Rename module from lyrics-dl to lrc-dl

lrc_dl/providers/__init__.py

@@ -0,0 +1,3 @@
+from lrc_dl.providers import musixmatch
+from lrc_dl.providers import kugou
+from lrc_dl.providers import youtube

lrc_dl/providers/kugou.py

@@ -0,0 +1,79 @@
+from typing import Optional, Iterable
+from base64 import b64decode
+import zlib
+import re
+from itertools import filterfalse, islice
+from datetime import datetime
+
+import httpx
+
+from lrc_dl.core import Song, AbstractProvider
+from lrc_dl.registry import lyrics_provider
+
+
+KRC_ENCODE_KEY = [64, 71, 97, 119, 94, 50, 116, 71, 81, 54, 49, 45, 206, 210, 110, 105]
+
+RE_KRC_JUNK = re.compile(r"^\[((id|ar|ti|by|hash|al|sign|qq|total|language):|offset:0\]|.*\](<.*>)?(Written by：|Lyrics by：|Composed by：|Producer：|作曲 :|作词 :)).*$")
+RE_WORD_TIMING = re.compile(r"<-?\d+,-?\d+,-?\d+>")
+
+
+def decode_krc(content: bytes) -> str:
+    content = b64decode(content)
+
+    buf = bytearray(len(content) - 4)
+    for i in range(4, len(content)):
+        buf[i - 4] = content[i] ^ KRC_ENCODE_KEY[(i - 4) % 16]
+
+    return zlib.decompress(buf).decode('utf-8-sig')
+
+
+def reformat_timings(lines: Iterable[str]) -> Iterable[str]:
+    for line in lines:
+        if not line.startswith("["):
+            yield line
+            continue
+
+        raw_timings, text = line.split("]", 1)
+        beginning, _ = map(int, raw_timings[1:].split(","))
+
+        timing = datetime.fromtimestamp(beginning / 1000).strftime("%M:%S.%f")[:8]
+
+        yield f"[{timing}]{text}"
+
+
+@lyrics_provider
+class Kugou(AbstractProvider):
+    name = "kugou"
+
+    def fetch_lyrics(self: AbstractProvider, song: Song) -> Optional[str]:
+        keyword = f"{song.artist} - {song.title}"
+
+        response = httpx.get("https://krcs.kugou.com/search", params={
+            "ver": 1,
+            "man": "yes",
+            "client": "mobi",
+            "keyword": keyword
+        }).json()
+
+        candidates = response["candidates"]
+
+        if not candidates:
+            return None
+
+        id_, accesskey = candidates[0]["id"], candidates[0]["accesskey"]
+
+        r = httpx.get("https://krcs.kugou.com/download", params={
+            "ver": 1,
+            "man": "yes",
+            "client": "mobi",
+            "format": "lrc",
+            "id": id_,
+            "accesskey": accesskey
+        }).json()
+
+        krc = decode_krc(r["content"])
+
+        krc = RE_WORD_TIMING.sub("", krc)
+        lines = reformat_timings(islice(filterfalse(RE_KRC_JUNK.match, krc.splitlines()), 1, None))
+
+        return "\n".join(lines)

lrc_dl/providers/musixmatch.py

@@ -0,0 +1,36 @@
+from typing import Optional
+import httpx
+
+from lrc_dl.core import Song, AbstractProvider
+from lrc_dl.registry import lyrics_provider
+
+
+@lyrics_provider
+class Musixmatch(AbstractProvider):
+    name = "musixmatch"
+
+    def __init__(self, token: str) -> None:
+        self.token = token
+
+    def fetch_lyrics(self, song: Song) -> Optional[str]:
+        response = httpx.get("https://apic-desktop.musixmatch.com/ws/1.1/macro.subtitles.get", params={
+            "format": "json",
+            "namespace": "lyrics_synched",
+            "part": "lyrics_crowd,user,lyrics_verified_by",
+            "user_language": "en",
+            "f_subtitle_length_max_deviation": 1,
+            "subtitle_format": "lrc",
+            "app_id": "web-desktop-app-v1.0",
+            "usertoken": self.token,
+
+            "q_artist": song.artist,
+            "q_track": song.title,
+            "q_album": song.album,
+        }, follow_redirects=True).json()
+
+        response = response["message"]["body"]["macro_calls"]["track.subtitles.get"]["message"]["body"]
+
+        if not response:
+            return None
+
+        return response["subtitle_list"][0]["subtitle"]["subtitle_body"]

lrc_dl/providers/youtube.py

@@ -0,0 +1,79 @@
+from typing import Optional, Dict
+from contextlib import redirect_stdout
+import subprocess
+import io
+import urllib
+import unittest.mock
+
+from yt_dlp import YoutubeDL
+
+from lrc_dl.core import Song, AbstractProvider
+from lrc_dl.registry import lyrics_provider
+from lrc_dl import utils
+
+
+@lyrics_provider
+class Youtube(AbstractProvider):
+    name = "youtube"
+
+    def _craft_search_link(self, song: Song) -> str:
+        query = f"{song.artist} - {song.title}"
+        query = urllib.parse.quote(query)
+        # sp=... means search only videos with subtitles
+        url = f"https://www.youtube.com/results?search_query={query}&sp=EgIoAQ%253D%253D"
+        return url
+
+    def _download_subtitles(self, video_id: str) -> str:
+        # buffer = io.BytesIO()
+        buffer = io.StringIO()
+
+        # A dirty monkey patch; youtube-dl does not
+        # support "-" filename for subtitles, so we
+        # just force it to use it here.
+        with unittest.mock.patch("yt_dlp.YoutubeDL.subtitles_filename", new=lambda *_: "-"):
+            with redirect_stdout(buffer):
+                with YoutubeDL({"writesubtitles": True, "skip_download": True, "subtitlesformat": "srt/vtt/best", 'logtostderr': True}) as ydl:
+                    ydl.download(video_id)
+
+        return buffer.getvalue()
+
+    def _subtitles_to_lyrics(self, subtitles: str) -> str:
+        # "-fflags +bitexact" prevents ffmpeg from
+        # writing metadata to .lrc file
+        # TODO: use `with` statement
+        process = subprocess.Popen(["ffmpeg", "-loglevel", "quiet", "-i", "-", "-f", "lrc", "-fflags", "+bitexact", "-"], stdin=subprocess.PIPE, stdout=subprocess.PIPE)
+
+        if not process.stdin or not process.stdout:
+            return ""
+
+        process.stdin.write(subtitles.encode())
+        process.stdin.close()
+        process.wait()
+        return process.stdout.read().decode()[1:]
+
+    def fetch_lyrics(self, song: Song) -> Optional[str]:
+        search_link = self._craft_search_link(song)
+        with YoutubeDL({"extract_flat": True, "playlistend": 10}) as ydl:
+            videos = ydl.extract_info(search_link)["entries"]
+
+        if song.duration:
+            def match_duration(video: Dict) -> bool:
+                return utils.threshold_equal(video["duration"], song.duration, 2)
+            videos = filter(match_duration, videos)
+
+        def match_title(video: Dict) -> bool:
+            return True
+        videos = filter(match_title, videos)
+
+        video = utils.next_or_none(videos)
+
+        if not video:
+            return None
+
+        subtitles = self._download_subtitles(video["id"])
+        lyrics = self._subtitles_to_lyrics(subtitles)
+
+        if lyrics != "":
+            return lyrics
+
+        return None