[FEATURE] Embed thumbnails into audio files as cover art (#345)

* [FEATURE] Embed thumbnails into audio files as cover art

* lint

* fix ytdl options

examples/music_audio_from_videos.yaml

@@ -17,6 +17,7 @@ presets:
       quality: 128
 
     music_tags:
+      embed_thumbnail: False  # Set to True to embed album art
       tags:
         artist: "{custom_artist_name}"
         albumartist: "{custom_artist_name}"

src/ytdl_sub/plugins/music_tags.py

@@ -7,7 +7,9 @@ from ytdl_sub.entries.entry import Entry
 from ytdl_sub.plugins.plugin import Plugin
 from ytdl_sub.plugins.plugin import PluginOptions
 from ytdl_sub.utils.file_handler import FileMetadata
+from ytdl_sub.utils.thumbnail import convert_download_thumbnail
 from ytdl_sub.validators.string_formatter_validators import DictFormatterValidator
+from ytdl_sub.validators.validators import BoolValidator
 
 
 class MusicTagsOptions(PluginOptions):
@@ -31,9 +33,12 @@ class MusicTagsOptions(PluginOptions):
                artist: "{artist}"
                album: "{album}"
                genre: "ytdl downloaded music"
+             # Optional
+             embed_thumbnail: False
     """
 
     _required_keys = {"tags"}
+    _optional_keys = {"embed_thumbnail"}
 
     @classmethod
     def partial_validate(cls, name: str, value: Any) -> None:
@@ -48,6 +53,9 @@ class MusicTagsOptions(PluginOptions):
         super().__init__(name, value)
 
         self._tags = self._validate_key(key="tags", validator=DictFormatterValidator)
+        self._embed_thumbnail = self._validate_key_if_present(
+            key="embed_thumbnail", validator=BoolValidator, default=False
+        ).value
 
     @property
     def tags(self) -> DictFormatterValidator:
@@ -56,6 +64,13 @@ class MusicTagsOptions(PluginOptions):
         """
         return self._tags
 
+    @property
+    def embed_thumbnail(self) -> bool:
+        """
+        Optional. Whether to embed the thumbnail into the audio file.
+        """
+        return self._embed_thumbnail
+
 
 class MusicTagsPlugin(Plugin[MusicTagsOptions]):
     plugin_options_type = MusicTagsOptions
@@ -85,7 +100,23 @@ class MusicTagsPlugin(Plugin[MusicTagsOptions]):
             audio_file = mediafile.MediaFile(entry.get_download_file_path())
             for tag_name, tag_value in tags_to_write.items():
                 setattr(audio_file, tag_name, tag_value)
+
+            if self.plugin_options.embed_thumbnail:
+                # convert the entry thumbnail so it is embedded as jpg
+                convert_download_thumbnail(entry=entry)
+
+                with open(entry.get_download_thumbnail_path(), "rb") as thumb:
+                    mediafile_img = mediafile.Image(
+                        data=thumb.read(), desc="cover", type=mediafile.ImageType.front
+                    )
+
+                audio_file.images = [mediafile_img]
+
             audio_file.save()
 
         # report the tags written
-        return FileMetadata.from_dict(value_dict=tags_to_write, title="Music Tags")
+        title = f"{'Embedded Thumbnail, ' if self.plugin_options.embed_thumbnail else ''}Music Tags"
+        return FileMetadata.from_dict(
+            value_dict=tags_to_write,
+            title=title,
+        )

src/ytdl_sub/subscriptions/subscription_ytdl_options.py

@@ -53,17 +53,11 @@ class SubscriptionYTDLOptions:
         """
         ytdl_options = {
             # Download all files in the format of {id}.{ext}
-            "outtmpl": str(Path(self._working_directory) / "%(id)s.%(ext)s")
+            "outtmpl": str(Path(self._working_directory) / "%(id)s.%(ext)s"),
+            # Always write thumbnails
+            "writethumbnail": True,
         }
 
-        if (
-            self._downloader.supports_download_archive
-            and self._preset.output_options.maintain_download_archive
-        ):
-            ytdl_options["download_archive"] = str(
-                Path(self._working_directory) / self._enhanced_download_archive.archive_file_name
-            )
-
         return ytdl_options
 
     @property
@@ -71,7 +65,6 @@ class SubscriptionYTDLOptions:
         return {
             "skip_download": True,
             "writethumbnail": False,
-            # TODO: find a way to not write subtitles; using `simulate: True` breaks tests
         }
 
     @property
@@ -85,10 +78,14 @@ class SubscriptionYTDLOptions:
     @property
     def _output_options(self) -> Dict:
         ytdl_options = {}
-        output_options = self._preset.output_options
 
