如何使用 OpenAI for PDF

查克尼特·宾

2024年一月22日

更新 2024年十二月11日

Translated

View the article in English

OpenAI是一个人工智能研究实验室，由营利性的OpenAI LP和其非营利性的母公司OpenAI Inc组成。它的成立目标是以一种造福全人类的方式推进数字智能的发展。 OpenAI在不同领域进行人工智能（AI）研究，旨在开发安全、有益且可及的AI技术。

IronPdf.Extensions.AI NuGet 包现已支持将 OpenAI 应用于 PDF 增强：摘要、查询和记忆。该软件包利用 Microsoft Semantic Kernel。

开始使用IronPDF

立即在您的项目中开始使用IronPDF，并享受免费试用。

第一步：

如何使用 OpenAI for PDF

下载 C# 库以使用 OpenAI 处理 PDF
为 OpenAI 准备 Azure 端点和 API 密钥
导入目标 PDF 文档
使用Summarize方法生成PDF的摘要
使用Query方法进行连续查询

立即在您的项目中开始使用IronPDF，并享受免费试用。

第一步：

除了IronPdf软件包之外，您还需要以下两个软件包：

总结PDF示例

要使用 OpenAI 功能，需要一个 Azure 端点和一个 API 密钥。根据下面的代码示例配置语义内核。导入 PDF 文档并利用 Summarize 方法生成 PDF 文档的摘要。您可以从OpenAI PDF摘要示例下载示例PDF文件。

请注意

您将遇到 SKEXP0001、SKEXP0010 和 SKEXP0050 错误。这是因为语义内核方法是试验性的。在您的 .csproj 文件中添加以下代码，以消除这些错误。

<PropertyGroup&gt；

<NoWarn>$(NoWarn);SKEXP0001,SKEXP0010,SKEXP0050</NoWarn>

</PropertyGroup&gt；

:path=/static-assets/pdf/content-code-examples/how-to/openai-summarize.cs

using IronPdf;
using IronPdf.AI;
using Microsoft.SemanticKernel;
using Microsoft.SemanticKernel.Connectors.OpenAI;
using Microsoft.SemanticKernel.Memory;
using System;
using System.Threading.Tasks;

// Setup OpenAI
var azureEndpoint = "<<enter your azure endpoint here>>";
var apiKey = "<<enter your azure API key here>>";
var builder = Kernel.CreateBuilder()
    .AddAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey)
    .AddAzureOpenAIChatCompletion("oaichat", azureEndpoint, apiKey);
var kernel = builder.Build();

// Setup Memory
var memory_builder = new MemoryBuilder()
    // optionally use new ChromaMemoryStore("http://127.0.0.1:8000") (see https://github.com/microsoft/semantic-kernel/blob/main/dotnet/notebooks/09-memory-with-chroma.ipynb)
    .WithMemoryStore(new VolatileMemoryStore())
    .WithAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey);
var memory = memory_builder.Build();

// Initialize IronAI
IronDocumentAI.Initialize(kernel, memory);

License.LicenseKey = "<<enter your IronPdf license key here";

// Import PDF document
PdfDocument pdf = PdfDocument.FromFile("wikipedia.pdf");

// Summarize the document
Console.WriteLine("Please wait while I summarize the document...");
string summary = await pdf.Summarize(); // optionally pass AI instance or use AI instance directly
Console.WriteLine($"Document summary: {summary}\n\n");

Imports Microsoft.VisualBasic
Imports IronPdf
Imports IronPdf.AI
Imports Microsoft.SemanticKernel
Imports Microsoft.SemanticKernel.Connectors.OpenAI
Imports Microsoft.SemanticKernel.Memory
Imports System
Imports System.Threading.Tasks

' Setup OpenAI
Private azureEndpoint = "<<enter your azure endpoint here>>"
Private apiKey = "<<enter your azure API key here>>"
Private builder = Kernel.CreateBuilder().AddAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey).AddAzureOpenAIChatCompletion("oaichat", azureEndpoint, apiKey)
Private kernel = builder.Build()

