epicyon/schedule.py

244 lines
10 KiB
Python
Raw Normal View History

2020-04-04 10:52:30 +00:00
__filename__ = "schedule.py"
__author__ = "Bob Mottram"
__license__ = "AGPL3+"
2023-01-21 23:03:30 +00:00
__version__ = "1.4.0"
2020-04-04 10:52:30 +00:00
__maintainer__ = "Bob Mottram"
2021-09-10 16:14:50 +00:00
__email__ = "bob@libreserver.org"
2020-04-04 10:52:30 +00:00
__status__ = "Production"
2021-06-15 15:08:12 +00:00
__module_group__ = "Calendar"
2020-01-12 20:13:44 +00:00
import os
2020-01-12 20:16:33 +00:00
import time
2020-01-12 20:13:44 +00:00
import datetime
2022-12-18 15:29:54 +00:00
from utils import acct_handle_dir
2021-12-26 10:57:03 +00:00
from utils import has_object_dict
2021-12-27 17:42:35 +00:00
from utils import get_status_number
2021-12-26 15:13:34 +00:00
from utils import load_json
2021-12-26 18:46:43 +00:00
from utils import is_account_dir
2021-12-26 12:02:29 +00:00
from utils import acct_dir
2022-06-21 11:58:50 +00:00
from utils import remove_eol
2021-12-29 21:55:09 +00:00
from outbox import post_message_to_outbox
from session import create_session
2022-07-28 09:59:18 +00:00
from threads import begin_thread
2020-01-12 20:13:44 +00:00
2020-04-04 10:52:30 +00:00
2021-12-29 21:55:09 +00:00
def _update_post_schedule(base_dir: str, handle: str, httpd,
2021-12-31 10:29:28 +00:00
max_scheduled_posts: int) -> None:
2022-09-21 20:17:21 +00:00
"""Checks if posts are due to be delivered and if so moves them to
the outbox
2020-01-12 20:13:44 +00:00
"""
2021-12-31 10:29:28 +00:00
schedule_index_filename = \
2022-12-18 15:29:54 +00:00
acct_handle_dir(base_dir, handle) + '/schedule.index'
2021-12-31 10:29:28 +00:00
if not os.path.isfile(schedule_index_filename):
2020-01-12 20:13:44 +00:00
return
# get the current time as an int
2021-12-26 13:17:46 +00:00
curr_time = datetime.datetime.utcnow()
2021-12-31 10:29:28 +00:00
days_since_epoch = (curr_time - datetime.datetime(1970, 1, 1)).days
2020-01-12 20:13:44 +00:00
2022-12-18 15:29:54 +00:00
schedule_dir = acct_handle_dir(base_dir, handle) + '/scheduled/'
2021-12-31 10:29:28 +00:00
index_lines = []
delete_schedule_post = False
2020-04-04 10:52:30 +00:00
nickname = handle.split('@')[0]
2022-06-09 14:46:30 +00:00
with open(schedule_index_filename, 'r', encoding='utf-8') as sched_file:
2021-12-31 10:29:28 +00:00
for line in sched_file:
2020-01-12 20:13:44 +00:00
if ' ' not in line:
continue
2021-12-31 10:29:28 +00:00
date_str = line.split(' ')[0]
if 'T' not in date_str:
2020-01-12 20:13:44 +00:00
continue
2022-06-21 11:58:50 +00:00
post_id1 = line.split(' ', 1)[1]
post_id = remove_eol(post_id1)
2021-12-31 10:29:28 +00:00
post_filename = schedule_dir + post_id + '.json'
if delete_schedule_post:
2020-01-12 20:13:44 +00:00
# delete extraneous scheduled posts
2021-12-26 23:41:34 +00:00
if os.path.isfile(post_filename):
try:
2021-12-26 23:41:34 +00:00
os.remove(post_filename)
2021-11-25 18:42:38 +00:00
except OSError:
2021-12-29 21:55:09 +00:00
print('EX: _update_post_schedule unable to delete ' +
2021-12-26 23:41:34 +00:00
str(post_filename))
2020-01-12 20:13:44 +00:00
continue
# create the new index file
2021-12-31 10:29:28 +00:00
index_lines.append(line)
2020-01-12 20:13:44 +00:00
# convert string date to int
2021-12-31 10:29:28 +00:00
post_time = \
datetime.datetime.strptime(date_str, "%Y-%m-%dT%H:%M:%S%z")
post_time = post_time.replace(tzinfo=None)
post_days_since_epoch = \
(post_time - datetime.datetime(1970, 1, 1)).days
if days_since_epoch < post_days_since_epoch:
2020-01-12 20:13:44 +00:00
continue
2021-12-31 10:29:28 +00:00
if days_since_epoch == post_days_since_epoch:
if curr_time.time().hour < post_time.time().hour:
2020-01-13 11:16:35 +00:00
continue
2021-12-31 10:29:28 +00:00
if curr_time.time().minute < post_time.time().minute:
2020-01-13 11:16:35 +00:00
continue
2021-12-26 23:41:34 +00:00
if not os.path.isfile(post_filename):
print('WARN: schedule missing post_filename=' + post_filename)
2021-12-31 10:29:28 +00:00
index_lines.remove(line)
2020-01-12 20:13:44 +00:00
continue
# load post
2021-12-26 23:41:34 +00:00
post_json_object = load_json(post_filename)
2021-12-25 22:09:19 +00:00
if not post_json_object:
2020-01-13 11:13:30 +00:00
print('WARN: schedule json not loaded')
2021-12-31 10:29:28 +00:00
index_lines.remove(line)
2020-01-12 20:13:44 +00:00
continue
2020-01-13 11:36:42 +00:00
# set the published time
# If this is not recent then http checks on the receiving side
# will reject it
2021-12-31 10:29:28 +00:00
_, published = get_status_number()
2021-12-25 22:09:19 +00:00
if post_json_object.get('published'):
post_json_object['published'] = published
2021-12-26 10:57:03 +00:00
if has_object_dict(post_json_object):
2021-12-25 22:09:19 +00:00
if post_json_object['object'].get('published'):
post_json_object['published'] = published
2020-01-13 11:36:42 +00:00
2021-12-26 19:47:06 +00:00
print('Sending scheduled post ' + post_id)
2020-01-12 20:13:44 +00:00
2020-01-13 12:27:27 +00:00
if nickname:
2021-12-31 23:07:23 +00:00
httpd.post_to_nickname = nickname
# create session if needed
curr_session = httpd.session
curr_proxy_type = httpd.proxy_type
if not curr_session:
curr_session = create_session(httpd.proxy_type)
httpd.session = curr_session
if not curr_session:
continue
if not post_message_to_outbox(curr_session,
2021-12-29 21:55:09 +00:00
httpd.translate,
post_json_object, nickname,
httpd, base_dir,
httpd.http_prefix,
httpd.domain,
httpd.domain_full,
httpd.onion_domain,
httpd.i2p_domain,
httpd.port,
httpd.recent_posts_cache,
httpd.followers_threads,
httpd.federation_list,
httpd.send_threads,
httpd.postLog,
httpd.cached_webfingers,
httpd.person_cache,
httpd.allow_deletion,
curr_proxy_type,
2021-12-29 21:55:09 +00:00
httpd.project_version,
httpd.debug,
httpd.yt_replace_domain,
httpd.twitter_replacement_domain,
httpd.show_published_date_only,
httpd.allow_local_network_access,
httpd.city, httpd.system_language,
httpd.shared_items_federated_domains,
2022-01-01 15:11:42 +00:00
httpd.shared_item_federation_tokens,
2021-12-29 21:55:09 +00:00
httpd.low_bandwidth,
httpd.signing_priv_key_pem,
httpd.peertube_instances,
httpd.theme_name,
httpd.max_like_count,
httpd.max_recent_posts,
httpd.cw_lists,
httpd.lists_enabled,
2022-07-05 14:40:26 +00:00
httpd.content_license_url,
httpd.dogwhistles,
2023-01-13 15:04:48 +00:00
httpd.min_images_for_accounts,
httpd.buy_sites):
2021-12-31 10:29:28 +00:00
index_lines.remove(line)
try:
2021-12-26 23:41:34 +00:00
os.remove(post_filename)
2021-11-25 18:42:38 +00:00
except OSError:
2021-12-29 21:55:09 +00:00
print('EX: _update_post_schedule unable to delete ' +
2021-12-26 23:41:34 +00:00
str(post_filename))
2020-01-12 20:13:44 +00:00
continue
# move to the outbox
2021-12-31 10:29:28 +00:00
outbox_post_filename = \
2021-12-26 23:41:34 +00:00
post_filename.replace('/scheduled/', '/outbox/')
2021-12-31 10:29:28 +00:00
os.rename(post_filename, outbox_post_filename)
2020-01-12 20:13:44 +00:00
2021-12-26 19:47:06 +00:00
print('Scheduled post sent ' + post_id)
2020-01-12 20:13:44 +00:00
2021-12-31 10:29:28 +00:00
index_lines.remove(line)
if len(index_lines) > max_scheduled_posts:
delete_schedule_post = True
2020-01-12 20:13:44 +00:00
# write the new schedule index file
2021-12-31 10:29:28 +00:00
schedule_index_file = \
2022-12-18 15:29:54 +00:00
acct_handle_dir(base_dir, handle) + '/schedule.index'
