spring-ai-examples/kotlin/kotlin-function-callback/README.md

Spring AI Kotlin Function Callback Demo

This project demonstrates the implementation of Function Callbacks in Spring AI using Kotlin, specifically showing how to integrate OpenAI's API to handle weather-related queries.

Overview

The application showcases how to:

• Set up Spring AI with Kotlin
• Implement Function Callbacks for weather information
• Use mock weather service for demonstration
• Handle OpenAI API integration through Spring AI

Prerequisites

• JDK 17 or higher
• Kotlin
• Spring Boot
• Spring AI dependencies
• OpenAI API key (configured in your application properties)

Key Components

1. Main Application Class

The KotlinFunctionCallbackApplication class initializes the Spring Boot application and sets up the chat client to interact with OpenAI's API. It includes a simple demo that queries weather conditions for San Francisco, Tokyo, and Paris.

2. Configuration

The Config class sets up the necessary beans for function callbacks:

• Configures the weather function information
• Sets up the mock weather service
• Defines the function callback structure

3. Weather Service

The MockKotlinWeatherService provides a simple mock implementation that returns predefined weather data for different cities:

4. Data Models

• WeatherRequest: Represents the weather query parameters

  • location (city name)
  • latitude
  • longitude
  • temperature unit (C/F)

• WeatherResponse: Contains weather information

  • current temperature
  • feels like temperature
  • minimum temperature
  • maximum temperature
  • pressure
  • humidity
  • unit (C/F)

Usage

1. Add your OpenAI API key to application.properties:

spring.ai.openai.api-key=your-api-key-here

2. Run the application:

mvn spring-boot:run

The application will execute a sample weather query for San Francisco, Tokyo, and Paris, demonstrating how function callbacks work with Spring AI.

Example Output

Response: The current weather conditions in the three cities are:

San Francisco: 30°C
Tokyo: 10°C
Paris: 15°C

Implementation Details

1. The application uses Spring AI's ChatClient to interact with OpenAI's API
2. Function callbacks are registered through the FunctionCallback builder
3. Weather queries are handled by the mock service for demonstration purposes
4. The application uses Kotlin data classes for type-safe request/response handling

Customization

To modify the mock weather service:

1. Update the temperature values in MockKotlinWeatherService
2. Add new city conditions as needed
3. Implement actual weather API integration by replacing the mock service