From 6edc6841bf1dce98c8c6c37e48a5749530529701 Mon Sep 17 00:00:00 2001
From: Colin <colin@uninsane.org>
Date: Wed, 29 Nov 2023 12:52:35 +0000
Subject: [PATCH] sane-tag-music: support directory/tree operations

---
 .../sane-scripts/src/sane-tag-music           | 60 +++++++++++++++----
 1 file changed, 48 insertions(+), 12 deletions(-)

diff --git a/pkgs/additional/sane-scripts/src/sane-tag-music b/pkgs/additional/sane-scripts/src/sane-tag-music
index e0a384cae..b07dedeb5 100755
--- a/pkgs/additional/sane-scripts/src/sane-tag-music
+++ b/pkgs/additional/sane-scripts/src/sane-tag-music
@@ -1,5 +1,8 @@
 #!/usr/bin/env nix-shell
 #!nix-shell -i python3 -p "python3.withPackages (ps: [ ps.mutagen ])"
+#
+# mutagen docs:
+# - <https://mutagen.readthedocs.io/en/latest/>
 
 from dataclasses import dataclass
 
@@ -41,16 +44,28 @@ class Tags:
         return f"artist:{self.artist}/{self.albumartist}, album:{self.album}, title:{self.title}, trackno:{self.tracknumber}"
 
     def union(self, fallback: 'Tags') -> 'Tags':
-        def merge_field(first: list[str], second: list[str]) -> list[str]:
-            first_lower = [i.lower() for i in first]
-            return first + [i for i in second if i.lower() not in first_lower]
+        def merge_field(primary: list[str], secondary: list[str]) -> list[str]:
+            # primary_lower = [i.lower() for i in primary]
+            # return primary + [i for i in secondary if i.lower() not in primary_lower]
+            return primary or secondary
+
+
+        artist=merge_field(self.artist, fallback.artist)
+        album=merge_field(self.album, fallback.album)
+        title=merge_field(self.title, fallback.title)
+        albumartist=merge_field(self.albumartist, fallback.albumartist)
+        tracknumber=merge_field(self.tracknumber, fallback.tracknumber)
+
+        if artist == albumartist:
+            # if extraneous, then keep the album artist whatever it originally was
+            albumartist = self.albumartist
 
         return Tags(
-            artist=merge_field(self.artist, fallback.artist),
-            album=merge_field(self.album, fallback.album),
-            title=merge_field(self.title, fallback.title),
-            albumartist=merge_field(self.albumartist, fallback.albumartist),
-            tracknumber=merge_field(self.tracknumber, fallback.tracknumber),
+            artist=artist,
+            album=album,
+            title=title,
+            albumartist=albumartist,
+            tracknumber=tracknumber,
         )
 
     def promote_albumartist(self) -> None:
@@ -75,7 +90,7 @@ class Tags:
         comps = p.split('/')
         tags = Tags()
         if len(comps) == 3:
-            tags.albumartist = comps[0]
+            tags.albumartist = [comps[0]]
             album_part = comps[1].split('-')
             if len(album_part) == 2:
                 # artist/artist-album/track
@@ -85,7 +100,7 @@ class Tags:
                 tags.album = [comps[1]]
             track_part = comps[2].split('-')
             if len(track_part) == 1:
-                tags.title = [comps[2]]
+                tags.title = [os.path.splitext(comps[2])[0]]
             # TODO: handle the else case
 
         return tags
@@ -104,6 +119,12 @@ class AudioFile:
         else:
             self.muta = None
 
+    @staticmethod
+    def new(path_: str) -> 'AudioFile':
+        f = AudioFile(path_)
+        if f.muta is not None:
+            return f
+
     def tags_on_disk(self) -> Tags:
         return Tags(
             artist=self.muta.get('artist', []) if self.muta else [],
@@ -131,7 +152,11 @@ class Tagger:
         self.dry_run = dry_run
 
     def tag_file(self, path_: str) -> None:
-        file_ = AudioFile(path_)
+        file_ = AudioFile.new(path_)
+        if not file_:
+            logger.debug(f"skipping unsupported file: {path_}")
+            return
+
         old_tags = file_.tags_on_disk()
 
         path_tags = Tags.from_path(path_)
@@ -147,6 +172,17 @@ class Tagger:
             if self.guard_dry_run("writing tags"):
                 file_.write_tags(new_tags)
 
+    def tag_file_tree(self, root: str) -> None:
+        for dir_, subdirs, files_ in os.walk(root):
+            for f in files_:
+                self.tag_file(os.path.join(dir_, f))
+
+    def tag_file_or_tree(self, path_: str) -> None:
+        if os.path.isdir(path_):
+            self.tag_file_tree(path_)
+        else:
+            self.tag_file(path_)
+
     def show_tagdif(self, path_: str, old_tags: Tags, new_tags: Tags):
         logger.info(f"updating tags for {path_}")
         logger.info(f"    {old_tags}")
@@ -183,7 +219,7 @@ def main():
         logging.getLogger().setLevel(logging.DEBUG)
 
     tagger = Tagger(dry_run=args.dry_run)
-    tagger.tag_file(args.path)
+    tagger.tag_file_or_tree(args.path)
 
 if __name__ == '__main__':
     main()