- Implement ClaudeReader for parsing JSON exports from Claude
- Add claude_rag.py following BaseRAGExample pattern
- Support both concatenated conversations and individual messages
- Handle multiple JSON formats and structures
- Include comprehensive error handling and user guidance
- Add metadata extraction (titles, timestamps, roles)
- Integrate with existing LEANN chunking and embedding systems
Features:
✅ JSON parsing from Claude exports
✅ ZIP file extraction support
✅ Multiple JSON format support (list, single object, wrapped)
✅ Conversation detection and structuring
✅ Message role identification (user/assistant)
✅ Metadata extraction and preservation
✅ Dual processing modes (concatenated/separate)
✅ Command-line interface with all LEANN options
✅ Comprehensive error handling
✅ Multiple input format support (.json, .zip, directories)
Usage:
python -m apps.claude_rag --export-path claude_export.json
python -m apps.claude_rag --export-path claude_export.zip --query 'Python help'
2dd4147de2