clean-tags.py: add support for cleaning vorbis comments in FLAC files

clean-tags.py

@@ -11,6 +11,7 @@ import logging
 import os
 
 from mutagen.id3 import ID3
+from mutagen.flac import FLAC
 from mutagen import MutagenError
 
 
@@ -30,14 +31,17 @@ class TagCleaner:
     them.
     """
     INDENTATION = "    "
-    FILE_EXTENSIONS = ['.mp3']
+    FILE_EXTENSIONS = ['.mp3', '.flac', '.ogg']
     TAGS_ID3 = ['TALB', 'TDRC', 'TIT2', 'TPE1', 'TRCK']
+    TAGS_VORBIS_COMMENTS = ['ALBUM', 'ARTIST', 'DATE', 'TITLE', 'TRACKNUMBER']
 
 
     def __init__(self, folder):
         """Construct a new tag cleaner instancce."""
         self._logger = logging.getLogger(__class__.__name__)
         self._dry = False
+        self._id3 = True
+        self._flac = True
         self._folder = folder
         self._files = []
         self._read_files(folder)
@@ -48,6 +52,16 @@ class TagCleaner:
         self._dry = dry
 
 
+    def set_id3(self, do):
+        """Set whether to clean ID3 tags or not."""
+        self._id3 = do
+
+
+    def set_flac(self, do):
+        """Set whether to clean FLAC vorbis comments or not."""
+        self._flac = do
+
+
     def clean_files(self):
         """Clean all found files."""
         self._logger.info("Clean files")
@@ -66,7 +80,7 @@ class TagCleaner:
                     continue
                 self._logger.debug("Found file \"%s\"", os.path.join(dirname, filename))
                 self._files.append(os.path.join(dirname, filename))
-        self._logger.info("Found %d files".format(len(self._files)))
+        self._logger.info("Found %d files", len(self._files))
 
 
     def _clean_file(self, filename):
@@ -74,7 +88,11 @@ class TagCleaner:
         self._logger.info("Clean file \"%s\"", filename)
         if os.path.isfile(filename):
             # ID3
-            self._clean_id3(filename)
+            if self._id3:
+                self._clean_id3(filename)
+            # FLAC
+            if self._flac:
+                self._clean_flac(filename)
         else:
             self._logger.info("Not a file: \"%s\"", filename)
 
@@ -136,7 +154,44 @@ class TagCleaner:
             else:
                 self._logger.info("Clean, nothing to do")
         except MutagenError as e:
-            self._logger.error("Cleaning of ID3 tags failed: %s", e)
+            self._logger.info("Cleaning of ID3 tags failed: %s", e)
+
+
+    def _clean_flac(self, filename):
+        """Clean FLAC vorbis comments."""
+        self._logger.info("Clean FLAC vorbis comments")
+        try:
+            flac = FLAC(filename)
+            invalid_comments = {}
+            if flac.tags:
+                invalid_comments = self._clean_vorbis_comments(flac.tags)
+            # Delete comments
+            if invalid_comments:
+                for key in invalid_comments.keys():
+                    print("Delete", key)
+                    del flac.tags[key]
+                if not self._dry:
+                    flac.save()
+                    print("File saved")
+                else:
+                    print("File not saved (running in dry mode)")
+            else:
+                self._logger.info("Clean, nothing to do")
+        except MutagenError as e:
+            self._logger.info("Cleaning of FLAC vorbis comments failed: %s", e)
+
+
+    def _clean_vorbis_comments(self, comments):
+        invalid_comments = {}
+        for key, value in comments:
+            if key not in TagCleaner.TAGS_VORBIS_COMMENTS:
+                invalid_comments[key] = value
+        if invalid_comments:
+            print("Unwanted comments:")
+            for key in invalid_comments.keys():
+                print("{}{}: {}".format(TagCleaner.INDENTATION, key, invalid_comments[key]))
+
+        return invalid_comments
 
 
 
@@ -147,7 +202,10 @@ if __name__ == "__main__":
     parser.add_argument('-d', '--dry', dest='dry', action='store_true', default=False, help="dry run, do not modify any file, just print information")
     parser.add_argument('-v', '--verbose', dest='verbosity', action='count', default=0, help="be verbose, show more information")
     parser.add_argument('-l', '--logfile', dest='logfile', help="specify name of logfile")
+    parser.add_argument('--no-id3', dest='id3', action='store_false', help="disable cleaning of ID3 tags")
+    parser.add_argument('--no-fac', dest='flac', action='store_false', help="disable cleaning of FLAC vorbis comments")
     parser.add_argument('folder', help="source folder to read audio files from")
+    parser.set_defaults(id3=True, flac=True)
     args = parser.parse_args()
 
     # Setup logging
@@ -160,6 +218,8 @@ if __name__ == "__main__":
     # Create tag cleaner instance
     tag_cleaner = TagCleaner(args.folder)
     tag_cleaner.set_dry(args.dry)
+    tag_cleaner.set_id3(args.id3)
+    tag_cleaner.set_flac(args.flac)
 
     # Run action
     tag_cleaner.clean_files()