C++ Unit Test Generator

Overview

Automatically generate, run, and evaluate Google Test unit tests for your C++ codebase using Github Models

Use these models for best results

• https://github.com/marketplace/models/azure-openai/gpt-4-1
• https://github.com/marketplace/models/azure-openai/gpt-4-1-mini
• https://github.com/marketplace/models/azure-openai/gpt-4-1-nano

---

Features

• Scans and parses your C++ codebase
• Uses a Github Hosted LLM to generate unit tests
• Builds the project using CMake
• Resolves build issues using LLM
• Compiles and runs Google Test test cases
• Measures test coverage with gcov
• Removes failing test cases automatically via LLM feedback

---

Project Structure

unit-test-generator
├── codebase    => C++ source files to test
├── tests       => Generated test files (Automatically generated)
├── build       => CMake build output (Automatically Generated)
├── src         => Node.js test generator logic

---

Requirements

C++ & Build

• g++
• CMake >= 3.10
• Google Test gtest

Node.js

• Node.js >= 16
• js-yaml, glob, dotenv, etc. (run npm install)

---

Installation and Setup

1. Create a folder and go inside it

mkdir unit-test-generator
cd unit-test-generator

2. Make codebase directory

mkdir codebase

3. Copy your C++ project into codebase folder

mv <path to your c++ project> ./codebase

4. Clone the repository as src:

git clone https://github.com/HarshitBamotra/cpp-unit-test-generator.git src

5. Change directory

cd src

6. Install Dependencies

npm install

7. Set up environment variables

Sample .env file

ENDPOINT = "https://models.github.ai/inference"
MODEL_NAME = "openai/gpt-4.1-mini"
LLM_ACCESS_TOKEN = <your github PAT token> (make sure to provide read access to models while generating PAT token)

8. Start the Application

npm start

---

Screenshots

Test Generation

Building Project And Fixing Build Errors

Removing Failing Test Cases

Test Results

Test Report

---

TODO

1. Make app run for generalized cpp project path (currently only works for ../codebase)
2. Make this into a CLI
3. Add more language generators