如何使用iText将带图像和超链接的HTML转换为PDF?

html-agility-pack html-parsing itext pdf xmlworker

我正在尝试使用iTextSharp在使用MVC Web表单ASP.NET Web应用程序中将HTML转换为PDF<img><a>元素具有绝对和相对 URL,而一些<img>元素是base64 。 SO和Google搜索结果的典型答案使用XMLWorkerHelper通用HTMLPDF代码,如下所示:

using (var stringReader = new StringReader(xHtml))
{
    using (Document document = new Document())
    {
        PdfWriter writer = PdfWriter.GetInstance(document, stream);
        document.Open();
        XMLWorkerHelper.GetInstance().ParseXHtml(
            writer, document, stringReader
        );
    }
}

所以使用这样的示例HTML

<div>
    <h3>HTML Works, but Broken in Converted PDF</h3>
    <div>Relative local <img>: <img src='./../content/images/kuujinbo_320-30.gif' /></div>
    <div>
        Base64 <img>:
        <img src='data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAUAAAAFCAYAAACNbyblAAAAHElEQVQI12P4//8/w38GIAXDIBKE0DHxgljNBAAO9TXL0Y4OHwAAAABJRU5ErkJggg==' />
    </div>
    <div><a href='/somePage.html'>Relative local hyperlink, broken in PDF</a></div>
<div>

生成的PDF: (1)缺少所有图像, (2)所有带有相对URL的超链接都被破坏并使用文件URI方案file///XXX... )而不是指向正确的网站。

SO的一些答案和Google搜索的其他答案建议用绝对URL替换相对URL,这对于一次性案例是完全可以接受的。但是,全局用硬编码字符串替换所有<img src><a href>属性对于这个问题是不可接受的,所以请不要发布这样的答案,因为它会相应地被低估。

我正在寻找适用于测试,开发和生产环境中的许多不同Web应用程序的解决方案。

一般承认的答案

开箱即用的XMLWorker 只能理解绝对URI ,因此所描述的问题是预期的行为。如果没有其他信息,解析器无法自动推导出URI方案或路径。

实现ILinkProvider可以修复损坏的超链接问题,实现IImageProvider可以修复损坏的图像问题。由于两种实现都必须执行URI解析 ,这是第一步。下面的帮助器类可以做到这一点,并且还尝试尽可能简单地进行Web( ASP.NET )上下文调用(示例如下):

// resolve URIs for LinkProvider & ImageProvider
public class UriHelper
{
    /* IsLocal; when running in web context:
     * [1] give LinkProvider http[s] scheme; see CreateBase(string baseUri)
     * [2] give ImageProvider relative path starting with '/' - see:
     *     Join(string relativeUri)
     */
    public bool IsLocal { get; set; }
    public HttpContext HttpContext { get; private set; }
    public Uri BaseUri { get; private set; }

    public UriHelper(string baseUri) : this(baseUri, true) {}
    public UriHelper(string baseUri, bool isLocal)
    {
        IsLocal = isLocal;
        HttpContext = HttpContext.Current;
        BaseUri = CreateBase(baseUri);
    }

    /* get URI for IImageProvider to instantiate iTextSharp.text.Image for 
     * each <img> element in the HTML.
     */
    public string Combine(string relativeUri)
    {
        /* when running in a web context, the HTML is coming from a MVC view 
         * or web form, so convert the incoming URI to a **local** path
         */
        if (HttpContext != null && !BaseUri.IsAbsoluteUri && IsLocal)
        {
            return HttpContext.Server.MapPath(
                // Combine() checks directory traversal exploits
                VirtualPathUtility.Combine(BaseUri.ToString(), relativeUri)
            );
        }
        return BaseUri.Scheme == Uri.UriSchemeFile 
            ? Path.Combine(BaseUri.LocalPath, relativeUri)
            // for this example we're assuming URI.Scheme is http[s]
            : new Uri(BaseUri, relativeUri).AbsoluteUri;
    }

    private Uri CreateBase(string baseUri)
    {
        if (HttpContext != null)
        {   // running on a web server; need to update original value  
            var req = HttpContext.Request;
            baseUri = IsLocal
                // IImageProvider; absolute virtual path (starts with '/')
                // used to convert to local file system path. see:
                // Combine(string relativeUri)
                ? req.ApplicationPath
                // ILinkProvider; absolute http[s] URI scheme
                : req.Url.GetLeftPart(UriPartial.Authority)
                    + HttpContext.Request.ApplicationPath;
        }

