ff

src/ffx/media_descriptor_change_set.py

@@ -1,5 +1,6 @@
 import click
 
+from ffx.iso_language import IsoLanguage
 from ffx.media_descriptor import MediaDescriptor
 from ffx.track_descriptor import TrackDescriptor
 
@@ -117,7 +118,11 @@ class MediaDescriptorChangeSet():
                 sourceTrackDescriptor: TrackDescriptor = None):
 
         sourceTrackTags = sourceTrackDescriptor.getTags() if sourceTrackDescriptor is not None else {}
-        targetTrackTags = targetTrackDescriptor.getTags() if targetTrackDescriptor is not None else {}
+        targetTrackTags = (
+            self.normalizeTrackTags(targetTrackDescriptor.getTags())
+            if targetTrackDescriptor is not None
+            else {}
+        )
 
         trackCompareResult = {}
 
@@ -142,6 +147,25 @@ class MediaDescriptorChangeSet():
 
         return trackCompareResult
 
+    def normalizeTrackTagValue(self, tagKey, tagValue):
+        if tagKey != "language":
+            return tagValue
+
+        if isinstance(tagValue, IsoLanguage):
+            return tagValue.threeLetter()
+
+        trackLanguage = IsoLanguage.findThreeLetter(str(tagValue))
+        if trackLanguage != IsoLanguage.UNDEFINED:
+            return trackLanguage.threeLetter()
+
+        return tagValue
+
+    def normalizeTrackTags(self, trackTags: dict):
+        return {
+            tagKey: self.normalizeTrackTagValue(tagKey, tagValue)
+            for tagKey, tagValue in trackTags.items()
+        }
+
 
     def generateDispositionTokens(self):
         """
@@ -243,7 +267,7 @@ class MediaDescriptorChangeSet():
                 addedTracks: dict = self.__changeSetObj[MediaDescriptorChangeSet.TRACKS_KEY][DIFF_ADDED_KEY]
                 trackDescriptor: TrackDescriptor
                 for trackDescriptor in addedTracks.values():
-                    for tagKey, tagValue in trackDescriptor.getTags().items():
+                    for tagKey, tagValue in self.normalizeTrackTags(trackDescriptor.getTags()).items():
                         if not tagKey in self.__removeTrackKeys:
                             metadataTokens += [f"-metadata:s:{trackDescriptor.getType().indicator()}"
                                                 + f":{trackDescriptor.getSubIndex()}",
@@ -267,7 +291,7 @@ class MediaDescriptorChangeSet():
 
                         trackDescriptor = self.__targetTrackDescriptorsByIndex[trackIndex]
 
-                        for tagKey, tagValue in outputTrackTags.items():
+                        for tagKey, tagValue in self.normalizeTrackTags(outputTrackTags).items():
                             metadataTokens += [f"-metadata:s:{trackDescriptor.getType().indicator()}"
                                                 + f":{trackDescriptor.getSubIndex()}",
                                                 f"{tagKey}={tagValue}"]
@@ -285,7 +309,7 @@ class MediaDescriptorChangeSet():
                                 }
                                 | unchangedTrackTags
                             )
-                            for tagKey, tagValue in preservedTrackTags.items():
+                            for tagKey, tagValue in self.normalizeTrackTags(preservedTrackTags).items():
                                 metadataTokens += [f"-metadata:s:{trackDescriptor.getType().indicator()}"
                                                 + f":{trackDescriptor.getSubIndex()}",
                                                 f"{tagKey}={tagValue}"]

tests/unit/test_media_descriptor_change_set.py

@@ -27,6 +27,64 @@ class StaticConfig:
 
 
 class MediaDescriptorChangeSetTests(unittest.TestCase):
+    def test_non_primary_source_language_code_is_normalized_in_changed_track_metadata(self):
+        context = {
+            "logger": get_ffx_logger(),
+            "config": StaticConfig({}),
+        }
+
+        source_track = TrackDescriptor(
+            index=0,
+            source_index=0,
+            sub_index=0,
+            track_type=TrackType.AUDIO,
+            tags={"language": "ger", "title": "German Main"},
+        )
+        target_track = TrackDescriptor(
+            index=0,
+            source_index=0,
+            sub_index=0,
+            track_type=TrackType.AUDIO,
+            tags={"language": "ger", "title": "German Main"},
+        )
+
+        change_set = MediaDescriptorChangeSet(
+            context,
+            MediaDescriptor(track_descriptors=[target_track]),
+            MediaDescriptor(track_descriptors=[source_track]),
+        )
+
+        metadata_tokens = change_set.generateMetadataTokens()
+
+        self.assertIn("-metadata:s:a:0", metadata_tokens)
+        self.assertIn("language=deu", metadata_tokens)
+        self.assertNotIn("language=ger", metadata_tokens)
+
+    def test_target_only_track_language_metadata_uses_primary_code(self):
+        context = {
+            "logger": get_ffx_logger(),
+            "config": StaticConfig({}),
+        }
+
+        target_track = TrackDescriptor(
+            index=0,
+            source_index=0,
+            sub_index=0,
+            track_type=TrackType.AUDIO,
+            tags={"language": "ger", "title": "German Main"},
+        )
+
+        change_set = MediaDescriptorChangeSet(
+            context,
+            MediaDescriptor(track_descriptors=[target_track]),
+        )
+
+        metadata_tokens = change_set.generateMetadataTokens()
+
+        self.assertIn("-metadata:s:a:0", metadata_tokens)
+        self.assertIn("language=deu", metadata_tokens)
+        self.assertNotIn("language=ger", metadata_tokens)
+
     def test_external_subtitle_preserves_source_only_tags_except_removed_keys(self):
         context = {
             "logger": get_ffx_logger(),
@@ -79,6 +137,40 @@ class MediaDescriptorChangeSetTests(unittest.TestCase):
         self.assertNotIn("BPS=remove-me", metadata_tokens)
         self.assertIn("BPS=", metadata_tokens)
 
+    def test_external_subtitle_normalizes_preserved_source_language_metadata(self):
+        context = {
+            "logger": get_ffx_logger(),
+            "config": StaticConfig({}),
+        }
+
+        source_track = TrackDescriptor(
+            index=0,
+            source_index=0,
+            sub_index=0,
+            track_type=TrackType.SUBTITLE,
+            tags={"language": "ger", "title": "German Subtitle"},
+        )
+        target_track = TrackDescriptor(
+            index=0,
+            source_index=0,
+            sub_index=0,
+            track_type=TrackType.SUBTITLE,
+            tags={},
+            external_source_file="/tmp/external-subtitle.vtt",
+        )
+
+        change_set = MediaDescriptorChangeSet(
+            context,
+            MediaDescriptor(track_descriptors=[target_track]),
+            MediaDescriptor(track_descriptors=[source_track]),
+        )
+
+        metadata_tokens = change_set.generateMetadataTokens()
+
+        self.assertIn("-metadata:s:s:0", metadata_tokens)
+        self.assertIn("language=deu", metadata_tokens)
+        self.assertNotIn("language=ger", metadata_tokens)
+
     def test_target_only_tracks_still_emit_remove_tokens_for_configured_stream_keys(self):
         context = {
             "logger": get_ffx_logger(),