MPEG-4 Unicode freeform frames are now encoded as UTF-8 bytes

beets/mediafile.py

@@ -382,6 +382,11 @@ class MediaField(object):
         
         if style.packing:
             out = Packed(out, style.packing)[style.pack_pos]
+
+        # MPEG-4 freeform frames are (should be?) encoded as UTF-8.
+        if obj.type == 'mp4' and style.key.startswith('----:') and \
+                isinstance(out, str):
+            out = out.decode('utf8')
         
         return _safe_cast(self.out_type, out)
     
@@ -410,8 +415,8 @@ class MediaField(object):
                         out = u''
                     # We trust that packed values are handled above.
         
-                # convert to correct storage type (irrelevant for
-                # packed values)
+                # Convert to correct storage type (irrelevant for
+                # packed values).
                 if style.as_type == unicode:
                     if out is None:
                         out = u''
@@ -429,7 +434,13 @@ class MediaField(object):
                 elif style.as_type in (bool, str):
                     out = style.as_type(out)
         
-            # store the data
+            # MPEG-4 "freeform" (----) frames must be encoded as UTF-8
+            # byte strings.
+            if obj.type == 'mp4' and style.key.startswith('----:') and \
+                    isinstance(out, unicode):
+                out = out.encode('utf8')
+
+            # Store the data.
             self._storedata(obj, out, style)
 
 class CompositeDateField(object):

beets/util/__init__.py

@@ -180,7 +180,7 @@ CHAR_REPLACE = [
     (re.compile(r'[\\/\?]|^\.'), '_'),
     (re.compile(r':'), '-'),
 ]
-CHAR_REPLACE_WINDOWS = re.compile('["\*<>\|]|^\.|\.$| +$'), '_'
+CHAR_REPLACE_WINDOWS = re.compile(r'["\*<>\|]|^\.|\.$| +$'), '_'
 def sanitize_path(path, pathmod=None):
     """Takes a path and makes sure that it is legal. Returns a new path.
     Only works with fragments; won't work reliably on Windows when a

test/test_mediafile.py

@@ -17,6 +17,7 @@
 
 import unittest
 import os
+import shutil
 
 import _common
 import beets.mediafile
@@ -149,6 +150,23 @@ class SideEffectsTest(unittest.TestCase):
         new_mtime = os.stat(self.empty).st_mtime
         self.assertEqual(old_mtime, new_mtime)
 
+class EncodingTest(unittest.TestCase):
+    def setUp(self):
+        src = os.path.join(_common.RSRC, 'full.m4a')
+        self.path = os.path.join(_common.RSRC, 'test.m4a')
+        shutil.copy(src, self.path)
+
+        self.mf = beets.mediafile.MediaFile(self.path)
+
+    def tearDown(self):
+        os.remove(self.path)
+
+    def test_unicode_label_in_m4a(self):
+        self.mf.label = u'foo\xe8bar'
+        self.mf.save()
+        new_mf = beets.mediafile.MediaFile(self.path)
+        self.assertEqual(new_mf.label, u'foo\xe8bar')
+
 def suite():
     return unittest.TestLoader().loadTestsFromName(__name__)