抖音无水印视频下载和文案提取

抖音无水印视频下载和文案提取工具. 从抖音分享链接获取无水印视频下载链接, 下载视频, 提取视频中的语音文案并自动保存到文件. 适用场景包括获取抖音视频信息, 下载无水印视频, 批量提取视频文案. 当用户需要处理抖音视频链接或提取视频内容时触发.

Security Scan

The skill's code and instructions are consistent with a Douyin (TikTok China) video downloader + audio transcription tool — it only needs network access to Douyin pages, FFmpeg, and an external transcription API key — but there are minor inconsistencies and one truncated file view that reduce confidence.

Purpose & Capability

Name/description match what the code and SKILL.md do: parse Douyin share URLs, download video, extract audio, and call a transcription API. No unrelated credentials, binaries, or installs are requested.

Instruction Scope

SKILL.md instructs the agent to parse links, download video, extract audio, and call an external transcription API. The runtime instructions do not ask for unrelated filesystem or credential access. They describe only the files and temp dirs the script actually uses.

Install Mechanism

No install spec is provided (instruction-only install). Dependencies are standard Python libraries (requests, ffmpeg-python) and FFmpeg system binary; installation is via pip / package manager as documented. No suspicious external download or archive extraction is used.

Credentials

Only an API key for the transcription service is needed, which is proportionate. However, SKILL.md and top-of-file comments refer to API_KEY, while one code path raises an error asking for DOUYIN_API_KEY — this environment-variable naming inconsistency is a bug and should be fixed/clarified before use.

Persistence & Privilege

The skill does not request always:true, does not modify other skills or system config, and is not persistent. It runs on demand and cleans up temp files.

Assessment

This skill appears to do what it says: download Douyin videos (no watermark) and transcribe their audio via the SiliconFlow transcription API. Before installing, check: 1) You must have FFmpeg installed and python packages (requests, ffmpeg-python). 2) Provide a valid transcription API key (SKILL.md says API_KEY; code contains at least one message referencing DOUYIN_API_KEY) — confirm which environment variable the script actually reads and set that one. 3) Confirm trustworthiness and pricing/quota of https://api.siliconflow.cn/ and do not expose or reuse that API key elsewhere. 4) Legal: ensure you have rights to download/transcribe any videos you process. 5) Because the provided file view in this evaluation was truncated, review the full scripts/douyin_downloader.py file locally to ensure there are no hidden network endpoints or unexpected behaviors before running.

从抖音分享链接获取无水印视频下载链接, 下载视频, 并使用语音识别提取视频中的文案, 自动保存到文件.

功能概述

• 获取下载链接: 从抖音分享链接解析出无水印视频的直接下载地址 (无需 API 密钥)
• 下载视频: 将无水印视频下载到本地指定目录
• 提取文案: 通过语音识别从视频中提取文字内容 (需要硅基流动 API 密钥)
• 自动保存: 每个视频的文案自动保存到独立文件夹 (视频ID为文件夹名)

环境要求

依赖安装

pip install requests ffmpeg-python

系统要求

• FFmpeg 必须安装在系统中 (用于音视频处理)
• macOS: brew install ffmpeg
• Ubuntu: apt install ffmpeg

API 密钥配置 (仅文案提取需要)

文案提取功能使用硅基流动 API, 需要设置环境变量:

export API_KEY="your-siliconflow-api-key"

获取 API 密钥: https://cloud.siliconflow.cn/

使用方法

方法一: 使用脚本 (推荐)

# 获取视频信息和下载链接 (无需 API 密钥)
python douyin_downloader.py --link "抖音分享链接" --action info

# 下载视频到指定目录
python douyin_downloader.py --link "抖音分享链接" --action download --output ./videos

# 提取视频文案并保存到文件 (需要 API_KEY 环境变量)
python douyin_downloader.py --link "抖音分享链接" --action extract --output ./output

# 提取文案并同时保存视频
python douyin_downloader.py --link "抖音分享链接" --action extract --output ./output --save-video

# 安静模式 (减少输出)
python douyin_downloader.py --link "抖音分享链接" --action extract --output ./output --quiet

输出目录结构

提取文案后, 每个视频会保存到独立文件夹:

output/
├── 7600361826030865707/      # 视频ID为文件夹名
│   └── transcript.md         # Markdown 格式文案文件
├── 7581044356631612699/
│   ├── transcript.md
│   └── 7581044356631612699.mp4  # 使用 --save-video 时保存
└── ...

Markdown 文案格式

# 视频标题

| 属性 | 值 |
|------|-----|
| 视频ID | `7600361826030865707` |
| 提取时间 | 2026-01-30 14:19:00 |
| 下载链接 | [点击下载](url) |

---

## 文案内容

(语音识别的文字内容)

方法二: 在 Python 代码中调用

from douyin_downloader import get_video_info, download_video, extract_text

# 获取视频信息
info = get_video_info("抖音分享链接")
print(f"视频ID: {info['video_id']}")
print(f"标题: {info['title']}")
print(f"下载链接: {info['url']}")

# 下载视频
video_path = download_video("抖音分享链接", output_dir="./videos")

# 提取文案并保存到文件
result = extract_text("抖音分享链接", output_dir="./output")
print(f"文案已保存到: {result['output_path']}")
print(result['text'])

工作流程

获取视频信息

1. 解析抖音分享链接, 提取真实的视频 URL
2. 模拟移动端请求获取页面数据
3. 从页面 JSON 数据中提取无水印视频地址
4. 返回视频 ID, 标题和下载链接

提取视频文案

1. 解析分享链接获取视频信息
2. 下载视频到临时目录
3. 使用 FFmpeg 从视频中提取音频 (MP3 格式)
4. 调用硅基流动 SenseVoice API 进行语音识别
5. 清理临时文件, 返回识别的文本

常见问题

无法解析链接

• 确保链接是有效的抖音分享链接
• 链接格式通常为 https://v.douyin.com/xxxxx/ 或完整的抖音视频 URL

提取文案失败

• 检查 API_KEY 环境变量是否已设置
• 确保 API 密钥有效且有足够的配额
• 确保 FFmpeg 已正确安装

下载速度慢

• 这取决于网络条件和视频大小
• 脚本会显示下载进度

注意事项

• 本工具仅供学习和研究使用
• 使用时需遵守相关法律法规
• 请勿用于任何侵犯版权或违法的目的