Merge pull request #1156 from djl/lastgenreunicode

lastgenre: Optionally replace unicode characters when performing lookups

beetsplug/lastgenre/__init__.py

@@ -41,6 +41,10 @@ PYLAST_EXCEPTIONS = (
     pylast.NetworkError,
 )
 
+REPLACE = {
+    u'\u2010': '-',
+}
+
 
 def deduplicate(seq):
     """Remove duplicates from sequence wile preserving order.
@@ -238,7 +242,10 @@ class LastGenrePlugin(plugins.BeetsPlugin):
     def _cached_lookup(self, entity, method, *args):
         """Get a genre based on the named entity using the callable `method`
         whose arguments are given in the sequence `args`. The genre lookup
-        is cached based on the entity name and the arguments.
+        is cached based on the entity name and the arguments. Before the
+        lookup, each argument is has some Unicode characters replaced with
+        rough ASCII equivalents in order to return better results from the
+        Last.fm database.
         """
         # Shortcut if we're missing metadata.
         if any(not s for s in args):
@@ -248,7 +255,13 @@ class LastGenrePlugin(plugins.BeetsPlugin):
         if key in self._genre_cache:
             return self._genre_cache[key]
         else:
-            genre = self.fetch_genre(method(*args))
+            args_replaced = []
+            for arg in args:
+                for k, v in REPLACE.items():
+                    arg = arg.replace(k, v)
+                args_replaced.append(arg)
+
+            genre = self.fetch_genre(method(*args_replaced))
             self._genre_cache[key] = genre
             return genre