Codebase-Analysis-using-LLM

This project provides an automatic code analysis for Java-based repositories. It performs the following key tasks:

1. Clone a GitHub repository

   • If the repository is not already cloned, it will clone the repository into a specified directory.

2. Read and split Java code

   • Loads Java files from the repository and splits them into smaller chunks to handle token limits.

3. Build a vector store

   • Uses FAISS (Facebook AI Similarity Search) to build a vector store from the code chunks for efficient semantic search.

4. Query the vector store

   • Accepts a user query to search the vector store and retrieves the most relevant code chunks.

5. Analyze code with LLM

   • Sends the top-matching code chunks to Google's Generative AI model and extracts structured information in JSON format.

6. Save the output

   • Saves the structured summary into a JSON file.

---

Setup Instructions

1. Clone the repository

First, clone this project to your local machine:

git clone https://github.com/your-repository/Code-Analysis-Project.git
cd Code-Analysis-Project

---

2. Install Dependencies

Ensure you have Python 3.8 or higher installed.
Then install the required dependencies:

pip install -r requirements.txt

Dependencies include:

• langchain
• langchain-google-genai
• sentence-transformers
• faiss-cpu
• python-dotenv
• numpy

---

3. Set Up Environment Variables

You will need a Google API key to access Gemini models.

Create a .env file in the root directory and add:

GOOGLE_API_KEY=your_google_api_key

Replace your_google_api_key with your actual API key.

---

4. Run the Project

Run the following command:

python main.py

This will perform the following:

• Clone the Sakila repository (if not already cloned).
• Load and split Java files.
• Build the FAISS vector store.
• Accept a user query.
• Analyze the matching code using Gemini LLM.
• Save the output to a JSON file.

---

Example Query

When you run the project, you will see:

What would you like to know about the SakilaProject?

Example queries you can ask:

• "What are the key classes in the SakilaProject?"
• "What does the method getCustomer do?"
• "Explain the database connection part."

---

Project Structure

File Name	Purpose
clone_repo.py	Contains the function to clone a GitHub repository.
load_split.py	Loads Java files and splits code into manageable chunks.
buildvector_faiss.py	Builds the FAISS vector store and handles semantic search operations.
analyze_code.py	Sends code to LLM and processes the analysis response.
save_output.py	Saves the final JSON output.
main.py	Main entry point integrating all functionalities.

---

License

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

---