lastgenre: Load configured blacklist file

Uses a custom text file format since YAML, INI, TOML, ... all have their
flaws with parsing regex patterns.

Author: JOJ0

beetsplug/lastgenre/__init__.py

@@ -25,13 +25,15 @@
 import codecs
 import os
 import traceback
+from collections import defaultdict
 from typing import Union
 
 import pylast
 import yaml
 
 from beets import config, library, plugins, ui
 from beets.library import Album, Item
+from beets.ui import UserError
 from beets.util import normpath, plurality, unique_list
 
 LASTFM = pylast.LastFMNetwork(api_key=plugins.LASTFM_KEY)
@@ -105,6 +107,7 @@ def __init__(self):
                 "prefer_specific": False,
                 "title_case": True,
                 "extended_debug": False,
+                "blacklist": False,
             }
         )
         self.setup()
@@ -117,6 +120,7 @@ def setup(self):
         self._genre_cache = {}
         self.whitelist = self._load_whitelist()
         self.c14n_branches, self.canonicalize = self._load_c14n_tree()
+        self.blacklist = self._load_blacklist()
 
     def _load_whitelist(self) -> set[str]:
         whitelist = set()
@@ -151,6 +155,54 @@ def _load_c14n_tree(self) -> tuple[list[list[str]], bool]:
             flatten_tree(genres_tree, [], c14n_branches)
         return c14n_branches, canonicalize
 
+    def _load_blacklist(self) -> defaultdict[list[str]]:
+        """Load the blacklist from a configured file path.
+
+        For maximum compatibility with regex patterns, a custom format is used:
+        - Each section starts with an artist name, followed by a colon.
+        - Subsequent lines are indented and contain a regex pattern to match a genre.
+
+        Eg.:
+          artist name 1:
+            - genre pattern 1
+            - genre pattern 2
+          artist name 2:
+            - genre pattern 3
+
+        Raises:
+            UserError: if the file format is invalid.
+        """
+        blacklist = defaultdict(list)
+        if not (bl_filename := self.config["blacklist"].get()):
+            return blacklist
+        if not os.path.isfile(bl_filename := normpath(bl_filename)):
+            self._log.error("Blacklist file not found: {} .", bl_filename)
+            return blacklist
+
+        current = None
+        with open(bl_filename, "rb") as f:
+            for lineno, line in enumerate(f, 1):
+                line = line.strip()
+                if not line or line.startswith(b"#"):
+                    continue
+                if not line.startswith(b' '):
+                    if not line.endswith(b':'):
+                        raise UserError(
+                            f"Malformed blacklist section header (eg 'artist 1:') "
+                            f"at line {lineno}: {line.decode('utf-8', 'replace')}"
+                        )
+                    current = line.rstrip(b':').decode("utf-8", "replace")
+                else:
+                    if current is None:
+                        raise UserError(
+                            f"Blacklist regex pattern line before any section header "
+                            f"at line {lineno}: {line.decode('utf-8', 'replace')}"
+                        )
+                    blacklist[current].append(
+                        line.strip().decode("utf-8", "replace")
+                    )
+        return blacklist
+
     @property
     def sources(self) -> tuple[str, ...]:
         """A tuple of allowed genre sources. May contain 'track',