juanra/audiobook-forge v2.3.0

v2.3.0: Interactive Metadata Matching

on GitHub

🎯 Interactive Metadata Matching (BEETS-Inspired)

This release adds a powerful interactive metadata matching system inspired by BEETS, allowing you to search, score, and interactively select the best Audible metadata match for your M4B files.

✨ Highlights

• Interactive Matching Command - New audiobook-forge match command for intelligent metadata selection
• Fuzzy Scoring System - Weighted distance algorithm (Title 40%, Author 30%, Duration 20%, Year 10%)
• Visual Confidence Levels - Color-coded matches (🟢 >96%, 🟡 88-96%, 🔴 80-88%)
• Rich Terminal UI - Beautiful interactive prompts using inquire crate
• Batch Processing - Process multiple M4B files with progress tracking
• Smart Extraction - Embedded tags + filename parsing with automatic fallback

🚀 New Features

Interactive Matching

# Interactive mode (default)
audiobook-forge match --file "Book.m4b"

# Batch processing
audiobook-forge match --dir /path/to/audiobooks

# Auto mode (non-interactive)
audiobook-forge match --dir /path --auto

# Dry run
audiobook-forge match --dir /path --dry-run

Metadata Extraction

• Embedded M4B tags (mp4ameta) - primary source
• Filename parsing - "Author - Title.m4b" pattern
• Manual override via --title and --author flags
• Automatic merge strategies

Interactive UI Features

• Numbered candidate selection (1-10)
• Before/after metadata comparison
• Action menu: Skip, Manual Entry, Custom Search
• Confirmation dialog showing all changes
• Manual metadata entry form
• Custom search with new terms

📊 Match Scoring

Distance Calculation:

• Title: 40% weight (normalized Levenshtein distance)
• Author: 30% weight (best match from multiple authors)
• Duration: 20% weight (5% tolerance, >20% penalty)
• Year: 10% weight (±10 years tolerance)

Confidence Levels:

• 🟢 Strong: >96% match (distance < 0.04)
• 🟡 Medium: 88-96% match (distance < 0.12)
• 🔴 Low: 80-88% match (distance < 0.20)
• ⚪ None: <80% match

⚙️ Configuration

New match_mode setting in config.yaml:

metadata:
  match_mode: disabled  # Options: disabled, auto, interactive

• disabled - Don't match during build (default)
• auto - Automatically select best match
• interactive - Prompt user for selection

📦 Dependencies

• inquire 0.6.2 - Interactive terminal UI
• strsim 0.10.0 - String similarity (normalized Levenshtein)

📈 Code Statistics

• 7 new files (~1,200 lines of code)
• 11 files modified
• Zero compilation errors
• Full async/await integration

🔧 Technical Details

Files Added:

• src/models/match_models.rs - Match data structures
• src/utils/scoring.rs - Scoring engine
• src/utils/extraction.rs - Metadata extraction
• src/ui/mod.rs - UI module
• src/ui/prompts.rs - Interactive prompts

Architecture:

File/Directory → Metadata Extraction → Audible Search →
Scoring & Ranking → Interactive Selection → Metadata Application

📝 Usage Examples

Interactive Matching:

audiobook-forge match --file "Book.m4b"

Batch Processing:

audiobook-forge match --dir ~/audiobooks --region us

Auto Mode:

audiobook-forge match --dir ~/audiobooks --auto

Manual Override:

audiobook-forge match --file "Book.m4b" \
  --title "Project Hail Mary" \
  --author "Andy Weir"

⚠️ Known Limitations

• Audnexus API dependency (requires internet connection)
• API rate limiting (100 requests/minute)
• M4B files only (MP3 support planned)

🔮 Future Enhancements

• Build integration with match_mode config
• Support for MP3 files
• Custom scoring weights configuration
• Alternative metadata sources

📚 Documentation

• See CHANGELOG.md for full details
• See README.md for command reference

Full Changelog: v2.2.0...v2.3.0