Skip to content

Latest commit

 

History

History
187 lines (148 loc) · 5.46 KB

README.md

File metadata and controls

187 lines (148 loc) · 5.46 KB

Video Explorer

A web application that uses Kaltura and AWS Bedrock (Claude 3) to analyze videos and enable interactive conversations with their content. The application provides video analysis, topic extraction, and AI-powered chat capabilities.

Features

  • Video Management

    • Search Kaltura videos by category or text
    • Browse recent videos with thumbnails and descriptions
    • Multi-video selection for batch analysis

  • AI Analysis

    • Deep content analysis using AWS Bedrock (Claude 3)
    • Topic extraction and importance scoring
    • Key moment detection with timestamps
    • Comprehensive video summaries

  • Interactive Interface

    • Clean, responsive design using PicoCSS
    • Tabbed analysis view (Summary, Insights, Topics, Key Moments)
    • Interactive topic visualization
    • Video segment preview with thumbnails
    • Real-time chat with video content
    • Dark theme support

Prerequisites

  • Python 3.9 or higher
  • Kaltura account with API access
  • AWS account with Bedrock access (Claude 3 model)

Quick Start

  1. Clone the repository:
git clone [your-repo-url]
cd video-explorer
  1. Create a virtual environment and install dependencies:
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
  1. Set up configuration:
cp .env.example .env

Edit .env with your credentials:

  • Get Kaltura credentials from your Kaltura Management Console
  • Get AWS credentials with Bedrock access from AWS Console
  1. Run the application:
python main.py
  1. Open http://localhost:8000 in your browser

Configuration

The .env file supports the following configuration options:

# Kaltura Configuration
KALTURA_PARTNER_ID=your_partner_id
KALTURA_SECRET=your_secret_key
KALTURA_SERVICE_URL=https://cdnapisec.kaltura.com
KALTURA_SESSION_DURATION=86400

# AWS Configuration
AWS_ACCESS_KEY_ID=your_aws_access_key
AWS_SECRET_ACCESS_KEY=your_aws_secret_key
AWS_REGION=us-east-1  # Region where Bedrock is available

# Server Configuration
HOST=0.0.0.0
PORT=8000

# Analysis Configuration
MODEL_ID=anthropic.claude-3-sonnet-20240229-v1:0
MODEL_TIMEOUT=60
MODEL_MAX_TOKENS=4000
MODEL_CHUNK_SIZE=24000
MODEL_TEMPERATURE=0
PAGE_SIZE=10

# Logging Configuration
LOG_LEVEL=INFO  # Main application log level (DEBUG, INFO, WARNING, ERROR, CRITICAL)
KALTURA_LOG_LEVEL=WARNING  # Kaltura client log level
AWS_LOG_LEVEL=WARNING  # AWS/boto3 log level
HTTP_LOG_LEVEL=WARNING  # HTTP client log level
LITELLM_LOG_LEVEL=WARNING  # LiteLLM log level

Architecture

Backend

  • Python: Core application runtime
  • Kaltura API: Video content management and delivery
  • AWS Bedrock: AI analysis using Claude 3 model
  • litellm: LLM integration layer
  • instructor: Structured outputs from LLM responses

Frontend

  • HTML/JavaScript: Pure JavaScript for interactivity
  • PicoCSS: Minimal, semantic CSS framework
  • Responsive Design: Mobile-first approach
  • Dynamic UI: Real-time updates and animations

Logging System

The application uses a robust, configurable logging system with the following features:

  • Color-coded log levels: Different colors for DEBUG, INFO, WARNING, ERROR, and CRITICAL logs
  • Timestamps: Precise timestamps for each log entry
  • Source location: File names and line numbers for easy debugging
  • Configurable levels: Separate log levels for main application and third-party libraries
  • Structured format: Consistent log format across all components
  • Third-party integration: Controlled logging for Kaltura, AWS, HTTP, and LiteLLM

Example log output:

2024-03-20 10:15:30.123 [INFO] main:125 - Starting video analysis...
2024-03-20 10:15:30.456 [DEBUG] kaltura_utils:89 - Successfully retrieved video metadata
2024-03-20 10:15:31.789 [WARNING] main:256 - Processing delay detected
2024-03-20 10:15:32.012 [ERROR] main:789 - Failed to process video segment

Key Features

  1. Parallel processing of video analysis
  2. Intelligent chunking for long videos
  3. In-memory caching of analysis results
  4. Real-time progress tracking
  5. Structured AI responses for consistent output
  6. Comprehensive logging system

API Endpoints

  • GET /: Main application interface
  • GET /api/videos: Search and list videos
  • POST /api/analyze: Analyze selected videos
  • GET /api/analysis-progress/{task_id}: Check analysis progress
  • POST /api/chat: Chat with analyzed video content

Development

Local Development

# Run with auto-reload
python main.py

Code Structure

video-explorer/
├── main.py           # Application logic and API endpoints
├── kaltura_utils.py  # Kaltura integration utilities
├── logger_config.py  # Logging configuration
├── static/
│   └── style.css    # Application styling
├── templates/
│   └── index.html   # Frontend interface
└── requirements.txt  # Python dependencies

Browser Support

  • Chrome/Edge (latest 2 versions)
  • Firefox (latest 2 versions)
  • Safari (latest 2 versions)
  • Mobile browsers (iOS Safari, Android Chrome)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

  1. Fork the repository
  2. Create your feature branch (git checkout -b feature/AmazingFeature)
  3. Commit your changes (git commit -m 'Add some AmazingFeature')
  4. Push to the branch (git push origin feature/AmazingFeature)
  5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.