Move mutagen PP to ffmpeg PP (WIP)

yt_dlp/__init__.py

@@ -690,10 +690,7 @@ def get_postprocessors(opts):
             'add_chapters': opts.addchapters,
             'add_metadata': opts.addmetadata,
             'add_infojson': opts.embed_infojson,
-        }
+            'prefer_mutagen': opts.prefer_mutagen,
-    if opts.prefer_mutagen:
-        yield {
-            'key': 'Mutagen',
         }
     # Deprecated
     # This should be above EmbedThumbnail since sponskrub removes the thumbnail attachment

yt_dlp/postprocessor/__init__.py

@@ -30,7 +30,6 @@ from .metadataparser import (
 )
 from .modify_chapters import ModifyChaptersPP
 from .movefilesafterdownload import MoveFilesAfterDownloadPP
-from .mutagen import MutagenPP
 from .sponskrub import SponSkrubPP
 from .sponsorblock import SponsorBlockPP
 from .xattrpp import XAttrMetadataPP

yt_dlp/postprocessor/ffmpeg.py

@@ -32,6 +32,26 @@ from ..utils import (
     variadic,
     write_json_file,
 )
+from ..dependencies import mutagen
+
+if mutagen:
+    import mutagen
+    from mutagen import (
+        FileType,
+        aiff,
+        dsdiff,
+        dsf,
+        flac,
+        id3,
+        mp3,
+        mp4,
+        oggopus,
+        oggspeex,
+        oggtheora,
+        oggvorbis,
+        trueaudio,
+        wave,
+    )
 
 EXT_TO_OUT_FORMATS = {
     'aac': 'adts',
@@ -668,11 +688,56 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 
 class FFmpegMetadataPP(FFmpegPostProcessor):
 
-    def __init__(self, downloader, add_metadata=True, add_chapters=True, add_infojson='if_exists'):
+    _MUTAGEN_SUPPORTED_EXTS = ('alac', 'aiff', 'flac', 'mp3', 'm4a', 'ogg', 'opus', 'vorbis', 'wav')
+    _VORBIS_METADATA = {
+        'title': 'title',
+        'artist': 'artist',
+        'genre': 'genre',
+        'date': 'date',
+        'album': 'album',
+        'albumartist': 'album_artist',
+        'description': 'description',
+        'comment': 'comment',
+        'composer': 'composer',
+        'tracknumber': 'track',
+        'WWWAUDIOFILE': 'purl',  # https://getmusicbee.com/forum/index.php?topic=39759.0
+    }
+    _ID3_METADATA = {
+        'TIT2': 'title',
+        'TPE1': 'artist',
+        'COMM': 'description',
+        'TCON': 'genre',
+        'WFED': 'purl',
+        'WOAF': 'purl',
+        'TDAT': 'date',
+        'TALB': 'album',
+        'TPE2': 'album_artist',
+        'TRCK': 'track',
+        'TCOM': 'composer',
+        'TPOS': 'disc',
+    }
+    _MP4_METADATA = {
+        '\251ART': 'artist',
+        '\251nam': 'title',
+        '\251gen': 'genre',
+        '\251day': 'date',
+        '\251alb': 'album',
+        'aART': 'album_artist',
+        '\251cmt': 'description',
+        '\251wrt': 'composer',
+        'disk': 'disc',
+        'tvsh': 'show',
+        'tvsn': 'season_number',
+        'egid': 'episode_id',
+        'tven': 'episode_sort',
+    }
+
+    def __init__(self, downloader, add_metadata=True, add_chapters=True, add_infojson='if_exists', prefer_mutagen=False):
         FFmpegPostProcessor.__init__(self, downloader)
         self._add_metadata = add_metadata
         self._add_chapters = add_chapters
         self._add_infojson = add_infojson
+        self._prefer_mutagen = prefer_mutagen
 
     @staticmethod
     def _options(target_ext):
@@ -681,8 +746,91 @@ class FFmpegMetadataPP(FFmpegPostProcessor):
         if audio_only:
             yield from ('-vn', '-acodec', 'copy')
 
