C#根据html生成PDF

使用iTextSharp 控件

iTextSharp包括几个dll。

主要iTextSharp版本包含:
——itextsharp。dll:核心库
——itextsharp.xtra。dll:额外的功能(PDF 2 !)
——itextsharp.pdfa。dll:PDF /相关的功能
这个项目是托管在http://sourceforge.net/projects/itextsharp/
你可以在这里找到的最新版本:
http://sourceforge.net/projects/itextsharp/files/itextsharp/

在某些情况下,您将需要额外的dll。
这些dll是可用的:
http://sourceforge.net/projects/itextsharp/files/extras/

对于XML和HTML功能,您需要这个dll:
——itextsharp.xmlworker.dll
这是可以在http://sourceforge.net/projects/itextsharp/files/xmlworker/

最后,我们也有一个Java工具,可以帮助您调试pdf文档:
——itext-rups-x.y.z.jar
这个项目是托管在http://sourceforge.net/projects/itextrups/

下列代码需要导入两个.dll

itextsharp.dll:核心库

itextsharp.xmlworker.dll

using System;
using System.Collections.Generic;

using System.Web;
using System.Web.UI;
using System.Web.UI.WebControls;
using iTextSharp.text;
using iTextSharp.text.pdf;
using System.IO;
using iTextSharp.tool.xml;
using System.Text;
using System.Net;

namespace PDFTools
{
    public partial class CreatePdf : System.Web.UI.Page
    {
        string INPATH = System.Web.HttpContext.Current.Server.MapPath("~/bin/temp.html");

        protected void Page_Load(object sender, EventArgs e)
        {
            string path = System.Web.HttpContext.Current.Server.MapPath("~/UploadPdf");
            if (!Directory.Exists(path))
                Directory.CreateDirectory(path);

            WebClient wc = new WebClient();
            //从网址下载Html字串
            wc.Encoding = System.Text.Encoding.UTF8;
            string htmlText = getWebContent();
            byte[] pdfFile = this.ConvertHtmlTextToPDF(htmlText);

            string fileId = "/file_" + DateTime.Now.ToString("yyyyMMddHHmmss") + ".pdf";
            System.IO.File.WriteAllBytes(path + fileId, pdfFile);
        }

        /// <summary>
        /// 判断是否有乱码
        /// </summary>
        /// <param name="txt"></param>
        /// <returns></returns>
        public bool isMessyCode(string txt)
        {
            var bytes = Encoding.UTF8.GetBytes(txt);            //239 191 189
            for (var i = ; i < bytes.Length; i++)
            {
                if (i < bytes.Length - )
                    if (bytes[i] ==  && bytes[i + ] ==  && bytes[i + ] == )
                    {
                        return true;
                    }
            }
            return false;
        }
        /// <summary>
        /// 获取网站内容，包含了 HTML+CSS+JS
        /// </summary>
        /// <returns>String返回网页信息</returns>
        public string getWebContent()
        {
            try
            {
                WebClient MyWebClient = new WebClient();
                MyWebClient.Credentials = CredentialCache.DefaultCredentials;
                //获取或设置用于向Internet资源的请求进行身份验证的网络凭据
                Byte[] pageData = MyWebClient.DownloadData(INPATH);
                //从指定网站下载数据
                string pageHtml = Encoding.UTF8.GetString(pageData);
                //如果获取网站页面采用的是GB2312，则使用这句
                bool isBool = isMessyCode(pageHtml);//判断使用哪种编码 读取网页信息
                if (!isBool)
                {
                    string pageHtml1 = Encoding.UTF8.GetString(pageData);
                    pageHtml = pageHtml1;
                }
                else
                {
                    string pageHtml2 = Encoding.Default.GetString(pageData);
                    pageHtml = pageHtml2;
                }
                return pageHtml;
            }

            catch (WebException webEx)
            {
                Console.WriteLine(webEx.Message.ToString());
                return webEx.Message;
            }
        }

        /// <summary>
        /// 将Html文字 输出到PDF档里
        /// </summary>
        /// <param name="htmlText"></param>
        /// <returns></returns>
        public byte[] ConvertHtmlTextToPDF(string htmlText)
        {
            if (string.IsNullOrEmpty(htmlText))
            {
                return null;
            }
            //避免当htmlText无任何html tag标签的纯文字时，转PDF时会挂掉，所以一律加上<p>标签
            htmlText = "<p>" + htmlText + "</p>";

            MemoryStream outputStream = new MemoryStream();//要把PDF写到哪个串流
            byte[] data = Encoding.UTF8.GetBytes(htmlText);//字串转成byte[]
            MemoryStream msInput = new MemoryStream(data);
            Document doc = new Document();//要写PDF的文件，建构子没填的话预设直式A4
            PdfWriter writer = PdfWriter.GetInstance(doc, outputStream);
            //指定文件预设开档时的缩放为100%

            PdfDestination pdfDest = new PdfDestination(PdfDestination.XYZ, , doc.PageSize.Height, 1f);
            //开启Document文件
            doc.Open();

            //使用XMLWorkerHelper把Html parse到PDF档里
            // XMLWorkerHelper.GetInstance().ParseXHtml(writer, doc, msInput, null, Encoding.UTF8, new UnicodeFontFactory());
            XMLWorkerHelper.GetInstance().ParseXHtml(writer, doc, msInput, null, Encoding.UTF8);

            //将pdfDest设定的资料写到PDF档
            PdfAction action = PdfAction.GotoLocalPage(, pdfDest, writer);
            writer.SetOpenAction(action);
            doc.Close();
            msInput.Close();
            outputStream.Close();
            //回传PDF档案
            return outputStream.ToArray();

        }

        //设置字体类
        public class UnicodeFontFactory : FontFactoryImp
        {
            private static readonly string arialFontPath = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Fonts),
      "arialuni.ttf");//arial unicode MS是完整的unicode字型。
            private static readonly string 标楷体Path = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Fonts),
            "KAIU.TTF");//标楷体

            public override Font GetFont(string fontname, string encoding, bool embedded, float size, int style, BaseColor color, bool cached)
            {
                BaseFont bfChiness = BaseFont.CreateFont(@"C:\Windows\Fonts\SIMSUN.TTC,1", BaseFont.IDENTITY_H, BaseFont.NOT_EMBEDDED);
                //可用Arial或标楷体，自己选一个
                BaseFont baseFont = BaseFont.CreateFont(标楷体Path, BaseFont.IDENTITY_H, BaseFont.EMBEDDED);
                return new Font(bfChiness, size, style, color);
            }
        }
    }
}