Populate artists in SoundCloud from author, fall back to author in artist match if no artists

spotdl/providers/audio/bandcamp.py

@@ -170,15 +170,15 @@ class BandCamp(AudioProvider):
 
     def get_results(self, search_term: str, *_args, **_kwargs) -> List[Result]:
         """
-        Get results from slider.kz
+        Get results from BandCamp
 
         ### Arguments
         - search_term: The search term to search for.
         - args: Unused.
         - kwargs: Unused.
 
         ### Returns
-        - A list of slider.kz results if found, None otherwise.
+        - A list of BandCamp results if found, None otherwise.
         """
 
         try:

spotdl/providers/audio/soundcloud.py

@@ -40,15 +40,15 @@ def __init__(self, *args: Any, **kwargs: Any) -> None:
 
     def get_results(self, search_term: str, *_args, **_kwargs) -> List[Result]:
         """
-        Get results from slider.kz
+        Get results from SoundCloud
 
         ### Arguments
         - search_term: The search term to search for.
         - args: Unused.
         - kwargs: Unused.
 
         ### Returns
-        - A list of slider.kz results if found, None otherwise.
+        - A list of SoundCloud results if found, None otherwise.
         """
 
         results = list(islice(self.client.search(search_term), 20))
@@ -83,6 +83,7 @@ def get_results(self, search_term: str, *_args, **_kwargs) -> List[Result]:
                     verified=result.user.verified,
                     duration=result.full_duration,
                     author=result.user.username,
+                    artists=(result.user.username,),
                     result_id=str(result.id),
                     isrc_search=False,
                     search_query=search_term,

spotdl/utils/matching.py

@@ -287,21 +287,26 @@ def get_best_matches(
 
 def calc_main_artist_match(song: Song, result: Result) -> float:
     """
-    Check if main artist is present in list of artists
+    Calculate how well the main artist of the song matches the result.
+    Falls back to result.author if result.artists is not populated.
 
     ### Arguments
-    - main_artist: main artist to check
-    - artists: list of artists to check
+    - song: song to match
+    - result: result to match
 
     ### Returns
-    - True if main artist is present in list of artists, False otherwise
+    - artist match percentage (0.0 to 100.0)
     """
 
     main_artist_match = 0.0
 
-    # Result has no artists, return 0.0
+    # Result has no artists, fall back to author
+    # If no author either, return 0.0
     if not result.artists:
-        return main_artist_match
+        if not result.author:
+            return main_artist_match
+
+        return ratio(slugify(song.artists[0]), slugify(result.author))
 
     song_artists, result_artists = list(map(slugify, song.artists)), list(
         map(slugify, result.artists)

tests/test_matching.py

@@ -2,7 +2,9 @@
 
 from spotdl.providers.audio.base import AudioProviderError
 from spotdl.providers.audio.ytmusic import YouTubeMusic
+from spotdl.types.result import Result
 from spotdl.types.song import Song
+from spotdl.utils.matching import calc_main_artist_match
 from spotdl.utils.spotify import SpotifyClient
 from tests.conftest import new_initialize
 
@@ -429,3 +431,93 @@ def test_ytmusic_matching(monkeypatch, query, expected, capsys):
 
     except AudioProviderError:
         pytest.skip("YouTube Music search failed")
+
+
+SONG_DICT = {
+    "name": "流転の果て",
+    "artists": ["tokiwa"],
+    "artist": "tokiwa",
+    "album_id": "test",
+    "album_name": "勿忘",
+    "album_artist": "tokiwa",
+    "album_type": "album",
+    "genres": [],
+    "disc_number": 1,
+    "disc_count": 1,
+    "duration": 222,
+    "year": 2023,
+    "date": "2023-01-01",
+    "track_number": 1,
+    "tracks_count": 1,
+    "isrc": "",
+    "song_id": "7yfMsNxODzSIysdZ47JrlG",
+    "cover_url": "https://example.com/cover.jpg",
+    "explicit": False,
+    "publisher": "test",
+    "url": "https://open.spotify.com/track/7yfMsNxODzSIysdZ47JrlG",
+    "copyright_text": "",
+}
+
+
+def test_calc_main_artist_match_with_artists():
+    song = Song.from_dict(SONG_DICT)
+    result = Result(
+        source="soundcloud",
+        url="https://soundcloud.com/tokiwa_anka/pnog8medav1f",
+        verified=False,
+        name="流転の果て",
+        duration=222041,
+        author="tokiwa",
+        artists=("tokiwa",),
+        result_id="941283319",
+    )
+
+    assert calc_main_artist_match(song, result) == 100.0
+
+
+def test_calc_main_artist_match_falls_back_to_author():
+    song = Song.from_dict(SONG_DICT)
+    result = Result(
+        source="soundcloud",
+        url="https://soundcloud.com/tokiwa_anka/pnog8medav1f",
+        verified=False,
+        name="流転の果て",
+        duration=222041,
+        author="tokiwa",
+        result_id="941283319",
+    )
+
+    assert result.artists is None
+    assert calc_main_artist_match(song, result) == 100.0
+
+
+def test_calc_main_artist_match_no_artists_no_author():
+    song = Song.from_dict(SONG_DICT)
+    result = Result(
+        source="soundcloud",
+        url="https://soundcloud.com/unknown/test",
+        verified=False,
+        name="流転の果て",
+        duration=222041,
+        author="",
+        result_id="000",
+    )
+
+    assert result.artists is None
+    assert calc_main_artist_match(song, result) == 0.0
+
+
+def test_calc_main_artist_match_author_fuzzy():
+    song = Song.from_dict({**SONG_DICT, "artists": ["Carly Rae Jepsen"], "artist": "Carly Rae Jepsen"})
+    result = Result(
+        source="youtube",
+        url="https://www.youtube.com/watch?v=abc",
+        verified=False,
+        name="I Really Like You",
+        duration=200,
+        author="Carly Rae Jepsen - Topic",
+        result_id="abc",
+    )
+
+    assert result.artists is None
+    assert calc_main_artist_match(song, result) > 70.0