MagicFridgeAI 🧠❄️

A Java/Spring Boot backend application that registers food items from the user's "fridge" and uses AI (GPT‑4O) to generate creative and personalized recipes based on the available ingredients.

---

🧰 Features

• 💾 CRUD for food items — Create, read, update, and delete ingredients.
• 🍲 AI-based recipe generation — Sends ingredients to ChatGPT and receives a detailed recipe.
• ☁️ Reactive webhook — Uses Spring WebFlux and WebClient for non-blocking calls to OpenAI.
• 📦 Persistence layer — H2 database (or PostgreSQL in production) with JPA mapping.
• 🛡️ Schema migration — Managed via Flyway.
• 📄 API documentation — Documented with Swagger/OpenAPI.

---

🛠️ Technologies

• Java 21 + Spring Boot
• Spring Data JPA, H2 / PostgreSQL
• Spring WebFlux
• Jackson, Lombok
• Flyway
• OpenAI API (gpt‑4o)
• Swagger

---

🚀 Getting Started

1. Clone the repository:

   git clone https://github.com/viniciusferreira7/magic-fridge-ai.git
   cd magic-fridge-ai

2. Set environment variables:

   export OPENAI_API_KEY=your_openai_key_here
   # Optional: PostgreSQL DB
   export SPRING_DATASOURCE_URL=jdbc:postgresql://localhost:5432/your_db

3. Run the application:

   ./mvnw spring-boot:run

4. Access API and Swagger docs:

   • Base URL: http://localhost:8080
   • Swagger UI: http://localhost:8080/swagger-ui.html

---

🧩 Main Endpoints

🍽️ Food

Method	Route	Description
POST	/food/save	Add a new food item
GET	/food/list	List all food items
GET	/food/list/{id}	Get food item by ID
PUT	/food/update	Update a food item
DELETE	/food/delete/{id}	Delete a food item by ID

🧑‍🍳 Recipe

Method	Route	Description
GET	/recipe/generate	Generate a recipe and image using available ingredients

---

⚙️ AI Integration Details

1. Formats the ingredients like:

   - Apple (FRUIT), 5 units, valid until 2025‑07‑15
   

2. Sends the formatted string as a prompt to the OpenAI API using WebClient.

3. Receives a response like:

   {
     "choices": [
       { "message": { "content": "Recipe steps..." } }
     ]
   }

4. Extracts choices[0].message.content and returns it as a Mono<String>.

---

📈 Possible Improvements

• Complete the DALL·E image generation integration.
• Add a web or mobile frontend.
• Use WebSocket or Server-Sent Events (SSE) for real-time streaming.
• Suggest complementary ingredients for better recipes.
• Improve validation and error handling.

---

📄 License

MIT License © 2025 Vinicius Ferreira

---

🔗 References

This project is inspired by similar open-source AI recipe generators on GitHub, including:

• Pontinn/magic-fridge-ai
• ViniciusVentura7/MagicFridgeAI

---

✨ Contributions welcome!

Feel free to open issues, suggest features, or submit pull requests to improve the project.