用Clipboard获取HTML格式内容

.NET的Clipboard.GetData(DataFormats.Html)方法有BUG。
网上查询了一下，有个解决方法，不过那个网站被封了（至少我这没法访问）
用某软件上去看了一下，照做，OK了

using  System;
using  System.Text;
using  System.Diagnostics;
using  System.Windows.Forms;
using  System.Runtime.InteropServices;
using  System.Runtime.InteropServices.ComTypes;



namespace  JX
{
    /**//// <summary>
    /// By JX
    /// Version 1.0.200511041900   
    /// This static class contains a method to extract Dataformat.Html data from a IDataObject 
    /// obtained from the Clipboard or a Drag and Drop operation.
    /// The method overcomes the UTF8 corruption problem encountered in .NET Framework version 2.0.50727.42
    /// Version 1.0 of the Dataformat.Html data has a header like this:
    /// 
    /// Version:1.0
    /// StartHTML:000000238
    /// EndHTML:000000546
    /// StartFragment:000000353
    /// EndFragment:000000510
    /// StartSelection:000000353
    /// EndSelection:000000510
    /// 
    /// The values refer to the BYTES of the string not the actual chars
    /// </summary>
    static class HtmlFromIDataObject
    {
        /**//// <summary>
        /// Extracts data of type Dataformat.Html from an IdataObject data container
        /// This method shouldn't throw any exception but writes relevant exception informations in the debug window
        /// </summary>
        /// <param name="data">IdataObject data container</param>
        /// <returns>A byte[] array with the decoded string or null if the method fails</returns>
        public static byte[] GetHtml(System.Windows.Forms.IDataObject data)
        {
            System.Runtime.InteropServices.ComTypes.IDataObject interopData = data as System.Runtime.InteropServices.ComTypes.IDataObject;

            FORMATETC format = new FORMATETC();          
            format.cfFormat = (short)DataFormats.GetFormat(DataFormats.Html).Id;
            format.dwAspect = DVASPECT.DVASPECT_CONTENT;
            format.lindex = -1;   
            format.tymed = TYMED.TYMED_HGLOBAL;

            STGMEDIUM stgmedium = new STGMEDIUM();
            stgmedium.tymed = TYMED.TYMED_HGLOBAL;
            stgmedium.pUnkForRelease = null;

            int queryResult = 0;

            try
            {
                queryResult = interopData.QueryGetData(ref format);
            }
            catch (Exception exp)
            {
                Debug.WriteLine("HtmlFromIDataObject.GetHtml -> QueryGetData(ref format) threw an exception: " 
                    + Environment.NewLine + exp.ToString());
                return null;
            }

            if (queryResult != 0)
            {
                Debug.WriteLine("HtmlFromIDataObject.GetHtml -> QueryGetData(ref format) returned a code != 0 code: " 
                    + queryResult.ToString());
                return null;
            }

            try
            {
                interopData.GetData(ref format, out stgmedium);
            }
            catch (Exception exp)
            {
                System.Diagnostics.Debug.WriteLine("HtmlFromIDataObject.GetHtml -> GetData(ref format, out stgmedium) threw this exception: " 
                    + Environment.NewLine + exp.ToString());
                return null;
            }

            if (stgmedium.unionmember == IntPtr.Zero)
            {
                Debug.WriteLine("HtmlFromIDataObject.GetHtml -> stgmedium.unionmember returned an IntPtr pointing to zero");
                return null;
            }

            IntPtr pointer = stgmedium.unionmember;
            
            HandleRef handleRef = new HandleRef(null, pointer);

            byte[] rawArray = null;

            try
            {
                IntPtr ptr1 = GlobalLock(handleRef);

                int length = GlobalSize(handleRef);

                rawArray = new byte[length];

                Marshal.Copy(ptr1, rawArray, 0, length);          

            }
            catch (Exception exp)
            {
                Debug.WriteLine("HtmlFromIDataObject.GetHtml -> Html Import threw an exception: " + Environment.NewLine + exp.ToString());
            }
            finally
            {
                GlobalUnlock(handleRef);

            }

            return rawArray;
        }


        [DllImport("kernel32.dll", CharSet = CharSet.Auto, ExactSpelling = true, SetLastError = true)]
        private static extern IntPtr GlobalLock(HandleRef handle);

        [DllImport("kernel32.dll", CharSet = CharSet.Auto, ExactSpelling = true, SetLastError = true)]
        private static extern bool GlobalUnlock(HandleRef handle);

        [DllImport("kernel32.dll", CharSet = CharSet.Auto, ExactSpelling = true, SetLastError = true)]
        private static extern int GlobalSize(HandleRef handle);
 
    }
}

byte [] rawHtmlBytes  =  JX.HtmlFromIDataObject.GetHtml(dataObject)

string  rawHtml  =  Encoding.UTF8.GetString(rawHtmlBytes);

原文地址：http://bakamachine.blogspot.com/2006/05/workarond-for-dataobject-html.html

转载于:https://www.cnblogs.com/yushiro/archive/2007/09/15/894080.html