        Uri uri;
        if (Uri.TryCreate(baseUri, UriKind.RelativeOrAbsolute, out uri)) return uri;

        throw new InvalidOperationException("cannot create a valid BaseUri");
    }
}

现在, UriHelper提供了基本URI,实现ILinkProvider非常简单。我们只需要正确的URI方案( filehttp[s] ):

// make hyperlinks with relative URLs absolute
public class LinkProvider : ILinkProvider
{
    // rfc1738 - file URI scheme section 3.10
    public const char SEPARATOR = '/';
    public string BaseUrl { get; private set; }

    public LinkProvider(UriHelper uriHelper)
    {
        var uri = uriHelper.BaseUri;
        /* simplified implementation that only takes into account:
         * Uri.UriSchemeFile || Uri.UriSchemeHttp || Uri.UriSchemeHttps
         */
        BaseUrl = uri.Scheme == Uri.UriSchemeFile
            // need trailing separator or file paths break
            ? uri.AbsoluteUri.TrimEnd(SEPARATOR) + SEPARATOR
            // assumes Uri.UriSchemeHttp || Uri.UriSchemeHttps
            : BaseUrl = uri.AbsoluteUri;
    }

    public string GetLinkRoot()
    {
        return BaseUrl;
    }
}

IImageProvider需要实现一个方法, Retrieve(string src) ,但Store(string src, Image img)很简单 - 注意那里的内联注释和GetImageRootPath()

// handle <img> elements in HTML  
public class ImageProvider : IImageProvider
{
    private UriHelper _uriHelper;
    // see Store(string src, Image img)
    private Dictionary<string, Image> _imageCache = 
        new Dictionary<string, Image>();

    public virtual float ScalePercent { get; set; }
    public virtual Regex Base64 { get; set; }

    public ImageProvider(UriHelper uriHelper) : this(uriHelper, 67f) { }
    //              hard-coded based on general past experience ^^^
    // but call the overload to supply your own
    public ImageProvider(UriHelper uriHelper, float scalePercent)
    {
        _uriHelper = uriHelper;
        ScalePercent = scalePercent;
        Base64 = new Regex( // rfc2045, section 6.8 (alphabet/padding)
            @"^data:image/[^;]+;base64,(?<data>[a-z0-9+/]+={0,2})$",
            RegexOptions.Compiled | RegexOptions.IgnoreCase
        );
    }

    public virtual Image ScaleImage(Image img)
    {
        img.ScalePercent(ScalePercent);
        return img;
    }

    public virtual Image Retrieve(string src)
    {
        if (_imageCache.ContainsKey(src)) return _imageCache[src];

        try
        {
            if (Regex.IsMatch(src, "^https?://", RegexOptions.IgnoreCase))
            {
                return ScaleImage(Image.GetInstance(src));
            }

            Match match;
            if ((match = Base64.Match(src)).Length > 0)
            {
                return ScaleImage(Image.GetInstance(
                    Convert.FromBase64String(match.Groups["data"].Value)
                ));
            }

            var imgPath = _uriHelper.Combine(src);
            return ScaleImage(Image.GetInstance(imgPath));
        }
        // not implemented to keep the SO answer (relatively) short
        catch (BadElementException ex) { return null; }
        catch (IOException ex) { return null; }
        catch (Exception ex) { return null; }
    }

    /*
     * always called after Retrieve(string src):
     * [1] cache any duplicate <img> in the HTML source so the image bytes
     *     are only written to the PDF **once**, which reduces the 
     *     resulting file size.
     * [2] the cache can also **potentially** save network IO if you're
     *     running the parser in a loop, since Image.GetInstance() creates
     *     a WebRequest when an image resides on a remote server. couldn't
     *     find a CachePolicy in the source code
     */
    public virtual void Store(string src, Image img)
    {
        if (!_imageCache.ContainsKey(src)) _imageCache.Add(src, img);
    }

