tubearchivist/tubearchivist/home/tasks.py

333 lines
10 KiB
Python
Raw Normal View History

2021-09-05 17:10:14 +00:00
"""
Functionality:
- initiate celery app
- collect tasks
- user config changes won't get applied here
because tasks are initiated at application start
2021-09-05 17:10:14 +00:00
"""
import os
2023-03-13 03:09:48 +00:00
from celery import Celery, Task, shared_task
2022-01-22 15:13:37 +00:00
from home.src.download.queue import PendingList
from home.src.download.subscriptions import (
2023-03-02 05:30:48 +00:00
SubscriptionHandler,
2023-03-01 12:12:27 +00:00
SubscriptionScanner,
)
2022-08-10 14:03:54 +00:00
from home.src.download.thumbnails import ThumbFilesystem, ThumbValidator
2022-01-22 15:13:37 +00:00
from home.src.download.yt_dlp_handler import VideoDownloader
from home.src.es.backup import ElasticBackup
from home.src.es.index_setup import ElasitIndexWrap
2022-01-22 15:13:37 +00:00
from home.src.index.channel import YoutubeChannel
2023-03-16 06:18:26 +00:00
from home.src.index.filesystem import Filesystem
from home.src.index.manual import ImportFolderScanner
from home.src.index.reindex import Reindex, ReindexManual, ReindexOutdated
2022-12-21 07:24:24 +00:00
from home.src.ta.config import AppConfig, ReleaseVersion, ScheduleBuilder
from home.src.ta.helper import clear_dl_cache
2022-01-22 15:13:37 +00:00
from home.src.ta.ta_redis import RedisArchivist, RedisQueue
2023-03-01 12:12:27 +00:00
from home.src.ta.task_manager import TaskManager
2021-09-05 17:10:14 +00:00
CONFIG = AppConfig().config
2021-09-30 11:03:23 +00:00
REDIS_HOST = os.environ.get("REDIS_HOST")
REDIS_PORT = os.environ.get("REDIS_PORT") or 6379
2021-09-05 17:10:14 +00:00
2021-12-02 08:54:29 +00:00
os.environ.setdefault("DJANGO_SETTINGS_MODULE", "config.settings")
2023-03-01 08:31:17 +00:00
app = Celery(
"tasks",
broker=f"redis://{REDIS_HOST}:{REDIS_PORT}",
backend=f"redis://{REDIS_HOST}:{REDIS_PORT}",
result_extended=True,
)
2021-10-27 11:07:35 +00:00
app.config_from_object("django.conf:settings", namespace="ta:")
2021-09-05 17:10:14 +00:00
app.autodiscover_tasks()
app.conf.timezone = os.environ.get("TZ") or "UTC"
2021-09-05 17:10:14 +00:00
2023-03-13 03:09:48 +00:00
class BaseTask(Task):
"""base class to inherit each class from"""
TASK_CONFIG = {
"update_subscribed": {
"title": "Rescan your Subscriptions",
"group": "message:download:scan",
},
"download_pending": {
"title": "Downloading",
"group": "message:download:run",
},
"extract_download": {
"title": "Add to download queue",
"group": "message:download:add",
},
"subscribe_to": {
"title": "Add Subscription",
"group": "message:download:subscribe",
},
"check_reindex": {
2023-03-15 09:59:20 +00:00
"title": "Reindex Documents",
2023-03-13 03:09:48 +00:00
"group": "message:settings:reindex",
},
"manual_import": {
"title": "Manual video import",
"group": "message:settings:import",
},
"run_backup": {
"title": "Index Backup",
"group": "message:settings:backup",
},
"restore_backup": {
"title": "Restore Backup",
"group": "message:settings:restore",
},
"rescan_filesystem": {
"title": "Rescan your Filesystem",
"group": "message:settings:filesystemscan",
},
"thumbnail_check": {
"title": "Check your Thumbnails",
"group": "message:settings:thumbnailcheck",
},
"resync_thumbs": {
"title": "Sync Thumbnails to Media Files",
"group": "message:settings:thumbnailsync",
},
"index_playlists": {
"title": "Index Channel Playlist",
"group": "message:channel:indexplaylist",
},
}
def on_failure(self, exc, task_id, args, kwargs, einfo):
"""callback for task failure"""
print(f"{task_id} Failed callback")
def on_success(self, retval, task_id, args, kwargs):
