// Copyright (c) Microsoft. All rights reserved. using System.ComponentModel; using Microsoft.SemanticKernel; using Microsoft.SemanticKernel.ChatCompletion; using Microsoft.SemanticKernel.Connectors.Google; using xRetry; namespace FunctionCalling; /// /// These examples demonstrate two ways functions called by the Gemini LLM can be invoked using the SK streaming and non-streaming AI API: /// /// 1. Automatic Invocation by SK (with and without nullable properties): /// Functions called by the LLM are invoked automatically by SK. The results of these function invocations /// are automatically added to the chat history and returned to the LLM. The LLM reasons about the chat history /// and generates the final response. /// This approach is fully automated and requires no manual intervention from the caller. /// /// 2. Manual Invocation by a Caller: /// Functions called by the LLM are returned to the AI API caller. The caller controls the invocation phase where /// they may decide which function to call, when to call them, how to handle exceptions, call them in parallel or sequentially, etc. /// The caller then adds the function results or exceptions to the chat history and returns it to the LLM, which reasons about it /// and generates the final response. /// This approach is manual and provides more control over the function invocation phase to the caller. /// public sealed class Gemini_FunctionCalling(ITestOutputHelper output) : BaseTest(output) { [RetryFact] public async Task GoogleAIChatCompletionWithFunctionCalling() { Console.WriteLine("============= Google AI - Gemini Chat Completion with function calling ============="); Assert.NotNull(TestConfiguration.GoogleAI.ApiKey); Assert.NotNull(TestConfiguration.GoogleAI.Gemini.ModelId); Kernel kernel = Kernel.CreateBuilder() .AddGoogleAIGeminiChatCompletion( modelId: TestConfiguration.GoogleAI.Gemini.ModelId, apiKey: TestConfiguration.GoogleAI.ApiKey) .Build(); await this.RunSampleAsync(kernel); } [RetryFact] public async Task VertexAIChatCompletionWithFunctionCalling() { Console.WriteLine("============= Vertex AI - Gemini Chat Completion with function calling ============="); Assert.NotNull(TestConfiguration.VertexAI.BearerKey); Assert.NotNull(TestConfiguration.VertexAI.Location); Assert.NotNull(TestConfiguration.VertexAI.ProjectId); Assert.NotNull(TestConfiguration.VertexAI.Gemini.ModelId); Kernel kernel = Kernel.CreateBuilder() .AddVertexAIGeminiChatCompletion( modelId: TestConfiguration.VertexAI.Gemini.ModelId, bearerKey: TestConfiguration.VertexAI.BearerKey, location: TestConfiguration.VertexAI.Location, projectId: TestConfiguration.VertexAI.ProjectId) .Build(); // To generate bearer key, you need installed google sdk or use Google web console with command: // // gcloud auth print-access-token // // Above code pass bearer key as string, it is not recommended way in production code, // especially if IChatCompletionService will be long-lived, tokens generated by google sdk lives for 1 hour. // You should use bearer key provider, which will be used to generate token on demand: // // Example: // // Kernel kernel = Kernel.CreateBuilder() // .AddVertexAIGeminiChatCompletion( // modelId: TestConfiguration.VertexAI.Gemini.ModelId, // bearerKeyProvider: () => // { // // This is just example, in production we recommend using Google SDK to generate your BearerKey token. // // This delegate will be called on every request, // // when providing the token consider using caching strategy and refresh token logic when it is expired or close to expiration. // return GetBearerKey(); // }, // location: TestConfiguration.VertexAI.Location, // projectId: TestConfiguration.VertexAI.ProjectId); await this.RunSampleAsync(kernel); } [RetryFact] public async Task GoogleAIFunctionCallingNullable() { Console.WriteLine("============= Google AI - Gemini Chat Completion with function calling (nullable properties) ============="); Assert.NotNull(TestConfiguration.GoogleAI.ApiKey); var kernelBuilder = Kernel.CreateBuilder() .AddGoogleAIGeminiChatCompletion( modelId: TestConfiguration.VertexAI.Gemini.ModelId, apiKey: TestConfiguration.GoogleAI.ApiKey); kernelBuilder.Plugins.AddFromType(); var promptExecutionSettings = new GeminiPromptExecutionSettings() { FunctionChoiceBehavior = FunctionChoiceBehavior.Auto(), }; var kernel = kernelBuilder.Build(); var response = await kernel.InvokePromptAsync("Hi, what's the weather in New York?", new(promptExecutionSettings)); Console.WriteLine(response.ToString()); } private sealed class MyWeatherPlugin { [KernelFunction] [Description("Get the weather for a given location.")] private string GetWeather(WeatherRequest request) { return $"The weather in {request?.Location} is sunny."; } } [RetryFact] public async Task VertexAIFunctionCallingNullable() { Console.WriteLine("============= Vertex AI - Gemini Chat Completion with function calling (nullable properties) ============="); Assert.NotNull(TestConfiguration.VertexAI.BearerKey); Assert.NotNull(TestConfiguration.VertexAI.Location); Assert.NotNull(TestConfiguration.VertexAI.ProjectId); var kernelBuilder = Kernel.CreateBuilder() .AddVertexAIGeminiChatCompletion( modelId: TestConfiguration.VertexAI.Gemini.ModelId, bearerKey: TestConfiguration.VertexAI.BearerKey, location: TestConfiguration.VertexAI.Location, projectId: TestConfiguration.VertexAI.ProjectId); // To generate bearer key, you need installed google sdk or use Google web console with command: // // gcloud auth print-access-token // // Above code pass bearer key as string, it is not recommended way in production code, // especially if IChatCompletionService will be long-lived, tokens generated by google sdk lives for 1 hour. // You should use bearer key provider, which will be used to generate token on demand: // // Example: // // Kernel kernel = Kernel.CreateBuilder() // .AddVertexAIGeminiChatCompletion( // modelId: TestConfiguration.VertexAI.Gemini.ModelId, // bearerKeyProvider: () => // { // // This is just example, in production we recommend using Google SDK to generate your BearerKey token. // // This delegate will be called on every request, // // when providing the token consider using caching strategy and refresh token logic when it is expired or close to expiration. // return GetBearerKey(); // }, // location: TestConfiguration.VertexAI.Location, // projectId: TestConfiguration.VertexAI.ProjectId); kernelBuilder.Plugins.AddFromType(); var promptExecutionSettings = new GeminiPromptExecutionSettings() { FunctionChoiceBehavior = FunctionChoiceBehavior.Auto(), }; var kernel = kernelBuilder.Build(); var response = await kernel.InvokePromptAsync("Hi, what's the weather in New York?", new(promptExecutionSettings)); Console.WriteLine(response.ToString()); } private async Task RunSampleAsync(Kernel kernel) { // Add a plugin with some helper functions we want to allow the model to utilize. kernel.ImportPluginFromFunctions("HelperFunctions", [ kernel.CreateFunctionFromMethod(() => DateTime.UtcNow.ToString("R"), "GetCurrentUtcTime", "Retrieves the current time in UTC."), kernel.CreateFunctionFromMethod((string cityName) => cityName switch { "Boston" => "61 and rainy", "London" => "55 and cloudy", "Miami" => "80 and sunny", "Paris" => "60 and rainy", "Tokyo" => "50 and sunny", "Sydney" => "75 and sunny", "Tel Aviv" => "80 and sunny", _ => "31 and snowing", }, "Get_Weather_For_City", "Gets the current weather for the specified city"), ]); Console.WriteLine("======== Example 1: Use automated function calling with a non-streaming prompt ========"); { GeminiPromptExecutionSettings settings = new() { ToolCallBehavior = GeminiToolCallBehavior.AutoInvokeKernelFunctions }; Console.WriteLine(await kernel.InvokePromptAsync( "Check current UTC time, and return current weather in Paris city", new(settings))); Console.WriteLine(); } Console.WriteLine("======== Example 2: Use automated function calling with a streaming prompt ========"); { GeminiPromptExecutionSettings settings = new() { ToolCallBehavior = GeminiToolCallBehavior.AutoInvokeKernelFunctions }; await foreach (var update in kernel.InvokePromptStreamingAsync( "Check current UTC time, and return current weather in Boston city", new(settings))) { Console.Write(update); } Console.WriteLine(); } Console.WriteLine("======== Example 3: Use manual function calling with a non-streaming prompt ========"); { var chat = kernel.GetRequiredService(); var chatHistory = new ChatHistory(); GeminiPromptExecutionSettings settings = new() { ToolCallBehavior = GeminiToolCallBehavior.EnableKernelFunctions }; chatHistory.AddUserMessage("Check current UTC time, and return current weather in London city"); while (true) { var result = (GeminiChatMessageContent)await chat.GetChatMessageContentAsync(chatHistory, settings, kernel); if (result.Content is not null) { Console.Write(result.Content); } if (result.ToolCalls is not { Count: > 0 }) { break; } chatHistory.Add(result); foreach (var toolCall in result.ToolCalls) { KernelArguments? arguments = null; if (kernel.Plugins.TryGetFunction(toolCall.PluginName, toolCall.FunctionName, out var function)) { // Add parameters to arguments if (toolCall.Arguments is not null) { arguments = []; foreach (var parameter in toolCall.Arguments) { arguments[parameter.Key] = parameter.Value?.ToString(); } } } else { Console.WriteLine("Unable to find function. Please try again!"); continue; } var functionResponse = await function.InvokeAsync(kernel, arguments); Assert.NotNull(functionResponse); var calledToolResult = new GeminiFunctionToolResult(toolCall, functionResponse); chatHistory.Add(new GeminiChatMessageContent(calledToolResult)); } } Console.WriteLine(); } /* Uncomment this to try in a console chat loop. Console.WriteLine("======== Example 4: Use automated function calling with a streaming chat ========"); { GeminiPromptExecutionSettings settings = new() { ToolCallBehavior = ToolCallBehavior.AutoInvokeKernelFunctions }; var chat = kernel.GetRequiredService(); var chatHistory = new ChatHistory(); while (true) { Console.Write("Question (Type \"quit\" to leave): "); string question = Console.ReadLine() ?? string.Empty; if (question == "quit") { break; } chatHistory.AddUserMessage(question); System.Text.StringBuilder sb = new(); await foreach (var update in chat.GetStreamingChatMessageContentsAsync(chatHistory, settings, kernel)) { if (update.Content is not null) { Console.Write(update.Content); sb.Append(update.Content); } } chatHistory.AddAssistantMessage(sb.ToString()); Console.WriteLine(); } } */ } private sealed class WeatherRequest { public string? Location { get; set; } } }