    /* XMLWorker documentation for ImageProvider recommends implementing
     * GetImageRootPath():
     * 
     * http://demo.itextsupport.com/xmlworker/itextdoc/flatsite.html#itextdoc-menu-10
     * 
     * but a quick run through the debugger never hits the breakpoint, so 
     * not sure if I'm missing something, or something has changed internally 
     * with XMLWorker....
     */
    public virtual string GetImageRootPath() { return null; }
    public virtual void Reset() { }
}

基于XML Worker文档IImageProvider上面的ILinkProviderIImageProvider的实现挂钩到一个简单的解析器类中非常简单:

/* a simple parser that uses XMLWorker and XMLParser to handle converting 
 * (most) images and hyperlinks internally
 */
public class SimpleParser
{
    public virtual ILinkProvider LinkProvider { get; set; }
    public virtual IImageProvider ImageProvider { get; set; }

    public virtual HtmlPipelineContext HtmlPipelineContext { get; set; }
    public virtual ITagProcessorFactory TagProcessorFactory { get; set; }
    public virtual ICSSResolver CssResolver { get; set; }

    /* overloads simplfied to keep SO answer (relatively) short. if needed
     * set LinkProvider/ImageProvider after instantiating SimpleParser()
     * to override the defaults (e.g. ImageProvider.ScalePercent)
     */
    public SimpleParser() : this(null) { }
    public SimpleParser(string baseUri)
    {
        LinkProvider = new LinkProvider(new UriHelper(baseUri, false));
        ImageProvider = new ImageProvider(new UriHelper(baseUri, true));

        HtmlPipelineContext = new HtmlPipelineContext(null);

        // another story altogether, and not implemented for simplicity 
        TagProcessorFactory = Tags.GetHtmlTagProcessorFactory();
        CssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);
    }

    /*
     * when sending XHR via any of the popular JavaScript frameworks,
     * <img> tags are **NOT** always closed, which results in the 
     * infamous iTextSharp.tool.xml.exceptions.RuntimeWorkerException:
     * 'Invalid nested tag a found, expected closing tag img.' a simple
     * workaround.
     */
    public virtual string SimpleAjaxImgFix(string xHtml)
    {
        return Regex.Replace(
            xHtml,
            "(?<image><img[^>]+)(?<=[^/])>",
            new MatchEvaluator(match => match.Groups["image"].Value + " />"),
            RegexOptions.IgnoreCase | RegexOptions.Multiline
        );
    }

    public virtual void Parse(Stream stream, string xHtml)
    {
        xHtml = SimpleAjaxImgFix(xHtml);

        using (var stringReader = new StringReader(xHtml))
        {
            using (Document document = new Document())
            {
                PdfWriter writer = PdfWriter.GetInstance(document, stream);
                document.Open();

                HtmlPipelineContext
                    .SetTagFactory(Tags.GetHtmlTagProcessorFactory())
                    .SetLinkProvider(LinkProvider)
                    .SetImageProvider(ImageProvider)
                ;
                var pdfWriterPipeline = new PdfWriterPipeline(document, writer);
                var htmlPipeline = new HtmlPipeline(HtmlPipelineContext, pdfWriterPipeline);
                var cssResolverPipeline = new CssResolverPipeline(CssResolver, htmlPipeline);

                XMLWorker worker = new XMLWorker(cssResolverPipeline, true);
                XMLParser parser = new XMLParser(worker);
                parser.Parse(stringReader);
            }
        }
    }
}

正如内联注释, SimpleAjaxImgFix(string xHtml)专门处理可能发送未封闭的<img>标签的XHR ,这些标签有效的 HTML ,但是无效的 XML 破坏XMLWorker可以在此处找到有关如何使用XHR和iTextSharp接收PDF或其他二进制数据的简单说明和实现。

SimpleAjaxImgFix(string xHtml)使用了SimpleAjaxImgFix(string xHtml) Regex ,因此任何使用( 复制/粘贴 ?)代码的人都不需要添加另一个nuget包,但是应该使用像HtmlAgilityPack这样的HTML解析器,因为它转为:

<div><img src='a.gif'><br><hr></div>

进入这个:

<div><img src='a.gif' /><br /><hr /></div>

只需几行代码:

var hDocument = new HtmlDocument()
{
    OptionWriteEmptyNodes = true,
    OptionAutoCloseOnEnd = true
};
hDocument.LoadHtml("<div><img src='a.gif'><br><hr></div>");
var closedTags  = hDocument.DocumentNode.WriteTo();

另外值得注意的是 - 使用上面的SimpleParser.Parse()作为另外实现自定义ICSSResolverITagProcessorFactory一般蓝图,这在文档中解释

现在应该处理问题中描述的问题。从MVC Action Method调用:

[HttpPost]  // some browsers have URL length limits
[ValidateInput(false)] // or throws HttpRequestValidationException
public ActionResult Index(string xHtml)
{
    Response.ContentType = "application/pdf";
    Response.AppendHeader(
        "Content-Disposition", "attachment; filename=test.pdf"
    );
    var simpleParser = new SimpleParser();
    simpleParser.Parse(Response.OutputStream, xHtml);

    return new EmptyResult();
}

或者从Web Form 服务器控件获取HTMLWeb Form

Response.ContentType = "application/pdf";
Response.AppendHeader("Content-Disposition", "attachment; filename=test.pdf");
using (var stringWriter = new StringWriter())
{
    using (var htmlWriter = new HtmlTextWriter(stringWriter))
    {
        ConvertControlToPdf.RenderControl(htmlWriter);
    }
    var simpleParser = new SimpleParser();
    simpleParser.Parse(Response.OutputStream, stringWriter.ToString());
}
Response.End();

或者在文件系统上包含超链接和图像的简单HTML文件:

<h1>HTML Page 00 on Local File System</h1>
<div>
    <div>
        Relative &lt;img&gt;: <img src='Images/alt-gravatar.png' />
    </div>
    <div>
        Hyperlink to file system HTML page: 
        <a href='file-system-html-01.html'>Page 01</a>
    </div>
</div>

或来自远程网站的HTML:

<div>
    <div>
        <img width="200" alt="Wikipedia Logo"
             src="portal/wikipedia.org/assets/img/Wikipedia-logo-v2.png">
    </div>
    <div lang="en">
        <a href="https://en.wikipedia.org/">English</a>
    </div>
    <div lang="en">
        <a href="wiki/IText">iText</a>
    </div>
</div>

以上两个HTML段从控制台应用程序运行:

var filePaths = Path.Combine(basePath, "file-system-html-00.html");
var htmlFile = File.ReadAllText(filePaths);
var remoteUrl = Path.Combine(basePath, "wikipedia.html");
var htmlRemote = File.ReadAllText(remoteUrl);
var outputFile = Path.Combine(basePath, "filePaths.pdf");
var outputRemote = Path.Combine(basePath, "remoteUrl.pdf");

using (var stream = new FileStream(outputFile, FileMode.Create))
{
    var simpleParser = new SimpleParser(basePath);
    simpleParser.Parse(stream, htmlFile);
}
using (var stream = new FileStream(outputRemote, FileMode.Create))
{
    var simpleParser = new SimpleParser("https://wikipedia.org");
    simpleParser.Parse(stream, htmlRemote);
}

相当长的答案,但在SO标记的htmlpdfitextsharp查看这里的问题,截至撰写本文时(2016-02-23),有776个结果反对4,063个总标记的itextsharp - 这是19%


热门答案

非常有用的帖子,

我在报告html中将图像渲染为pdf时遇到了问题。你的帖子我可以做到。

我正在使用asp.mvc 5。

我只需要改变ImageProviderClass的这个方法

public virtual string GetImageRootPath() { return null; }

public virtual string GetImageRootPath() { HostingEnvironment.MapPath("~/Content/Images/") }

谢谢!



Related

许可下: CC-BY-SA with attribution
不隶属于 Stack Overflow
许可下: CC-BY-SA with attribution
不隶属于 Stack Overflow