-        if output_options.thumbnail_name:
+        if (
-            ytdl_options["writethumbnail"] = True
+            self._downloader.supports_download_archive
+            and self._preset.output_options.maintain_download_archive
+        ):
+            ytdl_options["download_archive"] = str(
+                Path(self._working_directory) / self._enhanced_download_archive.archive_file_name
+            )
 
         return ytdl_options
 
@@ -111,6 +108,7 @@ class SubscriptionYTDLOptions:
         """
         return YTDLOptionsBuilder().add(
             self._global_options,
+            self._output_options,
             self._plugin_ytdl_options(DateRangePlugin),
             self._user_ytdl_options,  # user ytdl options...
             self._info_json_only_options,  # then info_json_only options
@@ -123,25 +121,18 @@ class SubscriptionYTDLOptions:
         YTDLOptionsBuilder
             Builder with values set based on the subscription for actual downloading
         """
-        ytdl_options_builder = YTDLOptionsBuilder().add(self._global_options)
+        ytdl_options_builder = YTDLOptionsBuilder().add(
+            self._global_options,
+            self._output_options,
+            self._plugin_ytdl_options(DateRangePlugin),
+            self._plugin_ytdl_options(FileConvertPlugin),
+            self._plugin_ytdl_options(SubtitlesPlugin),
+            self._plugin_ytdl_options(ChaptersPlugin),
+            self._plugin_ytdl_options(AudioExtractPlugin),
+            self._user_ytdl_options,  # user ytdl options...
+        )
+        # Add dry run options last if enabled
         if self._dry_run:
-            ytdl_options_builder.add(
+            ytdl_options_builder.add(self._dry_run_options)
-                self._plugin_ytdl_options(DateRangePlugin),
-                self._plugin_ytdl_options(FileConvertPlugin),
-                self._plugin_ytdl_options(SubtitlesPlugin),
-                self._plugin_ytdl_options(ChaptersPlugin),
-                self._user_ytdl_options,  # user ytdl options...
-                self._dry_run_options,  # then dry-run
-            )
-        else:
-            ytdl_options_builder.add(
-                self._output_options,
-                self._plugin_ytdl_options(DateRangePlugin),
-                self._plugin_ytdl_options(FileConvertPlugin),
-                self._plugin_ytdl_options(SubtitlesPlugin),
-                self._plugin_ytdl_options(ChaptersPlugin),
-                self._plugin_ytdl_options(AudioExtractPlugin),
-                self._user_ytdl_options,  # user ytdl options last
-            )
 
         return ytdl_options_builder

src/ytdl_sub/utils/thumbnail.py

@@ -1,3 +1,4 @@
+import os
 import tempfile
 from typing import Optional
 from urllib.request import urlopen
@@ -29,6 +30,11 @@ def convert_download_thumbnail(entry: Entry, error_if_not_found: bool = True) ->
     """
     download_thumbnail_path = entry.get_ytdlp_download_thumbnail_path()
     download_thumbnail_path_as_jpg = entry.get_download_thumbnail_path()
+
+    # If it was already converted, do not convert again
+    if os.path.isfile(download_thumbnail_path_as_jpg):
+        return
+
     if not download_thumbnail_path:
         if error_if_not_found:
             raise ValueError("Thumbnail not found")

tests/e2e/plugins/test_audio_extract.py

@@ -11,6 +11,7 @@ def single_song_preset_dict(output_directory):
         "preset": "song",
         "download": {"url": "https://www.youtube.com/watch?v=2lAe1cqCOXo"},
         "output_options": {"output_directory": output_directory},
+        "music_tags": {"embed_thumbnail": True},
         # download the worst format so it is fast
         "ytdl_options": {
             "format": "worst[ext=mp4]",
@@ -34,7 +35,6 @@ def multiple_songs_preset_dict(output_directory):
     }
 
 
-# TODO: Test album from chapters
 class TestAudioExtract:
     @pytest.mark.parametrize("dry_run", [True, False])
     def test_audio_extract_single_song(

tests/resources/expected_downloads_summaries/plugins/test_audio_extract_single.json

@@ -1,3 +1,3 @@
 {
-  "YouTube Rewind 2019： For the Record ｜ #YouTubeRewind.mp3": "d3687bf6c13a2a5f3a8a05fbce9be28c"
+  "YouTube Rewind 2019： For the Record ｜ #YouTubeRewind.mp3": "8e531bfb93b144fe652f44950e31b3f7"
 }

tests/resources/transaction_log_summaries/plugins/test_audio_extract_single.txt

@@ -2,7 +2,7 @@ Files created:
 ----------------------------------------
 {output_directory}
   YouTube Rewind 2019： For the Record ｜ #YouTubeRewind.mp3
-    Music Tags:
+    Embedded Thumbnail, Music Tags:
       album: Singles
       albumartist: YouTube
       artist: YouTube