利用com调用ie进行html解析

最新推荐文章于 2022-04-16 08:16:07 发布

weixin_33852020

最新推荐文章于 2022-04-16 08:16:07 发布

阅读量127

点赞数

文章标签： runtime

本文介绍了一个简单的HTML解析器实现过程，使用C#语言通过COM接口与MSHTML组件交互，完成网页内容的加载与解析。该解析器能够从指定URL加载网页，并等待页面完全加载完毕后获取IHTMLDocument3接口实例。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

别的就不多说了，直接上代码，代码很简单的，不懂的留言。

1using System;
 2using System.Collections.Generic;
 3using System.Linq;
 4using System.Text;
 5using System.Runtime.InteropServices;
 6using mshtml;
 7using System.Threading;
 8using System.Runtime.InteropServices.ComTypes;
 9using System.IO;
10
11namespace Eric.Utilities.Html
12{
13    public enum HRESULT
14    {
15        E_FAIL = -2147467259,
16        E_INVALIDARG = -2147024809,
17        E_NOINTERFACE = -2147467262,
18        E_NOTIMPL = -2147467263,
19        E_UNEXPECTED = -2147418113,
20        S_FALSE = 1,
21        S_OK = 0
22    }
23
24    [ComImport, Guid("0000010c-0000-0000-C000-000000000046"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown), ComVisible(true)]
25    public interface IPersist
26    {
27        void GetClassID(ref Guid pClassID);
28    }
29
30    [ComImport, InterfaceType(ComInterfaceType.InterfaceIsIUnknown), ComVisible(true), Guid("7FD52380-4E07-101B-AE2D-08002B2EC713")]
31    public interface IPersistStreamInit : IPersist
32    {
33        new void GetClassID(ref Guid pClassID);
34        [PreserveSig]
35        int IsDirty();
36        [PreserveSig]
37        HRESULT Load(IStream pstm);
38        [PreserveSig]
39        HRESULT Save(IStream pstm, [MarshalAs(UnmanagedType.Bool)] bool fClearDirty);
40        [PreserveSig]
41        HRESULT GetSizeMax([In, Out, MarshalAs(UnmanagedType.U8)] ref long pcbSize);
42        [PreserveSig]
43        HRESULT InitNew();
44    }
45
46    public class HtmlParser
47    {
48        public IHTMLDocument3 Parse(string url)
49        {
50            HTMLDocument objMSHTML = new HTMLDocument();
51            IHTMLDocument2 objMSHTML2;
52            IHTMLDocument3 objMSHTML3;
53
54            IPersistStreamInit objIPS;
55            objIPS = objMSHTML as IPersistStreamInit;
56            objIPS.InitNew();
57            objIPS = null;
58
59            objMSHTML2 = objMSHTML.createDocumentFromUrl(url, "null");
60            while (objMSHTML2.readyState != "complete")
61            {
62                Thread.Sleep(1000);
63            }
64            objMSHTML3 = objMSHTML2 as IHTMLDocument3;
65            return objMSHTML3;
66        }
67
68        public IHTMLDocument3 ParseHtml(string html, Encoding encoding)
69        {
70            string tmpFile = Path.GetTempFileName();
71            File.WriteAllText(tmpFile, html, encoding);
72            return Parse(tmpFile);
73        }
74    }
75}