Delete files via the scanner/monitor. Add an ignore folder

.env.example

@@ -3,6 +3,9 @@ LIBRARY_ROOT=./video
 CACHE_ROOT=/tmp/kyoo_cache
 LIBRARY_LANGUAGES=en
 
+# A pattern (regex) to ignore video files.
+LIBRARY_IGNORE_PATTERN=.*/[dD]ownloads?/.*
+
 # The following two values should be set to a random sequence of characters.
 # You MUST change thoses when installing kyoo (for security)
 AUTHENTICATION_SECRET=4c@mraGB!KRfF@kpS8739y9FcHemKxBsqqxLbdR?

back/src/Kyoo.Core/Controllers/Repositories/EpisodeRepository.cs

@@ -44,6 +44,8 @@ namespace Kyoo.Core.Controllers
 		/// </summary>
 		private readonly IProviderRepository _providers;
 
+		private readonly IShowRepository _shows;
+
 		/// <inheritdoc />
 		// Use absolute numbers by default and fallback to season/episodes if it does not exists.
 		protected override Sort<Episode> DefaultSort => new Sort<Episode>.Conglomerate(
@@ -65,6 +67,7 @@ namespace Kyoo.Core.Controllers
 		{
 			_database = database;
 			_providers = providers;
+			_shows = shows;
 
 			// Edit episode slugs when the show's slug changes.
 			shows.OnEdited += (show) =>
@@ -201,10 +204,13 @@ namespace Kyoo.Core.Controllers
 			if (obj == null)
 				throw new ArgumentNullException(nameof(obj));
 
+			int epCount = await _database.Episodes.Where(x => x.ShowID == obj.ShowID).Take(2).CountAsync();
 			_database.Entry(obj).State = EntityState.Deleted;
 			obj.ExternalIDs.ForEach(x => _database.Entry(x).State = EntityState.Deleted);
 			await _database.SaveChangesAsync();
 			await base.Delete(obj);
+			if (epCount == 1)
+				await _shows.Delete(obj.ShowID);
 		}
 	}
 }

scanner/scanner/__init__.py

@@ -30,8 +30,9 @@ async def main():
 		logging.basicConfig(level=logging.INFO)
 	if len(sys.argv) > 1 and sys.argv[1] == "-vv":
 		logging.basicConfig(level=logging.DEBUG)
+	logging.getLogger('watchfiles').setLevel(logging.WARNING)
 
-	jsons.set_serializer(lambda x, **_: format_date(x), Optional[date | int])
+	jsons.set_serializer(lambda x, **_: format_date(x), Optional[date | int]) # type: ignore
 	async with ClientSession(
 		json_serialize=lambda *args, **kwargs: jsons.dumps(
 			*args, key_transformer=jsons.KEY_TRANSFORMER_CAMELCASE, **kwargs

scanner/scanner/monitor.py

@@ -10,10 +10,13 @@ async def monitor(path: str, scanner: Scanner):
 			try:
 				if event == Change.added:
 					await scanner.identify(file)
+				elif event == Change.deleted:
+					await scanner.delete(file);
+				elif event == Change.modified:
+					pass
 				else:
 					print(f"Change {event} occured for file {file}")
 			except ProviderError as e:
 				logging.error(str(e))
 			except Exception as e:
 				logging.exception("Unhandled error", exc_info=e)
-		print("end", flush=True)

scanner/scanner/scanner.py

@@ -2,6 +2,7 @@ import os
 import asyncio
 import logging
 import jsons
+import re
 from aiohttp import ClientSession
 from pathlib import Path
 from guessit import guessit
@@ -19,6 +20,11 @@ class Scanner:
 		self._client = client
 		self._api_key = api_key
 		self._url = os.environ.get("KYOO_URL", "http://back:5000")
+		try:
+			self._ignore_pattern = re.compile(os.environ.get("LIBRARY_IGNORE_PATTERN", ""))
+		except Exception as e:
+			self._ignore_pattern = re.compile("")
+			logging.error(f"Invalid ignore pattern. Ignoring. Error: {e}")
 		self.provider = Provider.get_all(client)[0]
 		self.cache = {"shows": {}, "seasons": {}}
 		self.languages = languages
@@ -26,13 +32,20 @@ class Scanner:
 	async def scan(self, path: str):
 		logging.info("Starting the scan. It can take some times...")
 		self.registered = await self.get_registered_paths()
-		videos = (str(p) for p in Path(path).rglob("*") if p.is_file())
+		videos = [str(p) for p in Path(path).rglob("*") if p.is_file()]
+		deleted = [x for x in self.registered if x not in videos]
+
+		if len(deleted) != len(self.registered):
+			for x in deleted:
+				await self.delete(x)
+		else:
+			logging.warning("All video files are unavailable. Check your disks.")
+
 		# We batch videos by 20 because too mutch at once kinda DDOS everything.
-		for group in batch(videos, 20):
-			logging.info("Batch finished. Starting a new one")
+		for group in batch(iter(videos), 20):
 			await asyncio.gather(*map(self.identify, group))
 
-	async def get_registered_paths(self) -> List[Path]:
+	async def get_registered_paths(self) -> List[str]:
 		# TODO: Once movies are separated from the api, a new endpoint should be created to check for paths.
 		async with self._client.get(
 			f"{self._url}/episodes",
@@ -45,7 +58,7 @@ class Scanner:
 
 	@log_errors
 	async def identify(self, path: str):
-		if path in self.registered:
+		if path in self.registered or self._ignore_pattern.match(path):
 			return
 
 		raw = guessit(path, "--episode-prefer-number")
@@ -139,3 +152,13 @@ class Scanner:
 				r.raise_for_status()
 			ret = await r.json()
 			return ret["id"]
+
+	async def delete(self, path: str):
+		logging.info("Deleting %s", path)
+		# TODO: Adapt this for movies as well when they are split
+		async with self._client.delete(
+			f"{self._url}/episodes?path={path}", headers={"X-API-Key": self._api_key}
+		) as r:
+			if not r.ok:
+				logging.error(f"Request error: {await r.text()}")
+				r.raise_for_status()