Finish xem fixup rule

scanner/providers/implementations/thexem.py

@@ -124,3 +124,21 @@ class TheXem:
 
 		# Only tvdb has a proper absolute handling so we always use this one.
 		return (ep[provider]["season"], ep[provider]["episode"], ep["tvdb"]["absolute"])
+
+	@cache(ttl=timedelta(days=1))
+	async def get_expected_titles(
+		self, provider: Literal["tvdb"] | Literal["anidb"] = "tvdb"
+	) -> list[str]:
+		map = await self.get_map(provider)
+		titles = []
+
+		def clean(s: str):
+			return s.lower().replace(" ", "")
+
+		for x in map.values():
+			# Only the first element is a string (the show name) so we need to ignore the type hint
+			master_show_name: str = x[0]  # type: ignore
+			titles.append(clean(master_show_name))
+			for y in x[1:]:
+				titles.extend(clean(name) for name in y.keys())
+		return titles

scanner/providers/provider.py

@@ -3,6 +3,7 @@ from aiohttp import ClientSession
 from abc import abstractmethod, abstractproperty
 from typing import Optional, TypeVar
 
+from providers.implementations.thexem import TheXem
 from providers.utils import ProviderError
 
 from .types.show import Show
@@ -19,15 +20,12 @@ class Provider:
 	@classmethod
 	def get_all(
 		cls: type[Self], client: ClientSession, languages: list[str]
-	) -> list[Self]:
+	) -> tuple[list[Self], TheXem]:
 		providers = []
 
 		from providers.idmapper import IdMapper
 
 		idmapper = IdMapper()
-
-		from providers.implementations.thexem import TheXem
-
 		xem = TheXem(client)
 
 		from providers.implementations.themoviedatabase import TheMovieDatabase
@@ -46,7 +44,7 @@ class Provider:
 
 		idmapper.init(tmdb=tmdb, language=languages[0])
 
-		return providers
+		return providers, xem
 
 	@abstractproperty
 	def name(self) -> str:

scanner/scanner/__init__.py

@@ -1,8 +1,3 @@
-from providers.utils import ProviderError
-from .scanner import Scanner
-from .monitor import monitor
-
-
 async def main():
 	import os
 	import logging
@@ -11,7 +6,9 @@ async def main():
 	from datetime import date
 	from typing import Optional
 	from aiohttp import ClientSession
-	from providers.utils import format_date
+	from providers.utils import format_date, ProviderError
+	from .scanner import Scanner
+	from .monitor import monitor
 
 	path = os.environ.get("SCANNER_LIBRARY_ROOT", "/video")
 	languages = os.environ.get("LIBRARY_LANGUAGES")

scanner/scanner/cache.py

@@ -1,7 +1,7 @@
 import asyncio
 from datetime import datetime, timedelta
 from functools import wraps
-from typing import Any, Optional, Tuple, Final, Literal
+from typing import Any, Optional, Tuple, Final, Literal, TypeVar
 from enum import Enum
 
 
@@ -14,6 +14,7 @@ none: Final = Sentinel.NoneSentinel
 type Cache = dict[
 	Any, Tuple[asyncio.Event | Literal[none], datetime | Literal[none], Any]
 ]
+# Cache = TypeVar("Cache")  # type: ignore
 
 
 def cache(ttl: timedelta, cache: Optional[Cache] = None, typed=False):

scanner/scanner/parser/guess.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 
 from guessit.api import default_api
-from typing import cast
+from typing import cast, List
 from rebulk import Rebulk
 
 try:
@@ -14,9 +14,14 @@ rblk = cast(Rebulk, default_api.rebulk)
 rblk.rules(rules)
 
 
-def guessit(name: str):
+def guessit(name: str, *, xem_titles: List[str] = []):
 	return default_api.guessit(
-		name, {"episode_prefer_number": True, "excludes": "language"}
+		name,
+		{
+			"episode_prefer_number": True,
+			"excludes": "language",
+			"xem_titles": xem_titles,
+		},
 	)
 
 
@@ -24,7 +29,19 @@ def guessit(name: str):
 if __name__ == "__main__":
 	import sys
 	import json
+	from pathlib import Path
 	from guessit.jsonutils import GuessitEncoder
+	from aiohttp import ClientSession
+	import asyncio
 
-	ret = guessit(sys.argv[1])
-	print(json.dumps(ret, cls=GuessitEncoder, indent=4))
+	sys.path.append(str(Path(f"{__file__}/../../..").resolve()))
+	from providers.implementations.thexem import TheXem
+
+	async def main():
+		async with ClientSession() as client:
+			xem = TheXem(client)
+
+			ret = guessit(sys.argv[1], xem_titles=await xem.get_expected_titles())
+			print(json.dumps(ret, cls=GuessitEncoder, indent=4))
+
+	asyncio.run(main())

scanner/scanner/parser/rules.py

@@ -234,7 +234,7 @@ class XemFixup(Rule):
 	def when(self, matches: Matches, context) -> Any:
 		titles: List[Match] = matches.named("title", lambda m: m.tagged("title"))  # type: ignore
 
-		if not titles:
+		if not titles or not context["xem_titles"]:
 			return
 		title = titles[0]
 
@@ -249,6 +249,5 @@ class XemFixup(Rule):
 		new_title.end = nmatch[0].end
 		new_title.value = f"{title.value}{hole}{nmatch[0].value}"
 
-		# TODO: check if new_title exists on thexem, if not early return
-
-		return [[title, nmatch[0]], [new_title]]
+		if new_title.value.lower().replace(" ", "") in context["xem_titles"]:
+			return [[title, nmatch[0]], [new_title]]

scanner/scanner/scanner.py

@@ -31,7 +31,7 @@ class Scanner:
 		except Exception as e:
 			self._ignore_pattern = re.compile("")
 			logging.error(f"Invalid ignore pattern. Ignoring. Error: {e}")
-		self.provider = Provider.get_all(client, languages)[0]
+		[self.provider, *_], self._xem = Provider.get_all(client, languages)
 		self.languages = languages
 
 		self._collection_cache = {}
@@ -80,7 +80,7 @@ class Scanner:
 		if path in self.registered or self._ignore_pattern.match(path):
 			return
 
-		raw = guessit(path)
+		raw = guessit(path, xem_titles=await self._xem.get_expected_titles())
 
 		if "mimetype" not in raw or not raw["mimetype"].startswith("video"):
 			return
@@ -90,7 +90,7 @@ class Scanner:
 
 		if isinstance(raw.get("season"), List):
 			raise ProviderError(
-				f"An episode can't have multiple seasons (found {raw.get("season")} for {path})"
+				f"An episode can't have multiple seasons (found {raw.get('season')} for {path})"
 			)
 		if isinstance(raw.get("episode"), List):
 			raise ProviderError(