' Setup Memory
Private memory_builder = (New MemoryBuilder()).WithMemoryStore(New VolatileMemoryStore()).WithAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey)
Private memory = memory_builder.Build()

' Initialize IronAI
IronDocumentAI.Initialize(kernel, memory)

License.LicenseKey = "<<enter your IronPdf license key here"

' Import PDF document
Dim pdf As PdfDocument = PdfDocument.FromFile("wikipedia.pdf")

' Summarize the document
Console.WriteLine("Please wait while I summarize the document...")
Dim summary As String = Await pdf.Summarize() ' optionally pass AI instance or use AI instance directly
Console.WriteLine($"Document summary: {summary}" & vbLf & vbLf)

$vbLabelText $csharpLabel

产出摘要

连续查询示例

一种查询可能不适用于所有场景。 IronPdf.Extensions.AI 包还提供了一种查询方法，允许用户执行连续查询。

:path=/static-assets/pdf/content-code-examples/how-to/openai-query.cs

using IronPdf;
using IronPdf.AI;
using Microsoft.SemanticKernel;
using Microsoft.SemanticKernel.Connectors.OpenAI;
using Microsoft.SemanticKernel.Memory;
using System;
using System.Threading.Tasks;

// Setup OpenAI
var azureEndpoint = "<<enter your azure endpoint here>>";
var apiKey = "<<enter your azure API key here>>";
var builder = Kernel.CreateBuilder()
    .AddAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey)
    .AddAzureOpenAIChatCompletion("oaichat", azureEndpoint, apiKey);
var kernel = builder.Build();

// Setup Memory
var memory_builder = new MemoryBuilder()
    // optionally use new ChromaMemoryStore("http://127.0.0.1:8000") (see https://github.com/microsoft/semantic-kernel/blob/main/dotnet/notebooks/09-memory-with-chroma.ipynb)
    .WithMemoryStore(new VolatileMemoryStore())
    .WithAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey);
var memory = memory_builder.Build();

// Initialize IronAI
IronDocumentAI.Initialize(kernel, memory);

License.LicenseKey = "<<enter your IronPdf license key here";

// Import PDF document
PdfDocument pdf = PdfDocument.FromFile("wikipedia.pdf");

// Continuous query
while (true)
{
    Console.Write("User Input: ");
    var response = await pdf.Query(Console.ReadLine());
    Console.WriteLine($"\n{response}");
}

Imports Microsoft.VisualBasic
Imports IronPdf
Imports IronPdf.AI
Imports Microsoft.SemanticKernel
Imports Microsoft.SemanticKernel.Connectors.OpenAI
Imports Microsoft.SemanticKernel.Memory
Imports System
Imports System.Threading.Tasks

' Setup OpenAI
Private azureEndpoint = "<<enter your azure endpoint here>>"
Private apiKey = "<<enter your azure API key here>>"
Private builder = Kernel.CreateBuilder().AddAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey).AddAzureOpenAIChatCompletion("oaichat", azureEndpoint, apiKey)
Private kernel = builder.Build()

' Setup Memory
Private memory_builder = (New MemoryBuilder()).WithMemoryStore(New VolatileMemoryStore()).WithAzureOpenAITextEmbeddingGeneration("oaiembed", azureEndpoint, apiKey)
Private memory = memory_builder.Build()

' Initialize IronAI
IronDocumentAI.Initialize(kernel, memory)

License.LicenseKey = "<<enter your IronPdf license key here"

' Import PDF document
Dim pdf As PdfDocument = PdfDocument.FromFile("wikipedia.pdf")

' Continuous query
Do
	Console.Write("User Input: ")
	Dim response = Await pdf.Query(Console.ReadLine())
	Console.WriteLine($vbLf & "{response}")
Loop

$vbLabelText $csharpLabel

查克尼特·宾

立即与工程团队聊天

软件工程师

Chaknith 负责 IronXL 和 IronBarcode 的工作。他在 C# 和 .NET 方面拥有深厚的专业知识，帮助改进软件并支持客户。他从用户互动中获得的洞察力，有助于提升产品、文档和整体体验。