From 3f30a6b78ed331f843c8a6b6420371476a36b479 Mon Sep 17 00:00:00 2001
From: pukkandan <pukkandan.ytdlp@gmail.com>
Date: Tue, 27 Feb 2024 08:42:44 +0530
Subject: [PATCH] Abstract out lrc parsing

---
 yt_dlp/postprocessor/ffmpeg.py | 21 +++++---------------
 yt_dlp/utils/subtitles.py      | 36 ++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 16 deletions(-)
 create mode 100644 yt_dlp/utils/subtitles.py

diff --git a/yt_dlp/postprocessor/ffmpeg.py b/yt_dlp/postprocessor/ffmpeg.py
index d97930655..59ca879cf 100644
--- a/yt_dlp/postprocessor/ffmpeg.py
+++ b/yt_dlp/postprocessor/ffmpeg.py
@@ -33,6 +33,7 @@
     variadic,
     write_json_file,
 )
+from ..utils.subtitles import Subtitle, parse_lrc
 
 EXT_TO_OUT_FORMATS = {
     'aac': 'adts',
@@ -683,27 +684,15 @@ def _embed_lyrics(self, subtitles, filename, ext):
                 with open(sub['filepath'], encoding='utf-8') as f:
                     sub['data'] = f.read()
 
-        def totime(time):
-            time = time.split(":")
-            return int((int(time[0])*60 + float(time[1]))*1000)
-        def convert_lrc_to_sylt(lrc):
-            lrc = lrc.split("\n")
-            lrc = [i.strip() for i in lrc]
-            lrc = [i for i in lrc if i]
-            lrc = [i for i in lrc if i[-1] != "]"]
-            lrc = [i.split("]") for i in lrc]
-            lrc = [[i[0][1:], i[1]] for i in lrc]
-            lrc = [(i[1], totime(i[0])) for i in lrc]
-            return lrc
         if ext == 'mp3':
             metadata = mutagen.id3.ID3(filename)
             for lang, sub in subtitles.items():
                 metadata.add(mutagen.id3.SYLT(
-                    encoding=mutagen.id3.Encoding.UTF8,
+                    encoding=mutagen.id3.Encoding.UTF8, format=2, type=1,
                     lang=ISO639Utils.short2long(lang) or 'und',
-                    format=2,
-                    type=1,
-                    text=convert_lrc_to_sylt(sub['data'])))
+                    text=[(line.text, int(line.start * 1000))
+                          for line in parse_lrc(sub['data'])
+                          if isinstance(line, Subtitle)]))
         else:
             metadata = mutagen.File(filename)
             metadata['©lyr' if ext == 'm4a' else 'lyrics'] = [sub['data'] for sub in subtitles.values()]
diff --git a/yt_dlp/utils/subtitles.py b/yt_dlp/utils/subtitles.py
new file mode 100644
index 000000000..784acecde
--- /dev/null
+++ b/yt_dlp/utils/subtitles.py
@@ -0,0 +1,36 @@
+import re
+from dataclasses import dataclass
+from typing import TypeAlias
+
+Seconds: TypeAlias = float
+
+
+@dataclass
+class Metadata:
+    name: str
+    value: str
+
+
+@dataclass
+class Subtitle:
+    text: str
+    start: Seconds
+    end: Seconds = None
+
+
+def parse_lrc(text):
+    for line in text.split('\n'):
+        times = []
+        while mobj := re.fullmatch(r'\[(?P<time>((\d+:)?\d+:)?\d+(.\d+)?)\](?P<content>.*)', line):
+            times.append(sum(
+                float(t) * 60**i for i, t in enumerate(reversed(mobj.group('time').split(':')))))
+            line = mobj.group('content')
+
+        for t in times:
+            yield Subtitle(start=t, text=line.strip())
+
+        if not times:
+            if mobj := re.fullmatch(r'\[(?P<name>[^\]:]+):(?P<value>[^\]]+)\]', line):
+                yield Metadata(mobj.group('name'), mobj.group('value').strip())
+            elif line.strip():
+                yield ValueError(line)