feat: add lrclib and genius providers (#5)

* feat: lrclib and genius providers

* revert `prepend_header` to true

* change providers order

---------

Co-authored-by: mrsobakin <68982655+mrsobakin@users.noreply.github.com>

config.toml

@@ -1,7 +1,7 @@
 [providers]
-order = ["musixmatch", "kugou"]
+order = ["musixmatch", "kugou", "lrclib", "genius"]
 delay = 10
 prepend_header = true
 
 [providers.musixmatch]
-token = "123456"
+token = ""

lrc_dl/__init__.py

@@ -1,10 +1,10 @@
 import time
-from typing import Optional
-from pathlib import Path
 import traceback
+from pathlib import Path
+from typing import Optional
 
 # Initialize classes from lrc_dl/providers
-import lrc_dl.providers
+import lrc_dl.providers as _
 from lrc_dl.core import Song
 from lrc_dl.registry import Registry
 from lrc_dl.config import LyricsDlConfig
@@ -23,11 +23,12 @@ class LyricsDl:
         self.providers = []
 
         for name in config.order:
-            Provider = providers_classes[name]
-            provider_config = config.providers_configs.get(name)
+            Provider = providers_classes.get(name)
 
-            if not provider_config:
-                provider_config = {}
+            if not Provider:
+                continue
+
+            provider_config = config.providers_configs.get(name, {})
 
             try:
                 provider = Provider(**provider_config)
@@ -57,7 +58,7 @@ class LyricsDl:
 
                 return lyrics
 
-            self.logger.info(f"[{provider.name}] No lyrics was found!")
+            self.logger.info(f"[{provider.name}] No lyrics were found!")
 
         return None
 
@@ -78,7 +79,7 @@ class LyricsDl:
         lyrics = self.fetch_lyrics(song)
 
         if not lyrics:
-            self.logger.error("[lrc-dl] No lyrics was found!")
+            self.logger.error("[lrc-dl] No lyrics were found!")
             return True
 
         with open(lyrics_path, "w") as f:

lrc_dl/config.py

@@ -21,7 +21,7 @@ CONFIG_PATH = _get_config_file()
 
 @dataclass
 class LyricsDlConfig:
-    order: list[str] = field(default_factory=lambda: ["kugou", "youtube"])
+    order: list[str] = field(default_factory=lambda: ["kugou", "lrclib", "genius"])
     delay: float | None = 10
     prepend_header: bool = True
     providers_configs: dict[str, dict] = field(default_factory=lambda: {})

lrc_dl/providers/__init__.py

@@ -1,3 +1,5 @@
-from lrc_dl.providers import musixmatch
 from lrc_dl.providers import kugou
+from lrc_dl.providers import lrclib
+from lrc_dl.providers import musixmatch
+from lrc_dl.providers import genius
 from lrc_dl.providers import youtube

lrc_dl/providers/genius.py

@@ -0,0 +1,62 @@
+# https://github.com/jeffvli/feishin/blob/development/src/main/features/core/lyrics/genius.ts
+import re
+from typing import Optional
+import httpx
+import bs4
+
+from lrc_dl.core import Song, AbstractProvider
+from lrc_dl.registry import lyrics_provider
+
+
+def _format_div(div: bs4.Tag):
+    for br in div.find_all('br'):
+        br.replace_with('\n') # type: ignore
+    text = div.get_text().strip()
+    text = re.sub(r"\[.*\]\n", '', text).strip()
+
+    # remove extra newlines
+    return '\n\n'.join(['\n'.join(x.split('\n')) for x in text.split('\n\n')])
+
+
+@lyrics_provider
+class Genius(AbstractProvider):
+    name = "genius"
+
+    def fetch_lyrics(self, song: Song) -> Optional[str]:
+        r = httpx.get('https://genius.com/api/search/song', params={
+            'per_page': 1,
+            'q': f'{song.artist} {song.title}'
+        })
+
+        if r.status_code != 200 or 'application/json' not in r.headers.get('content-type', ''):
+            return
+
+        hits = r.json().get('response', {}).get('sections', [{}])[0].get('hits')
+
+        if not hits:
+            return
+
+        url: str = hits[0].get('result', {}).get('url')
+
+        if not url:
+            return
+
+        r = httpx.get(url, headers={
+            'user-agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:135.0) Gecko/20100101 Firefox/135.0'
+        })
+
+        if r.status_code != 200:
+            return
+
+        soup = bs4.BeautifulSoup(r.text, features='html.parser')
+        div = soup.select_one('div.lyrics')
+
+        if div:
+            return _format_div(div)
+
+        div = soup.select_one('div[class^=Lyrics__Container]')
+
+        if not div:
+            return
+
+        return _format_div(div)

lrc_dl/providers/lrclib.py

@@ -0,0 +1,18 @@
+from typing import Optional
+import httpx
+
+from lrc_dl.core import Song, AbstractProvider
+from lrc_dl.registry import lyrics_provider
+
+@lyrics_provider
+class LrcLib(AbstractProvider):
+    name = "lrclib"
+
+    def fetch_lyrics(self, song: Song, with_album = True) -> Optional[str]:
+        r = httpx.get("https://lrclib.net/api/get", params={
+            'track_name': song.title,
+            'artist_name': song.artist,
+            'album_name': song.album if with_album else None,
+        }).json()
+
+        return r.get('syncedLyrics') or r.get('plainLyrics') or (self.fetch_lyrics(song, False) if with_album else None)

lrc_dl/registry.py

@@ -4,14 +4,14 @@ from lrc_dl.core import AbstractProvider
 class Registry:
     providers: dict[str, type[AbstractProvider]] = {}
 
-    @staticmethod
-    def get_synced_providers() -> dict[str, type[AbstractProvider]]:
+    @classmethod
+    def get_synced_providers(cls) -> dict[str, type[AbstractProvider]]:
         # TODO: stub
-        return dict(Registry.providers)
+        return dict(cls.providers)
 
-    @staticmethod
-    def register_provider(provider_class: type[AbstractProvider]) -> None:
-        Registry.providers[provider_class.name] = provider_class
+    @classmethod
+    def register_provider(cls, provider_class: type[AbstractProvider]) -> None:
+        cls.providers[provider_class.name] = provider_class
 
 
 def lyrics_provider(cls: type[AbstractProvider]) -> type[AbstractProvider]:

setup.py

@@ -15,6 +15,7 @@ setup(
     },
     install_requires=[
         "httpx>=0.24.1",
+        "beautifulsoup4>=4.13.3",
         "mutagen>=1.46.0",
         "yt-dlp>=2023.10.13",
     ]