AI Assistant for Pippy and chat integration

[AdityaShome]

Chages made:

This PR introduces AI-powered code analysis and chat capabilities to Pippy using Code Llama locally through Ollama integration. The implementation includes a dedicated server for AI interactions and a new UI panel for the chat interface.

Example

Child friendly:

Explaination:

Chat interface:

Technical Details

New Components

1. Code Llama Server

   • Integration by codellama_server.py with Ollama's API for model inference
   • CodeLlamaHandler class for request handling
   • check_ollama_health() method for service verification
   • make_ollama_request() method with retry logic
   • Support for both code analysis and chat endpoints

2. Code Llama Helper

   • Manages communication with the Code Llama server
   • CodeLlamaHelper Handles JSON request/response formatting
   • Implements retry logic for failed requests

3. Code Llama Pane (CodeLlamaPane class)

   • GTK-based UI component for the chat interface
   • Features:
     • Text entry field for user messages
     • Send button for message submission
     • Color-coded message display (user/AI)
     • Automatic scrolling to latest messages
     • Thread-safe message handling

Key Features

• Code Analysis

  • Automatic code analysis on file changes
  • Structured response format (functionality, improvements, issues)
  • Configurable response length and quality parameters according to requirements

• Chat Interface

  • Real-time chat with Code Llama
  • Message history display
  • Error handling and user feedback
  • Optimized response parameters for natural conversation

Performance Optimizations

• Response timeout: 120 seconds
• Max tokens: 100
• Context window: 2048
• CPU threads: 4
• Temperature: 0.1 (code analysis), 0.7 (chat)
• Retry mechanism with 3 attempts and 2-second delays

Error Handling

• Health checks for Ollama service
• Retry logic for failed requests
• Detailed error logging
• User-friendly error messages
• Timeout handling with informative feedback

Testing

The implementation has been tested with:

• Chat interactions
• Error conditions (logs, timeouts, service unavailability)
• UI responsiveness and thread safety

Dependencies

• Python 3.x
• GTK 3
• Ollama service with Code Llama model
• llama-cpp-python package

Setup

• The Code Llama model should be pulled and running in Ollama before using these features
• The server runs on port 8080 by default
• Memory usage and response times might vary based on system resources

Kindly suggest the modifications needed and should I also give gemini or openai api support as fallback or configurable mode.