用 OpenAI SDK 访问大语言模型

在 https://github.com/openai/openai-dotnet 下载 OpenAI 的 dotnet 实现源码，编译生成 OpenAI.dll。

创建 dotnet 8 的 Console App 项目，将 OpenAI.dll 复制到项目目录，用 JetBrains Rider 打开项目，在项目 “Dependencies” 上右击，选 “Reference” 添加对 OpenAI.dll 的引用（Add Reference），程序代码如下：

using OpenAI;
using OpenAI.Chat;
using iText.Kernel.Pdf;
using iText.Kernel.Pdf.Canvas.Parser;
using iText.Kernel.Pdf.Canvas.Parser.Listener;
using System;
using System.Text;
using Xceed.Document.NET;
using Xceed.Words.NET;

// this app shows how to use OpenAI to translate a PDF file from English to Chinese
// Add OpenAI.dll build from https://github.com/openai/openai-dotnet OpenAI.csproj and reference to this project
// you should add the following NuGet packages to your project:
// - itext7
// - Xceed.Words.NET
// - Xceed.Document.NET

var endpoint = new Uri("https://models.inference.ai.azure.com");
var credential = System.Environment.GetEnvironmentVariable("GITHUB_TOKEN");
var model = "gpt-4o-mini";

var openAIOptions = new OpenAIClientOptions()
{
    Endpoint=endpoint
};

var pdfPath = "./main.pdf";
var pdfText = ExtractTextFromPDF(pdfPath);
int chunkSize = 4000;
List<string> chunks = SplitStringIntoChunks(pdfText, chunkSize);

foreach (var chunk in chunks)
{
    Thread.Sleep(60000); // 60 seconds = 60,000 milliseconds
    var client = new ChatClient(model, credential, openAIOptions);
    List<ChatMessage> messages = new List<ChatMessage>()
    {
        new SystemChatMessage("You are a helpful assistant."),
        new UserChatMessage("将英文翻译为中文："+chunk),
    };

    var requestOptions = new ChatCompletionOptions()
    {
        Temperature = 1,
        MaxTokens = 4096,
    };

    var response = client.CompleteChat(messages, requestOptions);
    String responseText = response.Value.Content[0].Text;
    System.Console.WriteLine(responseText);

}
// get text from pdf doc
static string ExtractTextFromPDF(string pdfPath)
{
    using (PdfReader reader = new PdfReader(pdfPath))
    using (PdfDocument pdfDoc = new PdfDocument(reader))
    {
        StringBuilder text = new StringBuilder();
        int numberOfPages = pdfDoc.GetNumberOfPages();
            
        for (int i = 1; i <= numberOfPages; i++)
        {
            ITextExtractionStrategy strategy = new SimpleTextExtractionStrategy();
            string pageText = PdfTextExtractor.GetTextFromPage(pdfDoc.GetPage(i), strategy);
            text.Append(pageText);
        }

        return text.ToString();
    }
}
// split a long string into List of substrings for LLM model inference
List<string> SplitStringIntoChunks(string str, int chunkSize)
{
    List<string> chunks = new List<string>();
    for (int i = 0; i < str.Length; i += chunkSize)
    {
        if (i + chunkSize > str.Length)
        {
            chunks.Add(str.Substring(i));
        }
        else
        {
            chunks.Add(str.Substring(i, chunkSize));
        }
    }
    return chunks;
}