Initial commit

src/watfag/parsers/tvboxset/__init__.py

@@ -0,0 +1,57 @@
+import importlib
+from pathlib import Path
+from pkgutil import iter_modules
+from typing import Optional
+
+from parsers.generic import Release, ParserManager
+from parsers.generic.watfag import *
+
+
+class TVBoxSetRelease(Release):
+    """Holds info representing a release of a TV box set."""
+    def __init__(self, unparsed_text, dl_link, **kwargs):
+        super().__init__(unparsed_text, dl_link, **kwargs)
+        self.show_title: str = ""
+        self.seasons: Optional[str] = None
+        self.group: Optional[Group] = None
+        self.group_name: Optional[str] = None
+        self.quality: Optional[Resolution] = None
+        self.source: Optional[Source] = None
+        self.streaming: Optional[StreamingService] = None
+        self.video_codec: Optional[VideoCodec] = None
+        self.audio_codec: Optional[AudioCodec] = None
+        self.audio_layout: Optional[AudioLayout] = None
+        self.dynamic_range: Optional[DynamicRange] = None
+        self.repack: Optional[Repack] = None
+        self.multi: Optional[Multi] = None
+
+    def __str__(self):
+        parts = [f"{self.show_title} (Seasons: {self.seasons})"]
+        for attr in ['quality', 'video_codec', 'audio_codec', 'audio_layout', 'dynamic_range', 'repack', 'multi', 'source']:
+            value = getattr(self, attr)
+            parts.append(f"{attr.capitalize()}: {value if value else 'Unknown'}")
+        if self.streaming:
+            parts.append(f"Streaming: {self.streaming}")
+        parts.append(f"Group: {self.group_name if self.group else 'Unknown'}")
+        if not self.fully_consumed():
+            parts.append(f"Unparsed: {self.metadata_text}")
+        parts.append(f"WATFAG: {self.watfag:.2f}")
+        return " | ".join(parts)
+
+class TVBoxSetParser:
+    """
+    This class can be inherited by any parser that is specific to TV box sets.
+    It allows dynamic importing of parser classes and provides a method to run all parsers on a given TV box set release.
+    """
+
+class TVBoxSetParserManager(ParserManager):
+    """Parses TV box set releases."""
+
+    def collect_parsers(self):
+        """Dynamically imports all TV box set parsers."""
+        super().collect_parsers()
+        package_dir = Path(__file__).parent
+        for _, module_name, _ in iter_modules([package_dir]):
+            importlib.import_module(f"{__package__}.{module_name}")
+
+        self.parsers.extend(TVBoxSetParser.__subclasses__())

src/watfag/parsers/tvboxset/title_seasons.py

@@ -0,0 +1,57 @@
+import regex as re
+
+from parsers.generic.parsers import DataParser
+from parsers.tvboxset import TVBoxSetParser, TVBoxSetRelease
+
+patterns = [
+    re.compile(  # Show Name S01-S02 (year)
+        r"^(?P<title>.+?)[-_. ]S(?:eason)?s?(?P<season_start>[0-9]{1,2}) ?[-_. ] ?(?:S(?:eason)?)?"
+        r"(?P<season_end>[0-9]{1,2})[-_. ]\(?(1(8|9)|20)\d{2}(?!p|i|(1(8|9)|20)\d{2}|\]|\W(1(8|9)|20)\d{2})\)?",
+        re.IGNORECASE | re.UNICODE
+    ),
+    re.compile(  # Show Name (year) S01-S02
+        r"^(?<title>.+?)[-_. ]\(?(1(8|9)|20)\d{2}(?!p|i|(1(8|9)|20)\d{2}|\]|\W(1(8|9)|20)\d{2})\)?[-_. ]S(?:eason)?s?"
+        r"(?<season_start>[0-9]{1,2}) ?[-_. ] ?(?:S(?:eason)?)?(?<season_end>[0-9]{1,2})[-_. ]",
+        re.IGNORECASE | re.UNICODE
+    ),
+    re.compile(  # Show Name (year) S01
+        r"^(?<title>.+?)[-_. ]\(?(1(8|9)|20)\d{2}(?!p|i|(1(8|9)|20)\d{2}|\]|\W(1(8|9)|20)\d{2})\)?[-_. ]"
+        r"S(?:eason)?s? ?(?<season_start>[0-9]{1,2}) ?[-_. ]",
+        re.IGNORECASE | re.UNICODE
+    ),
+    re.compile(  # Show Name S01 (year)
+        r"^(?<title>.+?)[-_. ]S(?:eason)?s? ?(?<season_start>[0-9]{1,2}) ?[-_. ]"
+        r"\(?(1(8|9)|20)\d{2}(?!p|i|(1(8|9)|20)\d{2}|\]|\W(1(8|9)|20)\d{2})\)?[-_. ]",
+        re.IGNORECASE | re.UNICODE
+    ),
+    re.compile(  # Show Name (Complete) S01-S02
+        r"^(?<title>.+?)[-_. ](?:Complete[-_. ]?(?:Series[-_. ])?)?\(?S(?:eason)?s?(?<season_start>[0-9]{1,2})"
+        r" ?[-_. ] ?(?:S(?:eason)?)?(?<season_end>[0-9]{1,2})\)?[-_. ](?:Complete[-_. ])?",
+        re.IGNORECASE | re.UNICODE
+    ),
+    re.compile(  # Nuclear option: Show Name S01
+        r"^(?<title>.+?)[-_. ]S(?:eason)?s? ?(?<season_start>[0-9]{1,2}) ?[-_. ]",
+        re.IGNORECASE | re.UNICODE
+    )
+]
+
+class TitleSeasonsParser(DataParser, TVBoxSetParser):
+    def __init__(self, release: TVBoxSetRelease):
+        super().__init__(release)
+        self.priority = 0  # First parser to run
+
+    """Parses the title and seasons from the unparsed text."""
+    def parse(self) -> bool:
+        for pattern in patterns:
+            match = pattern.match(self.release.original_text)
+            if match:
+                self.release.show_title = match.group("title").replace(".", " ").replace("_", " ").strip() if match.group("title") else ""
+                season_start = int(match.group("season_start")) if match.group("season_start") else 0
+                season_end = int(match.group("season_end")) if "season_end" in match.groupdict() and match.group("season_end") else season_start
+                self.release.seasons = f"{season_start}" if season_start == season_end else f"{season_start}-{season_end}"
+
+                self.release.metadata_text = self.release.original_text[:match.span()[0]] + self.release.original_text[match.span()[1]:]
+
+                return True
+
+        return False