liyading1818/obsidian-llm-wiki / .cursor/skills/video2wiki
video2wiki
VerifiedTurn a video link (Xiaohongshu / Bilibili / Douyin / YouTube) or an existing noisy web-clipping into a clean structured note that is ingested into the user's Obsidian LLM-Wiki vault (repository root). Use when the user invokes /video2wiki, drops a video URL or a clippings/ file and asks to 转成笔记 / 整理 / 沉淀 / 归档 / ingest, or asks to backfill the clippings folder. Downloads audio with yt-dlp, transcribes locally with faster-whisper, writes a structured source page, then runs the Karpathy-style wiki ingest. Adapted from the lark-video2note pattern but lands in Obsidian instead of Lark.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.0 KB |
| scripts/batch.py | script | 10.1 KB |
| scripts/fetch.py | script | 6.1 KB |
| scripts/transcribe.py | script | 4.5 KB |
scripts/batch.py
View on GitHub#!/usr/bin/env python3
"""batch.py — 批量把 clippings/ 里的视频剪藏跑完"下载 + 转写"。
只做机械的、耗时的前两步(fetch + transcribe),产出逐字稿到 raw/transcripts/,
并把每条的元信息写进 manifest.json,供 LLM 后续逐篇撰写成品笔记 + ingest。
特性:
- 断点续跑:已有同 source_url 逐字稿的剪藏自动跳过。
- 容错:单条失败(非视频/下载失败/转写失败)记录状态后继续下一条。
- 省空间:每条转写成功后删除临时 mp4(逐字稿已留底)。
用法(在仓库根目录运行):
python .cursor/skills/video2wiki/scripts/batch.py --model small
"""
from __future__ import annotations
import argparse
import datetime as dt
import json
import re
import subprocess
import sys
from pathlib import Path
SCRIPT_DIR = Path(__file__).resolve().parent
def find_root(start: Path) -> Path:
for p in [start, *start.parents]:
if (p / "AGENTS.md").is_file():
return p
sys.exit("ERROR: 未找到 AGENTS.md")
ROOT = find_root(Path.cwd())
# Windows 控制台默认 GBK;yt-dlp 日志也常是系统编码,subprocess 需与之对齐
_SUBPROC_ENC = "gbk" if sys.platform == "win32" else "utf-8"
def log(msg: str) -> None:
ts = dt.datetime.now().strftime("%H:%M:%S")
line = f"[{ts}] {msg}"
try:
print(line, flush=True)
except UnicodeEncodeError:
enc = sys.stdout.encoding or "utf-8"
print(line.encode(enc, errors="replace").decode(enc, errors="replace"), flush=True)
def rel_path(p: Path | str) -> str:
"""相对仓库根的路径(POSIX),避免 Windows subprocess 传中文绝对路径乱码。"""
pp = Path(p).resolve()
root = ROOT.resolve()
try:
return pp.relative_to(root).as_posix()
except ValueError:
ps, rs = pp.as_posix(), root.as_posix()
if ps.lower().startswith(rs.lower() + "/"):
return ps[len(rs) + 1 :]
return pp.as_posix()
def read(path: Path) -> str:
return path.read_text(encoding="utf-8", errors="replace")
def fm_field(text: str, key: str) -> str | None:
m = re.search(rf'^{key}:\s*["\']?(\S.*?)["\']?\s*$', text, re.MULTILINE)
return m.group(1).strip() if m else None
def extract_source(text: str) -> str | None:
s = fm_field(text, "source")
if s and s.startswith("http"):
return s
m = re.search(r"https?://[^\s\"')]+", text)
return m.group(0) if m else None
def slugify(name: str) -> str:
name = re.sub(r"[\s!!??~~,,。、::·\|/\\<>\"'*]+", "", name)
return name[:40] or "untitled"
def slug_from_clip(clip: Path, src: str, created: str) -> str:
"""工作目录 / 逐字稿 slug:优先用 URL 里的笔记 ID(纯 ASCII),避免 Windows subprocess 中文路径乱码。"""
m = re.search(r"/explore/([0-9a-f]+)", src, re.I)
if m:
return f"{created}-xhs-{m.group(1)}"
return f"{created}-{slugify(clip.stem)}"
def existing_source_urls(transcripts_dir: Path) -> set[str]:
urls = set()
if transcripts_dir.is_dir():
for p in transcripts_dir.glob("*.md"):
u = fm_field(read(p), "source_url")
if u:
# 去掉 query 串后比对,避免 xsec_token 变化导致重复
urls.add(u.split("?")[0])
return urls
def main() -> int:
ap = argparse.ArgumentParser()
ap.add_argument("--clippings-dir", default=str(ROOT / "clippings"))
ap.add_argument("--transcripts-dir", default=str(ROOT / "raw" / "transcripts"))
ap.add_argument("--work-dir", default=str(ROOT / "output" / ".video_work"))
ap.add_argument("--manifest", default=str(ROOT / "output" / ".video_work" / "manifest.json"))
ap.add_argument("--model", default="small")
ap.add_argument("--lang", default="zh")
ap.add_argument("--limit", type=int, default=0, help="最多处理几条(0=全部)")
args = ap.parse_args()
clip_dir = Path(args.clippings_dir)
trans_dir = Path(args.transcripts_dir)
work_dir = Path(args.work_dir)
trans_dir.mkdir(parents=True, exist_ok=True)
