doubao-capture（豆宝抓拍）

Doubao Capture

Usage

python scripts/capture_doubao_scroll_v2.py --hwnd <window_handle> --output-dir <dir> --stop-auto --json-output

Parameters

• window_handle (required): HWND from doubao-launch
• output_dir (optional): Output directory, default: "works/translations"
• stop_auto (optional): Auto-detect end, default: true
• no_new_threshold (optional): Consecutive empty reads threshold, default: 5

Returns

{
  "success": true,
  "text_file_path": "H:/works/translations/doubao_20240307_143022.txt",
  "line_count": 156,
  "char_count": 3847,
  "stopped_by": "auto_detect"
}

Tools

capture_translation

Capture translated subtitles from Doubao

Workflow Integration

This skill is part of the YouTube translation workflow:

1. youtube-audio-download: Download audio from YouTube
2. doubao-launch: Launch Doubao translation window
3. audio-play: Play the downloaded audio
4. doubao-capture: Capture translated subtitles

Execution

All skills execute on Windows Python via WSL cross-platform call:

wsl -> python.exe scripts/capture_doubao_scroll_v2.py ...

Error Handling

All skills return JSON with success field:

• success: true - Operation completed
• success: false - Check error_code and error_message

Notes

• Windows GUI automation requires visible desktop (no RDP disconnect)
• Output files are stored in Windows works/ directory
• WSL accesses Windows files via /mnt/h/...