+    def _use_mutagen(self, info):
+        if not self._prefer_mutagen:
+            return False
+        if info['ext'] not in self._MUTAGEN_SUPPORTED_EXTS:
+            return False
+        if self._add_chapters and info.get('chapters'):
+            # mutagen can't handle adding chapters to M4A
+            return False
+        if not mutagen:
+            self.report_warning('module mutagen was not found. Please install using `python3 -m pip install mutagen`')
+            return False
+        return True
+
+    if mutagen:
+        @functools.singledispatchmethod
+        def _assemble_metadata(self, file: FileType, meta: dict) -> None:
+            raise FFmpegPostProcessorError(f'Filetype {file.__class__.__name__} is not currently supported')
+
+        @_assemble_metadata.register(oggvorbis.OggVorbis)
+        @_assemble_metadata.register(oggtheora.OggTheora)
+        @_assemble_metadata.register(oggspeex.OggSpeex)
+        @_assemble_metadata.register(oggopus.OggOpus)
+        @_assemble_metadata.register(flac.FLAC)
+        def _(self, file: oggopus.OggOpus, meta: dict) -> None:
+            for file_key, meta_key in self._VORBIS_METADATA.items():
+                if meta.get(meta_key):
+                    file[file_key] = meta[meta_key]
+
+        @_assemble_metadata.register(trueaudio.TrueAudio)
+        @_assemble_metadata.register(dsf.DSF)
+        @_assemble_metadata.register(dsdiff.DSDIFF)
+        @_assemble_metadata.register(aiff.AIFF)
+        @_assemble_metadata.register(mp3.MP3)
+        @_assemble_metadata.register(wave.WAVE)
+        def _(self, file: wave.WAVE, meta: dict) -> None:
+            for file_key, meta_key in self._ID3_METADATA.items():
+                if meta.get(meta_key):
+                    id3_class = getattr(id3, file_key)
+                    if issubclass(id3_class, id3.UrlFrame):
+                        file[file_key] = id3_class(url=meta[meta_key])
+                    else:
+                        file[file_key] = id3_class(encoding=id3.Encoding.UTF8, text=meta[meta_key])
+
+        @_assemble_metadata.register(mp4.MP4)
+        def _(self, file: mp4.MP4, meta: dict) -> None:
+            for file_key, meta_key in self._MP4_METADATA.items():
+                if meta.get(meta_key):
+                    file[file_key] = meta[meta_key]
+
+            if meta.get('purl'):
+                # https://getmusicbee.com/forum/index.php?topic=39759.0
+                file['----:com.apple.iTunes:WWWAUDIOFILE'] = meta['purl'].encode()
+                file['purl'] = meta['purl'].encode()
+
+            if meta.get('track'):
+                file['trkn'] = [(meta['track'], 0)]
+
+    def _run_mutagen(self, info):
+        self.to_screen('Using mutagen to embed metadata')
+        filename = info['filepath']
+        metadata = self._get_metadata_dict(info)['common']
+        if not metadata:
+            self.to_screen('There isn\'t any metadata to add')
+            return [], info
+
+        self.to_screen(f'Adding metadata to "{filename}"')
+        try:
+            f = mutagen.File(filename)
+            self._assemble_metadata(f, metadata)
+            f.save()
+        except Exception as err:
+            raise FFmpegPostProcessorError(f'Unable to embed metadata; {err}')
+
+        return [], info
+
     @PostProcessor._restrict_to(images=False)
     def run(self, info):
+
+        if self._use_mutagen(info):
+            try:
+                self._run_mutagen(info)
+                return [], info
+            except Exception as err:
+                self.report_warning(f'Unable to embed metadata using mutagen; {err}')
+
         self._fixup_chapters(info)
         filename, metadata_filename = info['filepath'], None
         files_to_delete, options = [], []
@@ -732,7 +880,7 @@ class FFmpegMetadataPP(FFmpegPostProcessor):
             f.write(metadata_file_content)
         yield ('-map_metadata', '1')
 
-    def _get_metadata_opts(self, info):
+    def _get_metadata_dict(self, info):
         meta_prefix = 'meta'
         metadata = collections.defaultdict(dict)
 
@@ -774,6 +922,10 @@ class FFmpegMetadataPP(FFmpegPostProcessor):
             mobj = re.fullmatch(meta_regex, key)
             if value is not None and mobj:
                 metadata[mobj.group('i') or 'common'][mobj.group('key')] = value.replace('\0', '')
+        return metadata
+
+    def _get_metadata_opts(self, info):
+        metadata = self._get_metadata_dict(info)
 
