This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

# Build
go build ./cmd/vget

# Build to specific directory
go build -o build/vget ./cmd/vget

# Run directly
go run ./cmd/vget

# Build with version info (for releases)
go build -ldflags "-X github.com/guiyumin/vget/internal/version.Version=1.0.0" ./cmd/vget

vget is a media downloader CLI built with Go. It uses Cobra for command parsing and Bubbletea for interactive TUI elements (spinners, progress bars).

1. CLI Layer (internal/cli/) - Cobra commands parse flags and dispatch to handlers
2. Extractor Layer (internal/extractor/) - URL matching and media metadata extraction
3. Downloader Layer (internal/downloader/) - HTTP download with Bubbletea progress TUI

The MediaType enum in internal/extractor/extractor.go defines supported media types:

• MediaTypeVideo - Video files (Twitter, YouTube, etc.)
• MediaTypeAudio - Audio files (podcasts)
• MediaTypePDF - PDF documents
• MediaTypeEPUB - EPUB ebooks
• MediaTypeMOBI - MOBI ebooks
• MediaTypeAZW - AZW ebooks
• MediaTypeUnknown - Fallback (treated as video)

Each type has specific terminal output formatting in internal/cli/extract.go.

To add support for a new site, implement the Extractor interface in internal/extractor/:

type Extractor interface {
    Name() string
    Match(url string) bool
    Extract(url string) (*VideoInfo, error)
}

Set the appropriate MediaType in the returned VideoInfo:

return &VideoInfo{
    ID:        "...",
    Title:     "...",
    MediaType: MediaTypeAudio, // or MediaTypeVideo, etc.
    Formats:   []Format{...},
}, nil

Extractors are auto-registered via init() functions. See xiaoyuzhou.go or twitter.go for examples.

• vget <url> - Download media from URL
• vget init - Interactive config wizard (TUI)
• vget update - Self-update to latest version
• vget search --podcast <query> - Search Xiaoyuzhou podcasts
• vget ls <remote>:<path> - List WebDAV remote directory
• vget config show - Show current configuration
• vget config set <key> <value> - Set config value (non-interactive)
• vget config get <key> - Get config value
• vget config webdav ... - Manage WebDAV servers

Translations are embedded YAML files in internal/i18n/locales/. Supported: en, zh, jp, kr, es, fr, de.

Access translations via i18n.T(langCode) which returns a *Translations struct with typed fields.

User config lives in ~/.config/vget/config.yml. Two ways to configure:

1. Interactive (TUI): vget init - Bubbletea wizard for first-time setup
2. Non-interactive: vget config set <key> <value> - For scripting/Docker

Supported keys for vget config set:

• language - Language code (en, zh, jp, kr, es, fr, de)
• output_dir - Default download directory
• format - Preferred format (mp4, webm, best)
• quality - Default quality (1080p, 720p, best)
• twitter.auth_token - Twitter auth for NSFW content

IMPORTANT: Config is read fresh on every command execution (not cached at startup). This is intentional and MUST be preserved:

• Enables config changes without restart
• Critical for Docker UX (no container restart needed)
• Never change this behavior

The XHS extractor (internal/extractor/xiaohongshu.go) uses browser automation:

• Browser: Rod's auto-downloaded Chromium (NOT system Chrome)
• Binary location: ~/.cache/rod/browser/
• User data: ~/.config/vget/browser/ (persistent, shared by all extractors)
• Stealth: Uses go-rod/stealth for anti-bot detection

Important: Never use system Chrome profiles with browser automation - it can corrupt session data.

internal/updater/ uses go-selfupdate to fetch releases from GitHub (guiyumin/vget). Version is set in internal/version/version.go.

• BUILD OUTPUT DIRECTORY IS ./build
• YOU ONLY BUILD IT. I WILL TEST IT BY MYSELF.