Add custom guessit rules

scanner/scanner/parser/guess.py

@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+
+from guessit.api import default_api
+from typing import cast
+from rebulk import Rebulk
+
+import rules
+
+default_api.configure({})
+rblk = cast(Rebulk, default_api.rebulk)
+rblk.rules(rules)
+
+
+def guessit(name: str):
+	return default_api.guessit(
+		name, {"episode_prefer_number": True, "excludes": "language"}
+	)
+
+
+# Only used to test localy
+if __name__ == "__main__":
+	import sys
+	import json
+	from guessit.jsonutils import GuessitEncoder
+
+	ret = guessit(sys.argv[1])
+	print(json.dumps(ret, cls=GuessitEncoder, indent=4))

scanner/scanner/parser/rules.py

@@ -0,0 +1,63 @@
+# Read that for examples/rules: https://github.com/pymedusa/Medusa/blob/master/medusa/name_parser/rules/rules.py
+
+from typing import Any
+from rebulk import Rule, RemoveMatch, AppendMatch, POST_PROCESS
+from rebulk.match import Matches
+
+
+class MultipleSeasonRule(Rule):
+	"""Understand `abcd Season 2 - 5.mkv` as S2E5
+
+	Example: '[Erai-raws] Spy x Family Season 2 - 08 [1080p][Multiple Subtitle][00C44E2F].mkv'
+
+	Default:
+	```json
+	{
+	        "title": "Spy x Family",
+	        "season": [
+	                2,
+	                3,
+	                4,
+	                5,
+	                6,
+	                7,
+	                8
+	        ],
+	        "screen_size": "1080p",
+	        "release_group": "Multiple Subtitle",
+	        "crc32": "00C44E2F",
+	        "container": "mkv",
+	        "mimetype": "video/x-matroska",
+	        "type": "episode"
+	}
+	```
+
+	Expected:
+	```json
+	{
+	        "title": "Spy x Family Season 2",
+	        "season": null,
+	        "episode": 8,
+	        "screen_size": "1080p",
+	        "release_group": "Multiple Subtitle",
+	        "crc32": "00C44E2F",
+	        "container": "mkv",
+	        "mimetype": "video/x-matroska",
+	        "type": "episode"
+	}
+	```
+
+	We want `Season 2 ` to be parsed as part of the title since this format is
+	often used for animes (where season often does not match, we use thexem for that)
+	"""
+
+	priority = POST_PROCESS
+	consequence = [RemoveMatch, AppendMatch]
+
+	def when(self, matches: Matches, context) -> Any:
+		seasons = matches.named("season")
+		print(seasons)
+		print(seasons[0])
+		print(vars(seasons[0]))
+		print(seasons[0].initiator)
+		return

scanner/scanner/scanner.py

@@ -6,13 +6,13 @@ import jsons
 import re
 from aiohttp import ClientSession
 from pathlib import Path
-from guessit import guessit
 from typing import List, Literal, Any
 from providers.provider import Provider
 from providers.types.collection import Collection
 from providers.types.show import Show
 from providers.types.episode import Episode, PartialShow
 from providers.types.season import Season
+from .parser.guess import guessit
 from .utils import batch, log_errors
 from .cache import cache, exec_as_cache, make_key
 
@@ -80,7 +80,7 @@ class Scanner:
 		if path in self.registered or self._ignore_pattern.match(path):
 			return
 
-		raw = guessit(path, {"episode_prefer_number": True, "excludes": "language"})
+		raw = guessit(path)
 
 		if "mimetype" not in raw or not raw["mimetype"].startswith("video"):
 			return