         # Write id3v1 metadata also since Windows Explorer can't handle id3v2 tags
         yield ('-write_id3v1', '1')

yt_dlp/postprocessor/mutagen.py

@@ -1,204 +0,0 @@
-from __future__ import annotations
-import collections
-from functools import singledispatchmethod
-import re
-from typing import TypedDict
-
-from yt_dlp.utils._utils import PostProcessingError, variadic
-from ..dependencies import mutagen
-
-if mutagen:
-    import mutagen
-    from mutagen import (
-        FileType,
-        aiff,
-        dsdiff,
-        dsf,
-        flac,
-        id3,
-        mp3,
-        mp4,
-        oggopus,
-        oggspeex,
-        oggtheora,
-        oggvorbis,
-        trueaudio,
-        wave,
-    )
-
-from yt_dlp.postprocessor.common import PostProcessor
-
-
-class MutagenPPError(PostProcessingError):
-    pass
-
-
-class MutagenPP(PostProcessor):
-    def __init__(self, downloader=None):
-        PostProcessor.__init__(self, downloader)
-
-    class MetadataInfo(TypedDict):
-        title: str | None
-        date: str | None
-        description: str | None
-        synopsis: str | None
-        purl: str | None
-        comment: str | None
-        track: str | None
-        artist: str | None
-        composer: str | None
-        genre: str | None
-        album: str | None
-        album_artist: str | None
-        disc: str | None
-        show: str | None
-        season_number: str | None
-        episode_id: str | None
-        episode_sort: str | None
-
-    if mutagen:
-        @singledispatchmethod
-        @staticmethod
-        def _assemble_metadata(file: FileType, meta: MetadataInfo) -> None:
-            raise MutagenPPError(f'Filetype {file.__class__.__name__} is not currently supported')
-
-        @staticmethod
-        def _set_metadata(file: FileType, meta: MetadataInfo, file_name: str, meta_name: str):
-            if meta[meta_name]:
-                file[file_name] = meta[meta_name]
-
-        @_assemble_metadata.register(oggvorbis.OggVorbis)
-        @_assemble_metadata.register(oggtheora.OggTheora)
-        @_assemble_metadata.register(oggspeex.OggSpeex)
-        @_assemble_metadata.register(oggopus.OggOpus)
-        @_assemble_metadata.register(flac.FLAC)
-        @staticmethod
-        def _(file: oggopus.OggOpus, meta: MetadataInfo) -> None:
-            MutagenPP._set_metadata(file, meta, 'artist', 'artist')
-            MutagenPP._set_metadata(file, meta, 'title', 'title')
-            MutagenPP._set_metadata(file, meta, 'genre', 'genre')
-            MutagenPP._set_metadata(file, meta, 'date', 'date')
-            MutagenPP._set_metadata(file, meta, 'album', 'album')
-            MutagenPP._set_metadata(file, meta, 'albumartist', 'album_artist')
-            MutagenPP._set_metadata(file, meta, 'description', 'description')
-            MutagenPP._set_metadata(file, meta, 'comment', 'comment')
-            MutagenPP._set_metadata(file, meta, 'composer', 'composer')
-            MutagenPP._set_metadata(file, meta, 'tracknumber', 'track')
-
-            # https://getmusicbee.com/forum/index.php?topic=39759.0
-            MutagenPP._set_metadata(file, meta, 'WWWAUDIOFILE', 'purl')
-
-        @_assemble_metadata.register(trueaudio.TrueAudio)
-        @_assemble_metadata.register(dsf.DSF)
-        @_assemble_metadata.register(dsdiff.DSDIFF)
-        @_assemble_metadata.register(aiff.AIFF)
-        @_assemble_metadata.register(mp3.MP3)
-        @_assemble_metadata.register(wave.WAVE)
-        @staticmethod
-        def _(file: wave.WAVE, meta: MetadataInfo) -> None:
-
-            def _set_metadata(file_name: str, meta_name: str):
-                if meta[meta_name]:
-                    id3_class = getattr(id3, file_name)
-                    file[file_name] = id3_class(encoding=id3.Encoding.UTF8, text=meta[meta_name])
-
-            _set_metadata('TIT2', 'title')
-            _set_metadata('TPE1', 'artist')
-            _set_metadata('COMM', 'description')
-            _set_metadata('TCON', 'genre')
-            _set_metadata('WFED', 'purl')
-            _set_metadata('WOAF', 'purl')
-            _set_metadata('TDAT', 'date')
-            _set_metadata('TALB', 'album')
-            _set_metadata('TPE2', 'album_artist')
-            _set_metadata('TRCK', 'track')