"""callback task completed successfully"""
print(f"{task_id} success callback")
message, key = self._build_message()
message.update({"messages": ["Task completed successfully"]})
2023-03-13 03:09:48 +00:00
RedisArchivist().set_message(key, message, expire=5)
def before_start(self, task_id, args, kwargs):
"""callback before initiating task"""
print(f"{self.name} create callback")
message, key = self._build_message()
message.update({"messages": ["New task received."]})
2023-03-13 03:09:48 +00:00
RedisArchivist().set_message(key, message)
2023-03-15 06:28:19 +00:00
def send_progress(self, message_lines, progress=False, title=False):
"""send progress message"""
message, key = self._build_message()
message.update(
{
"messages": message_lines,
"progress": progress,
}
)
2023-03-15 06:28:19 +00:00
if title:
message["title"] = title
RedisArchivist().set_message(key, message)
2023-03-13 03:09:48 +00:00
def _build_message(self, level="info"):
2023-03-13 03:09:48 +00:00
"""build message dict"""
task_id = self.request.id
2023-03-13 03:09:48 +00:00
config = self.TASK_CONFIG.get(self.name)
message = {
"status": config.get("group"),
"title": config.get("title"),
"level": level,
"id": task_id,
}
key = f"{config.get('group')}:{task_id.split('-')[0]}"
return message, key
@shared_task(name="update_subscribed", bind=True, base=BaseTask)
2023-03-01 12:12:27 +00:00
def update_subscribed(self):
2021-09-21 09:25:22 +00:00
"""look for missing videos and add to pending"""
2023-03-01 12:12:27 +00:00
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] rescan already running")
message = {
"status": "message:rescan",
"level": "error",
"title": "Rescanning channels and playlists.",
"message": "Rescan already in progress.",
}
RedisArchivist().set_message("message:rescan", message, expire=True)
return
manager.init(self)
2023-03-14 09:40:05 +00:00
missing_videos = SubscriptionScanner(task=self).scan()
if missing_videos:
print(missing_videos)
extrac_dl.delay(missing_videos)
2023-03-01 12:12:27 +00:00
2023-03-15 06:28:19 +00:00
@shared_task(name="download_pending", bind=True, base=BaseTask)
def download_pending(self, from_queue=True):
2021-09-21 09:25:22 +00:00
"""download latest pending videos"""
2023-03-01 12:12:27 +00:00
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] download queue already running")
return
manager.init(self)
2023-03-15 06:28:19 +00:00
downloader = VideoDownloader(task=self)
if from_queue:
downloader.add_pending()
2023-03-01 12:12:27 +00:00
downloader.run_queue()
2021-09-05 17:10:14 +00:00
2023-03-14 09:40:05 +00:00
@shared_task(name="extract_download", bind=True, base=BaseTask)
def extrac_dl(self, youtube_ids):
2021-09-21 09:25:22 +00:00
"""parse list passed and add to pending"""
2023-03-14 09:40:05 +00:00
pending_handler = PendingList(youtube_ids=youtube_ids, task=self)
2022-03-18 11:27:25 +00:00
pending_handler.parse_url_list()
pending_handler.add_to_pending()
2023-03-15 09:59:20 +00:00
@shared_task(bind=True, name="check_reindex", base=BaseTask)
def check_reindex(self, data=False, extract_videos=False):
2021-09-21 09:25:22 +00:00
"""run the reindex main command"""
if data:
# started from frontend through API
print(f"[task][{self.name}] reindex {data}")
2023-03-15 09:59:20 +00:00
self.send_progress("Add items to the reindex Queue.")
ReindexManual(extract_videos=extract_videos).extract_data(data)
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] reindex queue is already running")
return
manager.init(self)
if not data:
# started from scheduler
print(f"[task][{self.name}] reindex outdated documents")
2023-03-15 09:59:20 +00:00
self.send_progress("Add outdated documents to the reindex Queue.")
