MCP Video Parser

# Changelog All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). ## [0.1.1] - 2025-06-12 ### Added - **Video cleanup utility** (`clean_videos.py`) - Complete system reset functionality - Remove all videos from database and file system - Optional backup of database before cleaning - Dry-run mode for safe preview of deletions - Selective cleanup (keep originals vs clean everything) - Interactive confirmation with override options - Detailed statistics on cleanup results - **Comprehensive unit tests** for video and audio processing - Video processing functionality tests with mocked dependencies - Audio transcription tests with Whisper integration - Integration tests for `process_new_video.py` script - Mock-based testing for various failure scenarios - **Enhanced help documentation** in MCP HTTP client - Organized help into clear categories (transcript, visual, combined analysis) - Specific query examples for different use cases - Pro tips about system capabilities - Updated welcome message highlighting transcript and visual features ### Fixed - **Video listing issue** - "list all videos" now shows all videos correctly - Fixed date parser to handle "recent" queries with 7-day window instead of 24 hours - Added `time_field` parameter to choose between `recording_timestamp` and `created_at` - Modified queries to use `created_at` for "recent/latest" to show recently processed videos - Root cause: New videos had old recording timestamps but recent creation dates - **ProcessingResult attribute error** in `process_new_video.py` - Fixed AttributeError where script tried to access non-existent 'summary' attribute - Changed to display transcript preview instead of summary - Verified video and audio processing works correctly with sample videos ### Enhanced - **Audio processing documentation** - Clarified that complete transcripts are preserved - System stores full transcripts (1000+ words per video) in database - Search functionality works across complete transcript content - Different query types access transcript vs visual analysis appropriately - **Date parsing improvements** - Better handling of natural language time queries - "Recent/recently/latest" now properly returns last 7 days of content - Enhanced query routing based on user intent (recent processing vs recording time) - **Utility scripts documentation** - Added comprehensive section to README - Video cleanup script usage and examples - Video processing script documentation - Clear instructions for different cleanup scenarios ### Technical Improvements - **Database query optimization** - Smart time field selection for better results - **Test coverage expansion** - Comprehensive test suite for core functionality - **Error handling improvements** - Better error messages and graceful failures - **Code organization** - Cleaner separation of concerns in video processing pipeline ## [0.1.0] - 2024-06-13 ### Added - Initial release of MCP Video Parser - Video processing with frame extraction and analysis - MCP server implementation with 8 tools - Natural language query support - Location-based video organization - Time-based video queries - Audio transcription with Whisper - Scene detection for intelligent frame extraction - Chat integration with video context awareness - HTTP-based MCP transport for remote access - Comprehensive test suite - Documentation and examples ### Features - `process_video` - Process and analyze video files - `query_location_time` - Query videos by location and time - `search_videos` - Search video content and transcripts - `get_video_summary` - Generate AI summaries - `ask_video` - Ask questions about specific videos - `analyze_moment` - Analyze specific timestamps - `get_video_stats` - System statistics - `get_video_guide` - Usage instructions ### Technical - FastMCP framework integration - Ollama LLM integration (Llava for vision, Mistral/Llama for chat) - SQLite storage with location-based file organization - Configurable frame extraction rates - Session management for HTTP connections - Proper MCP protocol handshake implementation ### Known Issues - Limited natural language time parsing (see INTELLIGENT_QUERY_PLAN.md) - Basic pattern matching for query routing - No real-time video processing support yet [0.1.1]: https://github.com/michaelbaker-dev/mcpVideoParser/releases/tag/v0.1.1 [0.1.0]: https://github.com/michaelbaker-dev/mcpVideoParser/releases/tag/v0.1.0