-            _set_metadata('TCOM', 'composer')
-            _set_metadata('TPOS', 'disc')
-
-        @_assemble_metadata.register(mp4.MP4)
-        @staticmethod
-        def _(file: mp4.MP4, meta: MetadataInfo) -> None:
-            MutagenPP._set_metadata(file, meta, '\251ART', 'artist')
-            MutagenPP._set_metadata(file, meta, '\251nam', 'title')
-            MutagenPP._set_metadata(file, meta, '\251gen', 'genre')
-            MutagenPP._set_metadata(file, meta, '\251day', 'date')
-            MutagenPP._set_metadata(file, meta, '\251alb', 'album')
-            MutagenPP._set_metadata(file, meta, 'aART', 'album_artist')
-            MutagenPP._set_metadata(file, meta, '\251cmt', 'description')
-            MutagenPP._set_metadata(file, meta, '\251wrt', 'composer')
-            MutagenPP._set_metadata(file, meta, 'disk', 'disc')
-            MutagenPP._set_metadata(file, meta, 'tvsh', 'show')
-            MutagenPP._set_metadata(file, meta, 'tvsn', 'season_number')
-            MutagenPP._set_metadata(file, meta, 'egid', 'episode_id')
-            MutagenPP._set_metadata(file, meta, 'tven', 'episode_sort')
-
-            if meta['purl']:
-                # https://getmusicbee.com/forum/index.php?topic=39759.0
-                file['----:com.apple.iTunes:WWWAUDIOFILE'] = meta['purl'].encode()
-                file['purl'] = meta['purl'].encode()
-
-            if meta['track']:
-                file['trkn'] = [(meta['track'], 0)]
-
-    def _get_metadata_from_info(self, info) -> MetadataInfo:
-        meta_prefix = 'meta'
-        metadata: dict[str, self.MetadataInfo] = collections.defaultdict(
-            lambda: collections.defaultdict(lambda: None),
-        )
-
-        def add(meta_list, info_list=None):
-            value = next((
-                info[key] for key in [f'{meta_prefix}_', *variadic(info_list or meta_list)]
-                if info.get(key) is not None), None)
-            if value not in ('', None):
-                value = ', '.join(map(str, variadic(value)))
-                value = value.replace('\0', '')  # nul character cannot be passed in command line
-                metadata['common'].update({meta_f: value for meta_f in variadic(meta_list)})
-
-        add('title', ('track', 'title'))
-        add('date', 'upload_date')
-        add(('description', 'synopsis'), 'description')
-        add(('purl', 'comment'), 'webpage_url')
-        add('track', 'track_number')
-        add('artist', ('artist', 'artists', 'creator', 'creators', 'uploader', 'uploader_id'))
-        add('composer', ('composer', 'composers'))
-        add('genre', ('genre', 'genres'))
-        add('album')
-        add('album_artist', ('album_artist', 'album_artists'))
-        add('disc', 'disc_number')
-        add('show', 'series')
-        add('season_number')
-        add('episode_id', ('episode', 'episode_id'))
-        add('episode_sort', 'episode_number')
-        if 'embed-metadata' in self.get_param('compat_opts', []):
-            add('comment', 'description')
-            metadata['common'].pop('synopsis', None)
-
-        meta_regex = rf'{re.escape(meta_prefix)}(?P<i>\d+)?_(?P<key>.+)'
-        for key, value in info.items():
-            mobj = re.fullmatch(meta_regex, key)
-            if value is not None and mobj:
-                metadata[mobj.group('i') or 'common'][mobj.group('key')] = value.replace('\0', '')
-
-        return metadata['common']
-
-    @PostProcessor._restrict_to(video=False, images=False)
-    def run(self, info):
-        if not mutagen:
-            raise MutagenPPError('module mutagen was not found. Please install using `python3 -m pip install mutagen`')
-        filename = info['filepath']
-        metadata = self._get_metadata_from_info(info)
-        if not metadata:
-            self.to_screen('There isn\'t any metadata to add')
-            return [], info
-
-        self.to_screen(f'Adding metadata to "{filename}"')
-        try:
-            f = mutagen.File(filename)
-            metadata = self._get_metadata_from_info(info)
-            self._assemble_metadata(f, metadata)
-            f.save()
-        except Exception as err:
-            raise MutagenPPError(f'Unable to embed metadata; {err}')
-
-        return [], info