lastgenre: Move file loading to helpers (#5979)

I split out this refactor from #5744 to make the blacklist PR just a
little bit easier to read.

- Separate _load methods for whitelist and canonicalization tree file
loading.
- Add return type hints for those methods.
- Use `pathlib.Path()` instead of `normpath()`
- Additionally respect None value to fall back to built-in settings for
`canonical` and `whitelist` files (was only `""` before) - this code
exists for backwards compatibility (see
233f71a457),
and might be better sanity checked and vanish in the future... (a None
value is set eg. `canoncial:`) -> #5994

beetsplug/lastgenre/__init__.py

@@ -22,9 +22,9 @@ The scraper script used is available here:
 https://gist.github.com/1241307
 """
 
-import codecs
 import os
 import traceback
+from pathlib import Path
 from typing import Union
 
 import pylast
@@ -32,7 +32,7 @@ import yaml
 
 from beets import config, library, plugins, ui
 from beets.library import Album, Item
-from beets.util import normpath, plurality, unique_list
+from beets.util import plurality, unique_list
 
 LASTFM = pylast.LastFMNetwork(api_key=plugins.LASTFM_KEY)
 
@@ -111,39 +111,49 @@ class LastGenrePlugin(plugins.BeetsPlugin):
             self.import_stages = [self.imported]
 
         self._genre_cache = {}
+        self.whitelist = self._load_whitelist()
+        self.c14n_branches, self.canonicalize = self._load_c14n_tree()
 
-        # Read the whitelist file if enabled.
-        self.whitelist = set()
+    def _load_whitelist(self) -> set[str]:
+        """Load the whitelist from a text file.
+
+        Default whitelist is used if config is True, empty string or set to "nothing".
+        """
+        whitelist = set()
         wl_filename = self.config["whitelist"].get()
-        if wl_filename in (True, ""):  # Indicates the default whitelist.
+        if wl_filename in (True, "", None):  # Indicates the default whitelist.
             wl_filename = WHITELIST
         if wl_filename:
-            wl_filename = normpath(wl_filename)
-            with open(wl_filename, "rb") as f:
-                for line in f:
-                    line = line.decode("utf-8").strip().lower()
-                    if line and not line.startswith("#"):
-                        self.whitelist.add(line)
+            self._log.debug("Loading whitelist {}", wl_filename)
+            text = Path(wl_filename).expanduser().read_text(encoding="utf-8")
+            for line in text.splitlines():
+                if (line := line.strip().lower()) and not line.startswith("#"):
+                    whitelist.add(line)
 
-        # Read the genres tree for canonicalization if enabled.
-        self.c14n_branches = []
+        return whitelist
+
+    def _load_c14n_tree(self) -> tuple[list[list[str]], bool]:
+        """Load the canonicalization tree from a YAML file.
+
+        Default tree is used if config is True, empty string, set to "nothing"
+        or if prefer_specific is enabled.
+        """
+        c14n_branches: list[list[str]] = []
         c14n_filename = self.config["canonical"].get()
-        self.canonicalize = c14n_filename is not False
-
+        canonicalize = c14n_filename is not False
         # Default tree
-        if c14n_filename in (True, ""):
-            c14n_filename = C14N_TREE
-        elif not self.canonicalize and self.config["prefer_specific"].get():
+        if c14n_filename in (True, "", None) or (
             # prefer_specific requires a tree, load default tree
+            not canonicalize and self.config["prefer_specific"].get()
+        ):
             c14n_filename = C14N_TREE
-
         # Read the tree
         if c14n_filename:
             self._log.debug("Loading canonicalization tree {}", c14n_filename)
-            c14n_filename = normpath(c14n_filename)
-            with codecs.open(c14n_filename, "r", encoding="utf-8") as f:
+            with Path(c14n_filename).expanduser().open(encoding="utf-8") as f:
                 genres_tree = yaml.safe_load(f)
-            flatten_tree(genres_tree, [], self.c14n_branches)
+            flatten_tree(genres_tree, [], c14n_branches)
+        return c14n_branches, canonicalize
 
     @property
     def sources(self) -> tuple[str, ...]:

docs/changelog.rst

@@ -131,6 +131,8 @@ Other changes:
   beets/library directory.
 - Added a test to check that all plugins can be imported without errors.
 - :doc:`/guides/main`: Add instructions to install beets on Void Linux.
+- :doc:`plugins/lastgenre`: Refactor loading whitelist and canonicalization
+  file. :bug:`5979`
 
 2.3.1 (May 14, 2025)
 --------------------