:21:33 +00:00
import datetime
import logging
2015-12-20 19:23:33 +00:00
import os
import time
2015-12-20 19:23:33 +00:00
from django.conf import settings
from django.core.management.base import BaseCommand, CommandError
2015-12-20 19:23:33 +00:00
2018-05-11 14:01:12 +02:00
from ...consumer import Consumer, ConsumerError, make_dirs
from ...mail import MailFetcher, MailFetcherError
2016-01-23 02:33:29 +00:00
class Command(BaseCommand):
2015-12-20 19:23:33 +00:00
"""
On every iteration of an infinite loop, consume what we can from the
consumption directory, and fetch any mail available.
2015-12-20 19:23:33 +00:00
"""
ORIGINAL_DOCS = os.path.join(settings.MEDIA_ROOT, "documents", "originals")
THUMB_DOCS = os.path.join(settings.MEDIA_ROOT, "documents", "thumbnails")
2015-12-20 19:23:33 +00:00
def __init__(self, *args, **kwargs):
2016-01-21 12:50:22 -05:00
2015-12-20 19:23:33 +00:00
self.verbosity = 0
self.file_consumer = None
self.mail_fetcher = None
self.first_iteration = True
2016-01-21 12:50:22 -05:00
2015-12-20 19:23:33 +00:00
BaseCommand.__init__(self, *args, **kwargs)
def add_arguments(self, parser):
2018-02-25 19:20:51 +01:00
parser.add_argument(
"directory",
default=settings.CONSUMPTION_DIR,
2018-02-26 18:52:46 +01:00
nargs="?",
help="The consumption directory."
2018-02-25 19:20:51 +01:00
)
parser.add_argument(
"--loop-time",
default=settings.CONSUMER_LOOP_TIME,
2018-02-26 18:52:46 +01:00
type=int,
help="Wait time between each loop (in seconds)."
)
parser.add_argument(
"--mail-delta",
default=10,
type=int,
help="Wait time between each mail fetch (in minutes)."
)
parser.add_argument(
"--oneshot",
action="store_true",
help="Run only once."
2018-02-25 19:20:51 +01:00
)
2015-12-20 19:23:33 +00:00
def handle(self, *args, **options):
self.verbosity = options["verbosity"]
2018-02-25 19:20:51 +01:00
directory = options["directory"]
loop_time = options["loop_time"]
mail_delta = datetime.timedelta(minutes=options["mail_delta"])
2015-12-20 19:23:33 +00:00
try:
self.file_consumer = Consumer(consume=directory)
self.mail_fetcher = MailFetcher(consume=directory)
except (ConsumerError, MailFetcherError) as e:
raise CommandError(e)
2018-05-11 14:01:12 +02:00
make_dirs(self.ORIGINAL_DOCS, self.THUMB_DOCS)
2015-12-20 19:23:33 +00:00
logging.getLogger(__name__).info(
"Starting document consumer at {}".format(directory)
)
2018-02-25 19:20:51 +01:00
if options["oneshot"]:
self.loop(mail_delta=mail_delta)
else:
try:
while True:
self.loop(mail_delta=mail_delta)
time.sleep(loop_time)
if self.verbosity > 1:
print(".", int(time.time()))
except KeyboardInterrupt:
print("Exiting")
def loop(self, mail_delta):
2016-01-01 16:13:59 +00:00
# Occasionally fetch mail and store it to be consumed on the next loop
# We fetch email when we first start up so that it is not necessary to
# wait for 10 minutes after making changes to the config file.
delta = self.mail_fetcher.last_checked + mail_delta
if self.first_iteration or delta < datetime.datetime.now():
self.first_iteration = False
self.mail_fetcher.pull()
self.file_consumer.consume_new_files()