work_dir.mkdir(parents=True, exist_ok=True)
manifest_path = Path(args.manifest)
manifest = {}
if manifest_path.is_file():
try:
manifest = json.loads(read(manifest_path))
except Exception: # noqa: BLE001
manifest = {}
done_urls = existing_source_urls(trans_dir)
clippings = sorted(clip_dir.glob("*.md"))
log(f"共 {len(clippings)} 篇剪藏,已有逐字稿 {len(done_urls)} 篇")
processed = 0
for clip in clippings:
text = read(clip)
src = extract_source(text)
if not src:
log(f"跳过(无 source):{clip.name}")
continue
src_key = src.split("?")[0]
if src_key in done_urls:
log(f"跳过(已转写):{clip.name}")
continue
if args.limit and processed >= args.limit:
log(f"达到 limit={args.limit},停止")
break
title = fm_field(text, "title") or clip.stem
created = fm_field(text, "created") or dt.date.today().isoformat()
slug = slug_from_clip(clip, src, created)
work = work_dir / slug
transcript = trans_dir / f"{slug}.md"
log(f"处理:{clip.name}")
entry = {"clipping": clip.name, "title": title, "source_url": src,
"slug": slug, "status": "", "transcript": "", "error": ""}
# 1. fetch
try:
r = subprocess.run(
[sys.executable, rel_path(SCRIPT_DIR / "fetch.py"),
"--clipping", rel_path(clip), "--out", rel_path(work)],
cwd=str(ROOT),
capture_output=True, text=True,
encoding=_SUBPROC_ENC, errors="replace", timeout=900,
)
except subprocess.TimeoutExpired:
entry["status"] = "fetch_timeout"
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(" 下载超时,跳过")
continue
if r.returncode == 2:
entry["status"] = "not_video"
entry["error"] = (r.stderr or "").strip()[:300]
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(" 非视频(图文),跳过,留待文本 ingest")
continue
if r.returncode != 0:
entry["status"] = "fetch_failed"
entry["error"] = (r.stderr or "").strip()[:300]
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(f" 下载失败:{entry['error'][:120]}")
continue
try:
# stdout 末行应为 JSON;stderr 可能混入 progress,取最后一条以 { 开头的行
out_lines = [ln.strip() for ln in (r.stdout or "").splitlines() if ln.strip()]
json_line = next((ln for ln in reversed(out_lines) if ln.startswith("{")), "")
meta = json.loads(json_line)
except Exception as e: # noqa: BLE001
entry["status"] = "fetch_badjson"
entry["error"] = str(e)
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(" 下载输出解析失败")
continue
dur = meta.get("duration_s", 0)
log(f" 已下载 {meta.get('size_bytes',0)//1024//1024}MB, 时长 {dur//60}:{dur%60:02d},开始转写……")
# 音频路径:优先从 work 目录取实际文件,避免 fetch JSON 里绝对路径在 Windows 上解析失败
audio_candidates = [p for p in work.iterdir() if p.is_file() and p.stat().st_size > 10000]
if audio_candidates:
audio_arg = rel_path(max(audio_candidates, key=lambda p: p.stat().st_size))
else:
audio_arg = rel_path(meta["file"])
# 2. transcribe
try:
r2 = subprocess.run(
[sys.executable, rel_path(SCRIPT_DIR / "transcribe.py"),
"--audio", audio_arg, "--out", rel_path(transcript),
"--model", args.model, "--lang", args.lang,
"--title", title, "--source-url", meta.get("source_url", src),
"--platform", meta.get("platform", ""), "--author", meta.get("author", "")],
cwd=str(ROOT),
capture_output=True, text=True,
encoding=_SUBPROC_ENC, errors="replace", timeout=14400,
)
except subprocess.TimeoutExpired:
entry["status"] = "transcribe_timeout"
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(" 转写超时,跳过")
continue
if r2.returncode != 0:
entry["status"] = "transcribe_failed"
entry["error"] = (r2.stderr or "").strip()[-300:]
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
log(f" 转写失败:{entry['error'][:120]}")
continue
entry["status"] = "transcribed"
entry["transcript"] = str(transcript)
entry["platform"] = meta.get("platform", "")
entry["duration_s"] = dur
manifest[slug] = entry
manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
done_urls.add(src_key)
processed += 1
# 3. 清理临时 mp4
try:
for f in work.glob("*"):
f.unlink()
work.rmdir()
except Exception: # noqa: BLE001
pass
log(f" ✓ 完成 → {transcript.name}")
# 汇总
by_status = {}
for e in manifest.values():
by_status[e["status"]] = by_status.get(e["status"], 0) + 1
log("批处理结束。状态汇总:" + json.dumps(by_status, ensure_ascii=False))
log(f"manifest: {manifest_path}")
return 0
if __name__ == "__main__":
sys.exit(main())
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |