protected override void TransformHtml2Pdf()
{
Document doc = new Document(PageSize.A4.Rotate());
PdfWriter pdfWriter = PdfWriter.GetInstance(doc, new FileStream(outPdf, FileMode.Create));
doc.SetMargins(doc.LeftMargin - 10, doc.RightMargin - 10, doc.TopMargin, doc.BottomMargin);
doc.Open();
CssFilesImpl cssFiles = new CssFilesImpl();
cssFiles.Add(
XMLWorkerHelper.GetCSS(
File.OpenRead(RESOURCES + @"\tool\xml\examples\" + "sampleTest.css")));
StyleAttrCSSResolver cssResolver = new StyleAttrCSSResolver(cssFiles);
HtmlPipelineContext hpc =
new HtmlPipelineContext(
new CssAppliersImpl(new XMLWorkerFontProvider(RESOURCES + @"\tool\xml\examples\fonts")));
hpc.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(Tags.GetHtmlTagProcessorFactory());
hpc.SetImageProvider(new SampleTest.SampleTestImageProvider());
HtmlPipeline htmlPipeline = new HtmlPipeline(hpc, new PdfWriterPipeline(doc, pdfWriter));
IPipeline pipeline = new CssResolverPipeline(cssResolver, htmlPipeline);
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser p = new XMLParser(true, worker, Encoding.GetEncoding("UTF-8"));
p.Parse(File.OpenRead(inputHtml), Encoding.GetEncoding("UTF-8"));
doc.Close();
}
protected override void MakePdf(string outPdf)
{
Document doc = new Document(PageSize.A4_LANDSCAPE);
PdfWriter pdfWriter = PdfWriter.GetInstance(doc, new FileStream(outPath + outPdf, FileMode.Create));
doc.SetMargins(0, 0, 0, 0);
doc.Open();
CssFilesImpl cssFiles = new CssFilesImpl();
cssFiles.Add(
XMLWorkerHelper.GetCSS(
File.OpenRead(RESOURCES + Path.DirectorySeparatorChar + testPath + testName +
Path.DirectorySeparatorChar + "sampleTest.css")));
StyleAttrCSSResolver cssResolver = new StyleAttrCSSResolver(cssFiles);
HtmlPipelineContext hpc =
new HtmlPipelineContext(
new CssAppliersImpl(new XMLWorkerFontProvider(RESOURCES + @"\tool\xml\examples\fonts")));
hpc.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(Tags.GetHtmlTagProcessorFactory());
hpc.SetImageProvider(new SampleTestImageProvider());
HtmlPipeline htmlPipeline = new HtmlPipeline(hpc, new PdfWriterPipeline(doc, pdfWriter));
IPipeline pipeline = new CssResolverPipeline(cssResolver, htmlPipeline);
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser p = new XMLParser(true, worker, Encoding.GetEncoding("UTF-8"));
p.Parse(File.OpenRead(inputHtml), Encoding.GetEncoding("UTF-8"));
doc.Close();
}
private static bool GeneratePdf(string htmlText, string fileFullName, string watermarkText, PdfFont font)
{
if (string.IsNullOrEmpty(htmlText))
{
return(false);
}
htmlText = "<p>" + htmlText + "</p>";
var document = new Document();
var writer = PdfWriter.GetInstance(document, new FileStream(fileFullName, FileMode.Create));
if (!string.IsNullOrEmpty(watermarkText))
{
writer.PageEvent = new PdfWatermarkPageEvent(watermarkText);
}
document.Open();
//pipeline
var htmlContext = new HtmlPipelineContext(null);
htmlContext.SetTagFactory(Tags.GetHtmlTagProcessorFactory());
htmlContext.SetImageProvider(new ChannelImageProvider());
htmlContext.SetCssAppliers(new CssAppliersImpl(GetFontProviderBy(font)));
var cssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);
var pipeline = new CssResolverPipeline(cssResolver,
new HtmlPipeline(htmlContext, new PdfWriterPipeline(document, writer)));
//parse
byte[] data = Encoding.UTF8.GetBytes(htmlText);
var msInput = new MemoryStream(data);
var worker = new XMLWorker(pipeline, true);
var parser = new XMLParser(worker);
parser.Parse(msInput); //XMLWorkerHelper.GetInstance().ParseXHtml(..)
var pdfDest = new PdfDestination(PdfDestination.XYZ, 0, document.PageSize.Height, 1f);
var action = PdfAction.GotoLocalPage(1, pdfDest, writer);
writer.SetOpenAction(action);
//close
document.Close();
msInput.Close();
return(true);
}
private void processHtml(IElementHandler elementsHandler)
{
var cssResolver = new StyleAttrCSSResolver();
if (CssFilesPath != null && CssFilesPath.Any())
{
foreach (var cssFile in CssFilesPath)
{
cssResolver.AddCss(XmlWorkerUtils.GetCssFile(cssFile));
}
}
if (!string.IsNullOrEmpty(InlineCss))
{
cssResolver.AddCss(InlineCss, "utf-8", true);
}
var htmlContext = new HtmlPipelineContext(new CssAppliersImpl(new UnicodeFontProvider(DefaultFont)));
if (!string.IsNullOrEmpty(ImagesPath))
{
htmlContext.SetImageProvider(new ImageProvider {
ImagesPath = ImagesPath
});
}
htmlContext.CharSet(Encoding.UTF8);
var tagsProcessorFactory = (DefaultTagProcessorFactory)Tags.GetHtmlTagProcessorFactory();
if (PdfElement != null)
{
tagsProcessorFactory.AddProcessor("totalpagesnumber", new TotalPagesNumberXmlWorkerProcessor(PdfElement));
}
htmlContext.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(tagsProcessorFactory);
var pipeline = new CssResolverPipeline(cssResolver,
new HtmlPipeline(htmlContext, new ElementHandlerPipeline(elementsHandler, null)));
var worker = new XMLWorker(pipeline, parseHtml: true);
var parser = new XMLParser();
parser.AddListener(worker);
parser.Parse(new StringReader(Html));
}
virtual public void AddingAnImageRoot()
{
Document doc = new Document(PageSize.A4);
PdfWriter writer = PdfWriter.GetInstance(doc,
new FileStream(TARGET + "columbus3.pdf", FileMode.Create));
doc.Open();
HtmlPipelineContext htmlContext = new HtmlPipelineContext(null);
htmlContext.SetImageProvider(new CustomImageProvider()).SetTagFactory(Tags.GetHtmlTagProcessorFactory());
ICSSResolver cssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);
IPipeline pipeline = new CssResolverPipeline(cssResolver, new HtmlPipeline(htmlContext,
new PdfWriterPipeline(doc, writer)));
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser p = new XMLParser(worker);
p.Parse(File.OpenRead(RESOURCES + @"\examples\columbus.html"));
doc.Close();
}
private XMLWorker CreateXmlWorker(Document doc, PdfWriter writer)
{
var cssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);
cssResolver.AddCss(content: css, isPersistent: true);
var htmlContext = new HtmlPipelineContext(null);
htmlContext.SetTagFactory(Tags.GetHtmlTagProcessorFactory());
htmlContext.SetImageProvider(imageProvider);
var pdfPipeline = new PdfWriterPipeline(doc, writer);
var htmlPipeline = new HtmlPipeline(htmlContext, pdfPipeline);
var cssPipeline = new CssResolverPipeline(cssResolver, htmlPipeline);
var worker = new XMLWorker(cssPipeline, true);
return(worker);
}
protected override void MakePdf(string outPdf)
{
Document doc = new Document(PageSize.A3.Rotate());
PdfWriter pdfWriter = PdfWriter.GetInstance(doc, new FileStream(outPdf, FileMode.Create));
pdfWriter.CreateXmpMetadata();
doc.SetMargins(200, 200, 0, 0);
doc.Open();
CssFilesImpl cssFiles = new CssFilesImpl();
cssFiles.Add(
XMLWorkerHelper.GetCSS(
File.OpenRead(RESOURCES + Path.DirectorySeparatorChar + testPath + Path.DirectorySeparatorChar + testName +
Path.DirectorySeparatorChar + "complexDiv_files" + Path.DirectorySeparatorChar +
"main.css")));
cssFiles.Add(
XMLWorkerHelper.GetCSS(
File.OpenRead(RESOURCES + Path.DirectorySeparatorChar + testPath + Path.DirectorySeparatorChar + testName +
Path.DirectorySeparatorChar + "complexDiv_files" + Path.DirectorySeparatorChar +
"widget082.css")));
StyleAttrCSSResolver cssResolver = new StyleAttrCSSResolver(cssFiles);
HtmlPipelineContext hpc =
new HtmlPipelineContext(
new CssAppliersImpl(new XMLWorkerFontProvider(RESOURCES + @"\tool\xml\examples\fonts")));
hpc.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(Tags.GetHtmlTagProcessorFactory());
hpc.SetImageProvider(new SampleTestImageProvider());
hpc.SetPageSize(doc.PageSize);
HtmlPipeline htmlPipeline = new HtmlPipeline(hpc, new PdfWriterPipeline(doc, pdfWriter));
IPipeline pipeline = new CssResolverPipeline(cssResolver, htmlPipeline);
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser p = new XMLParser(true, worker, Encoding.GetEncoding("UTF-8"));
p.Parse(File.OpenRead(inputHtml), Encoding.GetEncoding("UTF-8"));
//ICC_Profile icc = ICC_Profile.getInstance(ComplexDiv01Test.class.getResourceAsStream("sRGB Color Space Profile.icm"));
//pdfWriter.setOutputIntents("Custom", "", "http://www.color.org", "sRGB IEC61966-2.1", icc);
doc.Close();
}
private static void _add_to_table(Guid applicationId, HttpContext context, PdfPTable mainTable,
StyleAttrCSSResolver cssResolver, WikiPDFText text, Font font, PdfPCell cell = null)
{
if (cell == null)
{
cell = new PdfPCell();
}
cell.Border = 0;
cell.HorizontalAlignment = Element.ALIGN_JUSTIFIED;
cell.RunDirection = text.IsRTL ? PdfWriter.RUN_DIRECTION_RTL : PdfWriter.RUN_DIRECTION_LTR;
cell.SetLeading(0, 2);
itsXmlHtml.DefaultTagProcessorFactory tagProcessors =
(itsXmlHtml.DefaultTagProcessorFactory)itsXmlHtml.Tags.GetHtmlTagProcessorFactory();
tagProcessors.RemoveProcessor(itsXmlHtml.HTML.Tag.IMG); // remove the default processor
tagProcessors.AddProcessor(itsXmlHtml.HTML.Tag.IMG, new CustomImageTagProcessor(applicationId)); // use our new processor
//tagProcessors.AddProcessor(itsXmlHtml.HTML.Tag.TABLE, new TableTagProcessor());
HtmlPipelineContext htmlContext =
new HtmlPipelineContext(new itsXmlHtml.CssAppliersImpl(new IranSansFontProvider(font)));
htmlContext.SetImageProvider(new ImageProvider());
htmlContext.CharSet(Encoding.UTF8);
htmlContext.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(tagProcessors);
ElementsCollector elementsHandler = new ElementsCollector();
CssResolverPipeline pipeline = new CssResolverPipeline(cssResolver,
new HtmlPipeline(htmlContext, new ElementHandlerPipeline(elementsHandler, null)));
itsXml.XMLWorker worker = new itsXml.XMLWorker(pipeline, parseHtml: true);
XMLParser parser = new XMLParser(true, worker, Encoding.UTF8);
parser.Parse(new StringReader(text.Text));
cell.AddElement(elementsHandler.Paragraph);
mainTable.AddCell(cell);
}
private static void AddHTMLText(PdfWriter writer, Document document, string htmlString, List <string> stylesheetFilePath)
{
var cssResolver = new StyleAttrCSSResolver();
Stream cssStream = null;
if (stylesheetFilePath != null)
{
foreach (var stylesheetPath in stylesheetFilePath)
{
cssStream = new FileStream(stylesheetPath, FileMode.Open, FileAccess.Read);
var cssFile = XMLWorkerHelper.GetCSS(cssStream);
cssResolver.AddCss(cssFile);
}
}
var cssAppliersImpl = new CssAppliersImpl();
var htmlPipelineContext = new HtmlPipelineContext(cssAppliersImpl);
htmlPipelineContext.SetTagFactory(Tags.GetHtmlTagProcessorFactory());
htmlPipelineContext.SetImageProvider(new DownloadImageProvider());
var pdfWriterPipeline = new PdfWriterPipeline(document, writer);
var htmlPipe = new HtmlPipeline(htmlPipelineContext, pdfWriterPipeline);
var cssResolverPipeline = new CssResolverPipeline(cssResolver, htmlPipe);
var worker = new XMLWorker(cssResolverPipeline, true);
var parser = new XMLParser(worker);
var stringReader = new StringReader(htmlString);
parser.Parse(stringReader);
if (cssStream != null)
{
cssStream.Dispose();
}
}
static void Main(string[] args)
{
if (args.Length < 2)
{
Console.WriteLine("Invalid number of arguments.");
Console.WriteLine("Usage: html2Pdf.exe [input html_file/directory] [default css file]");
return;
}
List <FileStream> fileList = new List <FileStream>();
if (File.Exists(args[0]))
{
fileList.Add(new FileStream(args[0], FileMode.Open));
}
else if (Directory.Exists(args[0]))
{
CollectHtmlFiles(fileList, args[0]);
}
if (fileList.Count == 0)
{
Console.WriteLine("Invalid html_file/directory");
Console.WriteLine("Usage: html2Pdf.exe [input html_file/directory] [default css file]");
return;
}
foreach (FileStream fileStream in fileList)
{
Document doc = new Document(PageSize.LETTER);
doc.SetMargins(doc.LeftMargin, doc.RightMargin, 35, 0);
String path = Path.GetDirectoryName(Path.GetFullPath(fileStream.Name)) + Path.DirectorySeparatorChar +
Path.GetFileNameWithoutExtension(fileStream.Name) + ".pdf";
PdfWriter pdfWriter = PdfWriter.GetInstance(doc, new FileStream(path, FileMode.Create));
doc.Open();
Dictionary <String, String> substFonts = new Dictionary <String, String>();
substFonts["Arial Unicode MS"] = "Helvetica";
CssFilesImpl cssFiles = new CssFilesImpl();
cssFiles.Add(XMLWorkerHelper.GetCSS(new FileStream(args[1], FileMode.Open)));
StyleAttrCSSResolver cssResolver = new StyleAttrCSSResolver(cssFiles);
HtmlPipelineContext hpc = new HtmlPipelineContext(new CssAppliersImpl(new UnembedFontProvider(XMLWorkerFontProvider.DONTLOOKFORFONTS, substFonts)));
hpc.SetImageProvider(new ImageProvider(args[0]));
hpc.SetAcceptUnknown(true).AutoBookmark(true).SetTagFactory(Tags.GetHtmlTagProcessorFactory());
HtmlPipeline htmlPipeline = new HtmlPipeline(hpc, new PdfWriterPipeline(doc, pdfWriter));
IPipeline pipeline = new CssResolverPipeline(cssResolver, htmlPipeline);
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser xmlParse = new XMLParser(true, worker, null);
xmlParse.Parse(fileStream);
doc.Close();
String cmpPath = Path.GetDirectoryName(Path.GetFullPath(fileStream.Name)) + Path.DirectorySeparatorChar +
"cmp_" + Path.GetFileNameWithoutExtension(fileStream.Name) + ".pdf";
if (File.Exists(cmpPath))
{
CompareTool ct = new CompareTool(path, cmpPath);
String outImage = "<testName>-%03d.png".Replace("<testName>", Path.GetFileNameWithoutExtension(fileStream.Name));
String cmpImage = "cmp_<testName>-%03d.png".Replace("<testName>", Path.GetFileNameWithoutExtension(fileStream.Name));
String diffPath = Path.GetDirectoryName(Path.GetFullPath(fileStream.Name)) +
Path.DirectorySeparatorChar + "diff_" + Path.GetFileNameWithoutExtension(fileStream.Name) + ".png";
String errorMessage = ct.Compare(Path.GetDirectoryName(Path.GetFullPath(fileStream.Name)) + Path.DirectorySeparatorChar + "compare" + Path.DirectorySeparatorChar, diffPath);
if (errorMessage != null)
{
Console.WriteLine(errorMessage);
}
}
}
}
virtual public void SetUp()
{
ctx = new HtmlPipelineContext(null);
ctx.SetImageProvider(new CustomAbstractImageProvider());
clone = (HtmlPipelineContext)ctx.Clone();
}
public bool ConvertToPDF(string inputFile, string outputFile)
{
bool converted = false;
try
{
HtmlDocument doc = new HtmlDocument();
doc.OptionFixNestedTags = true;
doc.OptionWriteEmptyNodes = true;
doc.OptionAutoCloseOnEnd = true;
doc.Load(inputFile);
string rootInner = doc.DocumentNode.InnerHtml;
if (!rootInner.Contains("<html"))
{
doc.DocumentNode.InnerHtml = "<!DOCTYPE html>"
+ "\r\n<html lang=\"en\" xmlns=\"http://www.w3.org/1999/xhtml\">"
+ "\r\n\t<head>"
+ "\r\n\t\t<title>HTML to PDF</title>"
+ "\r\n\t</head>"
+ "\r\n\t<body>\r\n"
+ rootInner
+ "\r\n\t</body>"
+ "\r\n</html>";
}
rootInner = doc.DocumentNode.InnerHtml;
// Remove <meta> tag.
string metaTag = @"<\s*(meta)(\s[^>]*)?>\s*";
while (Regex.IsMatch(rootInner, metaTag))
{
string metaMatch = Regex.Match(rootInner, metaTag).Value;
rootInner = rootInner.Replace(metaMatch, string.Empty);
}
rootInner = rootInner.Replace("</meta>", string.Empty);
// Remove <form> tag.
string formTag = @"<\s*(form)(\s[^>]*)?>\s*";
while (Regex.IsMatch(rootInner, formTag))
{
string formMatch = Regex.Match(rootInner, formTag).Value;
rootInner = rootInner.Replace(formMatch, string.Empty);
}
rootInner = rootInner.Replace("</form>", string.Empty);
// Close br tag.
string brTag = @"<\s*(br)(\s[^>]*)?>";
MatchCollection brMatches = Regex.Matches(rootInner, brTag);
if (brMatches != null)
{
foreach (Match match in brMatches)
{
rootInner = rootInner.Replace(match.Value, "<br />");
}
}
// Replace <font> tag with div.
string fontTag = @"<\s*(font)(\s[^>]*)?>";
while (Regex.IsMatch(rootInner, fontTag))
{
string fontMatch = Regex.Match(rootInner, fontTag).Value;
string toSpan = fontMatch.Replace("font", "div");
rootInner = rootInner.Replace(fontMatch, toSpan);
}
rootInner = rootInner.Replace("</font>", "</div>");
doc.DocumentNode.InnerHtml = rootInner;
// Table elements
var tableNodes = doc.DocumentNode.SelectNodes("//table");
if (tableNodes != null)
{
foreach (HtmlNode node in tableNodes)
{
bool isValidTable = false;
if (node.HasChildNodes)
{
if ((node.SelectSingleNode("thead/tr/th") != null) ||
(node.SelectSingleNode("tbody/tr/td") != null) ||
(node.SelectSingleNode("tr/td") != null))
{
isValidTable = true;
}
}
// Remove invalid table (no tr, td tags).
if (!isValidTable)
{
HtmlNode parent = node.ParentNode;
parent.InnerHtml = " ";
}
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()))
{
if (!styleValue.ToString().EndsWith(";"))
{
styleValue.Append(";");
}
}
if (node.Attributes.Contains("cellspacing"))
{
if (!styleValue.ToString().Contains("border-collapse"))
{
styleValue.Append("border-collapse: collapse; ");
}
node.Attributes.Remove("cellspacing");
}
if (node.Attributes.Contains("cellpadding"))
{
if (!styleValue.ToString().Contains("cellpadding"))
{
styleValue.Append("padding: " + node.Attributes["cellpadding"].Value + "px; ");
}
node.Attributes.Remove("cellpadding");
}
if (node.Attributes.Contains("border"))
{
if (!styleValue.ToString().Contains("border"))
{
styleValue.Append("border: " + node.Attributes["border"].Value + "; ");
}
node.Attributes.Remove("border");
}
if (node.Attributes.Contains("width"))
{
string width = node.Attributes["width"].Value;
if (node.Attributes["width"].Value.EndsWith("%"))
{
width = node.Attributes["width"].Value;
}
else
{
if (node.Attributes["width"].Value.EndsWith("px"))
{
width = node.Attributes["width"].Value;
}
else
{
width = node.Attributes["width"].Value + "px";
}
}
styleValue.Append("width: " + width + "; ");
node.Attributes.Remove("width");
}
if (node.Attributes.Contains("height"))
{
string height = node.Attributes["height"].Value.EndsWith("px") ? node.Attributes["height"].Value : node.Attributes["height"].Value + "px; ";
styleValue.Append("height: " + height);
node.Attributes.Remove("height");
}
if (node.Attributes.Contains("align"))
{
styleValue.Append("text-align: " + node.Attributes["align"].Value + "; ");
node.Attributes.Remove("align");
}
node.Attributes["style"].Value = styleValue.ToString();
}
}
// Remove div from /div/img path.
var imgNodes = doc.DocumentNode.SelectNodes("//div/img");
while (imgNodes != null)
{
foreach (HtmlNode node in imgNodes)
{
node.InnerHtml = " ";
HtmlNode td = node.ParentNode.ParentNode;
td.RemoveChild(node.ParentNode, true);
}
imgNodes = doc.DocumentNode.SelectNodes("//div/img");
}
// Remove div with class="Top_Hidden".
var divHiddenNodes = doc.DocumentNode.SelectNodes("//div[@class='Top_Hidden']");
while (divHiddenNodes != null)
{
foreach (HtmlNode node in divHiddenNodes)
{
HtmlNode tatay = node.ParentNode;
tatay.RemoveChild(node, false);
}
divHiddenNodes = doc.DocumentNode.SelectNodes("//div[@class='Top_Hidden']");
}
// Remove children for div with class="blank".
var divBlankNodes = doc.DocumentNode.SelectNodes("//div[@class='blank']");
if (divBlankNodes != null)
{
foreach (HtmlNode node in divBlankNodes)
{
node.RemoveAllChildren();
}
}
// Close img tag from /td/img path.
var tdImgNodes = doc.DocumentNode.SelectNodes("//td/img");
if (tdImgNodes != null)
{
foreach (HtmlNode node in tdImgNodes)
{
node.InnerHtml = " ";
}
}
// Add style to div tag.
var divNodes = doc.DocumentNode.SelectNodes("//div");
if (divNodes != null)
{
foreach (HtmlNode node in divNodes)
{
if (node.HasAttributes)
{
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()) && !styleValue.ToString().EndsWith(";"))
{
styleValue.Append(";");
}
if (node.Attributes.Contains("face"))
{
string fontFamily = node.Attributes["face"].Value;
styleValue.Append("font-family: " + fontFamily.ToLower() + ";");
node.Attributes.Remove("face");
}
if (node.Attributes.Contains("size"))
{
string fontSize = node.Attributes["size"].Value;
string size = "9pt";
switch (fontSize)
{
case "1":
{
size = "7pt";
break;
}
case "2":
{
size = "9pt";
break;
}
case "3":
{
size = "10pt";
break;
}
case "4":
{
size = "12pt";
break;
}
case "5":
{
size = "16pt";
break;
}
case "6":
{
size = "20pt";
break;
}
case "7":
{
size = "30pt";
break;
}
default:
break;
}
styleValue.Append("font-size: " + size.ToLower() + ";");
node.Attributes.Remove("size");
}
node.Attributes["style"].Value = styleValue.ToString();
}
}
}
// Add td style.
var tdNodes = doc.DocumentNode.SelectNodes("//td");
if (tdNodes != null)
{
foreach (HtmlNode node in tdNodes)
{
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()))
{
if (!styleValue.ToString().EndsWith(";"))
{
styleValue.Append("; ");
}
}
if (node.Attributes.Contains("align"))
{
styleValue.Append("text-align: " + node.Attributes["align"].Value + "; ");
node.Attributes.Remove("align");
}
else
{
styleValue.Append("text-align: left;");
}
if (node.Attributes.Contains("valign"))
{
styleValue.Append("vertical-align: " + node.Attributes["valign"].Value + "; ");
node.Attributes.Remove("valign");
}
if (node.Attributes.Contains("width"))
{
string width = node.Attributes["width"].Value;
if (node.Attributes["width"].Value.EndsWith("%"))
{
width = node.Attributes["width"].Value;
}
else
{
if (node.Attributes["width"].Value.EndsWith("px"))
{
width = node.Attributes["width"].Value;
}
else
{
width = node.Attributes["width"].Value + "px";
}
}
styleValue.Append("width: " + width + "; ");
node.Attributes.Remove("width");
}
if (!string.IsNullOrEmpty(styleValue.ToString()))
{
node.Attributes["style"].Value = styleValue.ToString();
}
}
}
// Add style to p tag.
var pNodes = doc.DocumentNode.SelectNodes("//p");
if (pNodes != null)
{
foreach (HtmlNode node in pNodes)
{
if (node.HasAttributes)
{
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()) && !styleValue.ToString().EndsWith(";"))
{
styleValue.Append(";");
}
if (node.Attributes.Contains("align"))
{
string value = node.Attributes["align"].Value;
styleValue.Append("text-align: " + value.ToLower() + ";");
node.Attributes.Remove("align");
}
node.Attributes["style"].Value = styleValue.ToString();
}
}
}
// Remove u tag from //u/div path but put underline in div tag.
var uDivNodes = doc.DocumentNode.SelectNodes("//u/div");
while (uDivNodes != null)
{
foreach (HtmlNode node in uDivNodes)
{
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()) && !styleValue.ToString().EndsWith(";"))
{
styleValue.Append(";");
}
styleValue.Append("text-decoration: underline;");
node.Attributes["style"].Value = styleValue.ToString();
HtmlNode lolo = node.ParentNode.ParentNode;
lolo.RemoveChild(node.ParentNode, true);
}
uDivNodes = doc.DocumentNode.SelectNodes("//u/div");
}
// Remove strong tag from //strong/div path but put bold in div tag.
var strongDivNodes = doc.DocumentNode.SelectNodes("//strong/div");
while (strongDivNodes != null)
{
foreach (HtmlNode node in strongDivNodes)
{
bool hasStyle = node.Attributes.Contains("style");
if (!hasStyle)
{
node.Attributes.Add("style", string.Empty);
}
StringBuilder styleValue = new StringBuilder(node.Attributes["style"].Value.Trim());
if (!string.IsNullOrEmpty(styleValue.ToString()) && !styleValue.ToString().EndsWith(";"))
{
styleValue.Append(";");
}
styleValue.Append("font-weight: bold;");
node.Attributes["style"].Value = styleValue.ToString();
HtmlNode lolo = node.ParentNode.ParentNode;
lolo.RemoveChild(node.ParentNode, true);
}
strongDivNodes = doc.DocumentNode.SelectNodes("//strong/div");
}
// Replace p tag with div from //p/div path.
var pDivNodes = doc.DocumentNode.SelectNodes("//p/div");
while (pDivNodes != null)
{
foreach (HtmlNode node in pDivNodes)
{
node.ParentNode.Name = "div";
}
pDivNodes = doc.DocumentNode.SelectNodes("//p/div");
}
// Remove div tag from //ol/div path.
var olDivNodes = doc.DocumentNode.SelectNodes("//ol/div");
while (olDivNodes != null)
{
foreach (HtmlNode node in olDivNodes)
{
HtmlNode tatay = node.ParentNode;
bool hasStyle = node.Attributes.Contains("style");
if (hasStyle)
{
tatay.Attributes.Add(node.Attributes["style"]);
}
tatay.RemoveChild(node, true);
}
olDivNodes = doc.DocumentNode.SelectNodes("//ol/div");
}
// Remove div tag from //ul/div path.
var ulDivNodes = doc.DocumentNode.SelectNodes("//ul/div");
while (ulDivNodes != null)
{
foreach (HtmlNode node in ulDivNodes)
{
HtmlNode tatay = node.ParentNode;
bool hasStyle = node.Attributes.Contains("style");
if (hasStyle)
{
tatay.Attributes.Add(node.Attributes["style"]);
}
tatay.RemoveChild(node, true);
}
ulDivNodes = doc.DocumentNode.SelectNodes("//ul/div");
}
// Remove div tag from //li/div path.
var liDivNodes = doc.DocumentNode.SelectNodes("//li/div");
while (liDivNodes != null)
{
foreach (HtmlNode node in liDivNodes)
{
HtmlNode tatay = node.ParentNode;
bool hasStyle = node.Attributes.Contains("style");
if (hasStyle)
{
tatay.Attributes.Add(node.Attributes["style"]);
}
tatay.RemoveChild(node, true);
}
liDivNodes = doc.DocumentNode.SelectNodes("//li/div");
}
// Save the modified html to a new name.
string formattedHTMLFile = outputFile.Replace(".pdf", string.Empty) + "_TEMP.html";
doc.Save(formattedHTMLFile);
HtmlDocument docCPM = new HtmlDocument();
docCPM.OptionFixNestedTags = true;
docCPM.OptionWriteEmptyNodes = true;
docCPM.OptionAutoCloseOnEnd = true;
string newHTML = outputFile.Replace(".pdf", string.Empty) + "_PDF.html";
docCPM.Load(formattedHTMLFile);
docCPM.Save(newHTML);
FontFactory.RegisterDirectories();
Document document = new Document();
PdfWriter writer = PdfWriter.GetInstance(document, new FileStream(outputFile, FileMode.Create));
try
{
document.Open();
HtmlPipelineContext htmlContext = new HtmlPipelineContext(null);
htmlContext.SetTagFactory(Tags.GetHtmlTagProcessorFactory());
CustomImageProvider imageProvider = new CustomImageProvider();
htmlContext.SetImageProvider(imageProvider);
ICSSResolver cssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);
IPipeline pipeline =
new CssResolverPipeline(cssResolver,
new HtmlPipeline(htmlContext,
new PdfWriterPipeline(document, writer)));
XMLWorker worker = new XMLWorker(pipeline, true);
XMLParser p = new XMLParser(worker);
p.Parse(new StreamReader(newHTML));
//XMLWorkerHelper.GetInstance().ParseXHtml(writer, document, new StreamReader(newHTMLFile));
converted = true;
}
catch
{
converted = false;
throw;
}
finally
{
if (document.IsOpen())
{
document.Close();
}
if (writer != null)
{
writer.Close();
}
File.Delete(formattedHTMLFile);
File.Delete(newHTML);
}
}
catch
{
converted = false;
throw;
}
finally
{
}
return(converted);
}