ReindexOutdated().add_outdated()
2023-03-15 09:59:20 +00:00
Reindex(task=self).reindex_all()
2021-09-13 15:17:36 +00:00
2023-03-15 10:39:37 +00:00
@shared_task(bind=True, name="manual_import", base=BaseTask)
def run_manual_import(self):
2021-09-21 09:25:22 +00:00
"""called from settings page, to go through import folder"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] manual import is already running")
return
manager.init(self)
2023-03-15 10:39:37 +00:00
ImportFolderScanner(task=self).scan()
2021-09-18 13:02:54 +00:00
2023-03-16 03:11:10 +00:00
@shared_task(bind=True, name="run_backup", base=BaseTask)
def run_backup(self, reason="auto"):
2021-09-21 09:25:22 +00:00
"""called from settings page, dump backup to zip file"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] backup is already running")
return
2022-08-12 05:03:09 +00:00
manager.init(self)
2023-03-16 03:11:10 +00:00
ElasticBackup(reason=reason, task=self).backup_all_indexes()
2023-03-16 03:11:10 +00:00
@shared_task(bind=True, name="restore_backup", base=BaseTask)
def run_restore_backup(self, filename):
2021-09-21 09:25:22 +00:00
"""called from settings page, dump backup to zip file"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] restore is already running")
return
manager.init(self)
2023-03-16 03:11:10 +00:00
self.send_progress(["Reset your Index"])
ElasitIndexWrap().reset()
2023-03-16 03:11:10 +00:00
ElasticBackup(task=self).restore(filename)
2021-09-21 09:25:22 +00:00
print("index restore finished")
2021-09-24 16:37:26 +00:00
2023-03-01 08:31:17 +00:00
@shared_task(name="kill_download")
2021-09-24 16:37:26 +00:00
def kill_dl(task_id):
"""kill download worker task by ID"""
2021-10-11 08:26:31 +00:00
if task_id:
app.control.revoke(task_id, terminate=True)
_ = RedisArchivist().del_message("dl_queue_id")
RedisQueue(queue_name="dl_queue").clear()
2021-09-24 16:37:26 +00:00
_ = clear_dl_cache(CONFIG)
2021-09-24 16:37:26 +00:00
# notify
mess_dict = {
"status": "message:download",
2021-09-24 16:37:26 +00:00
"level": "error",
"title": "Canceling download process",
"message": "Canceling download queue now.",
2021-09-24 16:37:26 +00:00
}
RedisArchivist().set_message("message:download", mess_dict, expire=True)
2023-03-16 06:18:26 +00:00
@shared_task(bind=True, name="rescan_filesystem", base=BaseTask)
def rescan_filesystem(self):
2021-10-08 08:10:44 +00:00
"""check the media folder for mismatches"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] filesystem rescan already running")
return
manager.init(self)
2023-03-16 06:18:26 +00:00
Filesystem(task=self).process()
2023-03-16 10:12:50 +00:00
ThumbValidator(task=self).validate()
2023-03-16 10:12:50 +00:00
@shared_task(bind=True, name="thumbnail_check", base=BaseTask)
def thumbnail_check(self):
2021-12-02 08:54:29 +00:00
"""validate thumbnails"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] thumbnail check is already running")
return
manager.init(self)
2023-03-16 10:12:50 +00:00
ThumbValidator(task=self).validate()
2021-12-02 08:54:29 +00:00
2023-03-16 11:13:37 +00:00
@shared_task(bind=True, name="resync_thumbs", base=BaseTask)
def re_sync_thumbs(self):
"""sync thumbnails to mediafiles"""
manager = TaskManager()
if manager.is_pending(self):
print(f"[task][{self.name}] thumb re-embed is already running")
return
manager.init(self)
2023-03-16 11:13:37 +00:00
ThumbFilesystem(task=self).embed()
2023-03-01 08:31:17 +00:00
@shared_task(name="subscribe_to")
def subscribe_to(url_str):
"""take a list of urls to subscribe to"""
2023-03-02 05:30:48 +00:00
SubscriptionHandler(url_str).subscribe()
@shared_task(bind=True, name="index_playlists", base=BaseTask)
def index_channel_playlists(self, channel_id):
"""add all playlists of channel to index"""
channel = YoutubeChannel(channel_id, task=self)
channel.index_channel_playlists()
2021-12-02 08:54:29 +00:00
2022-12-21 07:24:24 +00:00
@shared_task(name="version_check")
def version_check():
"""check for new updates"""
ReleaseVersion().check()
# start schedule here
app.conf.beat_schedule = ScheduleBuilder().build_schedule()