k3nnyandClaude Opus 5.5
8145e5e985
Wrapper around yt-dlp to download whole arte.tv series into Series (year)/Season XX/Series - SxxEyy - Title.mkv: - resolve series / season / episode URLs through the Arte API - pick video, audio (VO, VF, AD...) and subtitle tracks from a TOML config - download with yt-dlp, convert WebVTT to SRT (ffmpeg 4.4 yields empty subtitles from Arte's CRLF files), remux with track languages, titles, default/forced flags and episode tags - optional TMDB matching: series name and year, numbering (including multi-episode files when Arte merges two episodes), localized titles and synopses following a language priority list Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
55 lines
1.9 KiB
Python
55 lines
1.9 KiB
Python
"""Minimal WebVTT -> SRT conversion.
|
||
|
||
Arte serves VTT files with CRLF line endings and STYLE blocks, which older
|
||
ffmpeg releases (e.g. 4.4, also used by yt-dlp --convert-subs) silently turn
|
||
into empty subtitles. Converting ourselves avoids depending on the ffmpeg version.
|
||
"""
|
||
from __future__ import annotations
|
||
|
||
import html
|
||
import re
|
||
|
||
_TIMING_RE = re.compile(
|
||
r'^\s*(?P<start>(?:\d+:)?\d{2}:\d{2}[.,]\d{3})\s+-->\s+(?P<end>(?:\d+:)?\d{2}:\d{2}[.,]\d{3})')
|
||
_KEEP_TAGS_RE = re.compile(r'</?(?:i|b|u)>')
|
||
_TAG_RE = re.compile(r'<[^>]*>')
|
||
|
||
|
||
def _timestamp(ts: str) -> str:
|
||
ts = ts.replace(',', '.')
|
||
parts = ts.split(':')
|
||
if len(parts) == 2:
|
||
parts.insert(0, '0')
|
||
h, m, s = parts
|
||
sec, ms = s.split('.')
|
||
return f'{int(h):02d}:{int(m):02d}:{int(sec):02d},{ms}'
|
||
|
||
|
||
def _clean(line: str) -> str:
|
||
# Keep <i>, <b>, <u>; drop <c.class>, <v Speaker>, <lang>, inline timestamps...
|
||
kept = []
|
||
|
||
def stash(m):
|
||
kept.append(m[0])
|
||
return f'\x00{len(kept) - 1}\x00'
|
||
|
||
line = _KEEP_TAGS_RE.sub(stash, line)
|
||
line = html.unescape(_TAG_RE.sub('', line)).replace(' ', ' ')
|
||
return re.sub(r'\x00(\d+)\x00', lambda m: kept[int(m[1])], line)
|
||
|
||
|
||
def vtt_to_srt(vtt: str) -> str:
|
||
text = vtt.lstrip('').replace('\r\n', '\n').replace('\r', '\n')
|
||
cues = []
|
||
for block in re.split(r'\n{2,}', text):
|
||
lines = block.strip('\n').split('\n')
|
||
timing_idx = next((i for i, l in enumerate(lines) if _TIMING_RE.match(l)), None)
|
||
if timing_idx is None: # header, STYLE, NOTE, REGION blocks
|
||
continue
|
||
m = _TIMING_RE.match(lines[timing_idx])
|
||
payload = [c for c in (_clean(l).strip() for l in lines[timing_idx + 1:]) if c]
|
||
if payload:
|
||
cues.append((_timestamp(m['start']), _timestamp(m['end']), payload))
|
||
return ''.join(f'{n}\n{start} --> {end}\n' + '\n'.join(payload) + '\n\n'
|
||
for n, (start, end, payload) in enumerate(cues, 1))
|