Dewiride.Azure.AI.OpenAI.Helper 1.0.3

Azure OpenAI Helper

The Azure OpenAI Helper provides utility methods to interact with Azure OpenAI endpoints for both standard API requests and streaming responses. It includes robust retry logic and detailed logging for handling errors and timeouts.

Features

• Non-Streaming Responses: Send a request to Azure OpenAI and receive a deserialized response.
• Streaming Responses: Stream responses from Azure OpenAI, useful for large or dynamic outputs.
• Retry Logic: Automatically retries requests on failure with configurable retry count and delay.
• Logging: Logs errors and retry attempts for easier debugging.

Installation

1. Ensure your project references Newtonsoft.Json and Microsoft.Extensions.Logging.
2. Add this helper to your project as part of a service.

Methods

1. GetChatCompletionResponseAsync

Send a standard request to the Azure OpenAI API and receive a deserialized response.

Method Signature

public async Task<TResponse?> GetChatCompletionResponseAsync<TRequest, TResponse>(
    string apiEndpoint,
    string apiKey,
    TRequest dataRequest,
    int maxRetryAttempts = 10,
    int retryDelayMs = 1000
);

Parameters

• apiEndpoint (string): The URL of the Azure OpenAI endpoint.
• apiKey (string): Your Azure OpenAI API key.
• dataRequest (TRequest): The request payload.
• maxRetryAttempts (int): Maximum number of retries in case of failure (default: 10).
• retryDelayMs (int): Delay between retries in milliseconds (default: 1000).

Example Usage

var helper = new AzureOpenAiHelper(logger);

var request = new
{
    model = "gpt-4",
    messages = new[]
    {
        new { role = "system", content = "You are a helpful assistant." },
        new { role = "user", content = "Hello, how are you?" }
    }
};

var response = await helper.GetChatCompletionResponseAsync<object, OpenAiDataResponse>(
    apiEndpoint: "https://<your-endpoint>.openai.azure.com/openai/deployments/<deployment-id>/chat/completions?api-version=2024-01-01",
    apiKey: "<your-api-key>",
    dataRequest: request
);

if (response != null)
{
    Console.WriteLine($"Response: {response.Choices.FirstOrDefault()?.Message?.Content}");
}
else
{
    Console.WriteLine("Failed to retrieve a response.");
}

2. GetChatCompletionStreamedResponseAsync

Stream the response from Azure OpenAI, processing each chunk of data as it arrives.

Method Signature

public async Task GetChatCompletionStreamedResponseAsync<TRequest>(
    string azureOpenAiEndpoint,
    string azureOpenAiKey,
    TRequest request,
    Action<string> onMessageReceived,
    int maxRetryAttempts = 5,
    int retryDelayMs = 1000
);

Parameters

• azureOpenAiEndpoint (string): The URL of the Azure OpenAI streaming endpoint.
• azureOpenAiKey (string): Your Azure OpenAI API key.
• request (TRequest): The request payload.
• onMessageReceived (Action<string>): A callback to handle streamed content.
• maxRetryAttempts (int): Maximum number of retries in case of failure (default: 5).
• retryDelayMs (int): Delay between retries in milliseconds (default: 1000).

Example Usage

var helper = new AzureOpenAiHelper(logger);

var request = new
{
    model = "gpt-4",
    messages = new[]
    {
        new { role = "system", content = "You are a helpful assistant." },
        new { role = "user", content = "Tell me a story!" }
    },
    stream = true
};

await helper.GetChatCompletionStreamedResponseAsync(
    azureOpenAiEndpoint: "https://<your-endpoint>.openai.azure.com/openai/deployments/<deployment-id>/chat/completions?api-version=2024-01-01",
    azureOpenAiKey: "<your-api-key>",
    request: request,
    onMessageReceived: message =>
    {
        Console.WriteLine($"Streamed content: {message}");
    }
);

Logging

The AzureOpenAiHelper relies on ILogger<AzureOpenAiHelper> for logging. To use this feature, ensure you inject an appropriate logger instance into the helper.

Configuration

• Retry Configuration: Customize maxRetryAttempts and retryDelayMs to suit your application's needs.
• Timeout Handling: Both methods handle timeouts gracefully using retry logic and logging warnings.

Dependencies

• Newtonsoft.Json: For JSON serialization and deserialization.
• Microsoft.Extensions.Logging: For logging errors and warnings.

License

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

Company

• Dewiride Technologies Private Limited

Repository

• GitHub Repository