2022-06-09 14:46:30 +00:00
with open(schedule_index_file, 'w+', encoding='utf-8') as schedule_file:
2021-12-31 10:29:28 +00:00
for line in index_lines:
schedule_file.write(line)
2020-01-12 20:13:44 +00:00
2020-04-04 10:52:30 +00:00
2021-12-31 10:29:28 +00:00
def run_post_schedule(base_dir: str, httpd, max_scheduled_posts: int):
2020-01-12 20:13:44 +00:00
"""Dispatches scheduled posts
"""
while True:
time.sleep(60)
# for each account
2021-12-31 10:29:28 +00:00
for _, dirs, _ in os.walk(base_dir + '/accounts'):
2020-01-12 20:13:44 +00:00
for account in dirs:
if '@' not in account:
continue
2021-12-26 18:46:43 +00:00
if not is_account_dir(account):
2021-01-09 21:47:52 +00:00
continue
2020-01-12 20:13:44 +00:00
# scheduled posts index for this account
2021-12-31 10:29:28 +00:00
schedule_index_filename = \
2021-12-25 16:17:53 +00:00
base_dir + '/accounts/' + account + '/schedule.index'
2021-12-31 10:29:28 +00:00
if not os.path.isfile(schedule_index_filename):
2020-01-12 20:13:44 +00:00
continue
2021-12-29 21:55:09 +00:00
_update_post_schedule(base_dir, account,
2021-12-31 10:29:28 +00:00
httpd, max_scheduled_posts)
2021-06-05 12:57:24 +00:00
break
2020-01-12 20:13:44 +00:00
2020-04-04 10:52:30 +00:00
2021-12-29 21:55:09 +00:00
def run_post_schedule_watchdog(project_version: str, httpd) -> None:
2020-01-12 20:13:44 +00:00
"""This tries to keep the scheduled post thread running even if it dies
"""
2022-06-14 10:51:40 +00:00
print('THREAD: Starting scheduled post watchdog ' + project_version)
2021-12-31 10:29:28 +00:00
post_schedule_original = \
2021-12-29 21:55:09 +00:00
httpd.thrPostSchedule.clone(run_post_schedule)
2022-07-28 09:59:18 +00:00
begin_thread(httpd.thrPostSchedule, 'run_post_schedule_watchdog')
2020-01-12 20:13:44 +00:00
while True:
2020-03-22 21:16:02 +00:00
time.sleep(20)
2021-06-05 12:43:57 +00:00
if httpd.thrPostSchedule.is_alive():
continue
httpd.thrPostSchedule.kill()
2022-03-13 11:01:07 +00:00
print('THREAD: restarting scheduled post watchdog')
2021-06-05 12:43:57 +00:00
httpd.thrPostSchedule = \
2021-12-31 10:29:28 +00:00
post_schedule_original.clone(run_post_schedule)
2022-07-28 09:59:18 +00:00
begin_thread(httpd.thrPostSchedule, 'run_post_schedule_watchdog')
2021-06-05 12:43:57 +00:00
print('Restarting scheduled posts...')
2020-01-14 10:23:17 +00:00
2020-04-04 10:52:30 +00:00
2021-12-29 21:55:09 +00:00
def remove_scheduled_posts(base_dir: str, nickname: str, domain: str) -> None:
2020-01-14 10:23:17 +00:00
"""Removes any scheduled posts
"""
# remove the index
2021-12-31 10:29:28 +00:00
schedule_index_filename = \
2021-12-26 12:02:29 +00:00
acct_dir(base_dir, nickname, domain) + '/schedule.index'
2021-12-31 10:29:28 +00:00
if os.path.isfile(schedule_index_filename):
try:
2021-12-31 10:29:28 +00:00
os.remove(schedule_index_filename)
2021-11-25 18:42:38 +00:00
except OSError:
2021-12-29 21:55:09 +00:00
print('EX: remove_scheduled_posts unable to delete ' +
2021-12-31 10:29:28 +00:00
schedule_index_filename)
2020-01-14 10:23:17 +00:00
# remove the scheduled posts
2021-12-31 10:29:28 +00:00
scheduled_dir = acct_dir(base_dir, nickname, domain) + '/scheduled'
if not os.path.isdir(scheduled_dir):
2020-01-14 10:23:17 +00:00
return
2021-12-31 10:29:28 +00:00
for scheduled_post_filename in os.listdir(scheduled_dir):
file_path = os.path.join(scheduled_dir, scheduled_post_filename)
if os.path.isfile(file_path):
2021-10-29 18:48:15 +00:00
try:
2021-12-31 10:29:28 +00:00
os.remove(file_path)
2021-11-25 18:42:38 +00:00
except OSError:
2021-12-29 21:55:09 +00:00
print('EX: remove_scheduled_posts unable to delete ' +
2021-12-31 10:29:28 +00:00
file_path)