CodePure: Smart Code Smell Detection extension on VS

CodePure is an advanced VS Code extension that detects and highlights code smells using extracted metrics. It leverages Tree-sitter for code extraction, processes the data into JSON, and applies iPlasma-inspired rules to identify problematic code structures.

🚀 Features

🔍 Code Smell Detection & Highlighting

CodePure analyzes your code and visually highlights problematic sections based on defined metrics:

• Brain Class
  • Highlights based on WMC > 87 (method heads) and LOC > 351 (class head).
  • Provides messages suggesting code improvements.
• Data Class
  • Detects classes with minimal behavior (low method complexity, high attribute usage).
• God Class
  • Flags excessively large classes with high complexity and responsibility.
• Schizo Class (Future Testing Needed)
  • Identifies classes with conflicting responsibilities.

🤖 Quick AI Fix

• Automatically suggests refactoring solutions for detected code smells.
• Helps developers quickly improve code quality with minimal manual effort.

✨ Advanced Highlighting System

• Real-time visual cues for detected code smells.
• Different colors and markers for various issues, ensuring quick identification.

📊 Metrics-Based Analysis

CodePure retrieves metrics from a precomputed JSON file instead of recalculating them, improving efficiency. Extracted metrics include:

• WMC (Weighted Methods per Class)
• LOC (Lines of Code)
• NrBm (Number of Brain Methods)
• CBO (Coupling Between Objects)
• NOA (Number of Attributes)

📐 Class Diagram Generation

• Generates a comprehensive class diagram for the entire project.
• Uses the extracted JSON data to visualize class structures and relationships.

🛠 Code Extraction & Parsing

• Uses Tree-sitter to extract code structure.
• Saves extracted data in JSON format for easy readability and further processing.
• Powers a Class Diagram Generator using the parsed JSON data.

🧠 Custom AI Model for Detection

• The AI model was trained by us specifically for CodePure.
• Optimized for detecting Brain Class, Data Class, God Class, and Schizo Class.
• Continuously improved based on real-world testing.

🧪 Accuracy Evaluation & Testing

CodePure was tested on medium-sized GitHub repositories:

• ✅ Data Class detection: Always accurate.
• ⚖ God Class detection: 50% accuracy.
• 🔬 Further testing required for Brain Class & Schizo Class.

📦 Installation

1. Download the extension from the VS Code Marketplace (link to be added).
2. Open VS Code and navigate to Extensions.
3. Search for CodePure and install.
4. Restart VS Code if necessary.

🚀 Usage

1. Open a project in VS Code.
2. CodePure automatically detects and highlights code smells.
3. Use Quick AI Fix to resolve issues faster.
4. Generate a Class Diagram for a better project overview.
5. Check the provided messages and refactor accordingly.

📌 Roadmap

• ✅ Current Features: Brain Class, Data Class, God Class detection.
• 🛠 In Progress: Schizo Class detection improvements.
• 🔍 Future: Enhanced visualization & customization options.

📞 Contact & Contributions

We welcome contributions! Feel free to:

• Submit issues on GitHub
• Suggest improvements & optimizations in our feedback in the extension itself
• Help refine detection accuracy with new test cases

---

Code smarter with CodePure! 🚀