/// <summary>
/// Using iTextSharp's limited HTML to PDF capabilities.
/// </summary>
public PdfPCell RenderHtml()
{
if (PdfFont == null)
{
throw new NullReferenceException($"{nameof(PdfFont)} of the HTML section is null.");
}
var pdfCell = new PdfPCell
{
UseAscender = true,
UseDescender = true,
VerticalAlignment = Element.ALIGN_MIDDLE
};
applyStyleSheet();
var props = new Hashtable
{
{ "img_provider", new HtmlImageFactory(this.PdfElement) },
{ "font_factory", new UnicodeFontProvider(PdfFont.Fonts[0]) } // Always use Unicode fonts
};
using (var reader = new StringReader(Html))
{
var parsedHtmlElements = HtmlWorker.ParseToList(reader, StyleSheet, props);
foreach (IElement htmlElement in parsedHtmlElements)
{
applyRtlRunDirection(htmlElement);
pdfCell.AddElement(htmlElement);
}
return(pdfCell);
}
}
static void Main(string[] args)
{
Console.WriteLine("ВВедите url адрес");
string url = Console.ReadLine();
HtmlWorker download = new HtmlWorker(url);
download.Dispose();
}
public void Verify_Html_To_Pdf_With_colspan_CanBeCreated()
{
var pdfFilePath = TestUtils.GetOutputFileName();
var stream = new FileStream(pdfFilePath, FileMode.Create);
// create a StyleSheet
var styles = new StyleSheet();
// set the default font's properties
styles.LoadTagStyle(HtmlTags.BODY, "encoding", "Identity-H");
styles.LoadTagStyle(HtmlTags.BODY, HtmlTags.FONT, "Tahoma");
styles.LoadTagStyle(HtmlTags.BODY, "size", "16pt");
FontFactory.Register(TestUtils.GetTahomaFontPath());
var unicodeFontProvider = FontFactoryImp.Instance;
unicodeFontProvider.DefaultEmbedding = BaseFont.EMBEDDED;
unicodeFontProvider.DefaultEncoding = BaseFont.IDENTITY_H;
var props = new Hashtable
{
{ "font_factory", unicodeFontProvider }
};
var document = new Document();
PdfWriter.GetInstance(document, stream);
document.AddAuthor(TestUtils.Author);
document.Open();
var objects = HtmlWorker.ParseToList(
new StringReader(@"<table style='width: 100%;font-size:9pt;font-family:Tahoma;'>
<tr>
<td align='center' bgcolor='#00FF00' colspan='2'>Hello</td>
</tr>
<tr>
<td align = 'right'> Month </td>
<td align = 'right'> Savings </td>
</tr>
<tr>
<td align = 'right'> January </td>
<td align = 'right'>$100 </td>
</tr>
</table>"),
styles,
props
);
foreach (IElement element in objects)
{
document.Add(element);
}
document.Close();
stream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
/// <summary>
/// Parse html to elements and adds elements to the Pdf
/// </summary>
/// <param name="html"><see cref="System.String"/>The HTML as string</param>
public Pdf AddHtml(string html)
{
using (var reader = new StringReader(html))
{
var parsedHtmlElements = HtmlWorker.ParseToList(reader, new StyleSheet());
foreach (IElement e in parsedHtmlElements)
{
this._elements.Add(new Builder <IElement>(e));
}
}
return(this);
}
public void Verify_Unicode_Html_To_Pdf_CanBeCreated()
{
var pdfFilePath = TestUtils.GetOutputFileName();
var stream = new FileStream(pdfFilePath, FileMode.Create);
// create a StyleSheet
var styles = new StyleSheet();
// set the default font's properties
styles.LoadTagStyle(HtmlTags.BODY, "encoding", "Identity-H");
styles.LoadTagStyle(HtmlTags.BODY, HtmlTags.FONT, "Tahoma");
styles.LoadTagStyle(HtmlTags.BODY, "size", "16pt");
FontFactory.Register(TestUtils.GetTahomaFontPath());
var unicodeFontProvider = FontFactoryImp.Instance;
unicodeFontProvider.DefaultEmbedding = BaseFont.EMBEDDED;
unicodeFontProvider.DefaultEncoding = BaseFont.IDENTITY_H;
var props = new Hashtable
{
{ "img_provider", new MyImageFactory() },
{ "font_factory", unicodeFontProvider } // Always use Unicode fonts
};
// step 1
var document = new Document();
// step 2
PdfWriter.GetInstance(document, stream);
// step 3
document.AddAuthor(TestUtils.Author);
document.Open();
// step 4
var objects = HtmlWorker.ParseToList(
new StringReader(createHtmlSnippet()),
styles,
props
);
foreach (IElement element in objects)
{
document.Add(element);
}
document.Close();
stream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
// Renders the PDF to an array of bytes
public async Task <MemoryStream> RenderPdf()
{
// Document is built-in class available in iTextSharp
var stream = new MemoryStream();
var document = new Document(_size);
var writer = PdfWriter.GetInstance(document, stream);
// Open document
document.Open();
// Render each page that has been added
foreach (string page in Pages)
{
document.NewPage();
// Generate this page of text
using (var ms = new MemoryStream()) {
using (var sw = new StreamWriter(ms, Encoding.UTF8)) {
await sw.WriteAsync(page); // Get the page output
}
// Read the created stream
using (MemoryStream generate = new MemoryStream(ms.ToArray())) {
using (StreamReader reader = new StreamReader(generate)) {
var providers = new Hashtable();
int i = -1;
foreach (IElement el in HtmlWorker.ParseToList(reader, _styles, providers))
{
if (el is iTextSharp.text.pdf.PdfPTable tbl)
{
i++;
if (i < _colWidths.Count && _colWidths[i].Length == tbl.NumberOfColumns)
{
tbl.SetWidths(_colWidths[i]);
}
}
document.Add(el);
}
}
}
}
}
// Close document
document.Close();
// Return the rendered PDF
stream.Seek(0, SeekOrigin.Begin);
return(stream);
}
public void Verify_Issue24_CanBe_Processed()
{
var pdfFilePath = TestUtils.GetOutputFileName();
var stream = new FileStream(pdfFilePath, FileMode.Create);
// create a StyleSheet
var styles = new StyleSheet();
// set the default font's properties
styles.LoadTagStyle(HtmlTags.BODY, "encoding", "Identity-H");
styles.LoadTagStyle(HtmlTags.BODY, HtmlTags.FONT, "Tahoma");
styles.LoadTagStyle(HtmlTags.BODY, "size", "16pt");
FontFactory.Register(TestUtils.GetTahomaFontPath());
var unicodeFontProvider = FontFactoryImp.Instance;
unicodeFontProvider.DefaultEmbedding = BaseFont.EMBEDDED;
unicodeFontProvider.DefaultEncoding = BaseFont.IDENTITY_H;
var props = new Hashtable
{
{ "font_factory", unicodeFontProvider }
};
var document = new Document();
PdfWriter.GetInstance(document, stream);
document.AddAuthor(TestUtils.Author);
document.Open();
var objects = HtmlWorker.ParseToList(
new StringReader(@"<b>This is a test</b>
<br/>
<span style='color:blue;font-size:20pt;font-family:tahoma;font-style:italic;font-weight:bold'>
<b>Hi<b/>
</span>"),
styles,
props
);
foreach (IElement element in objects)
{
document.Add(element);
}
document.Close();
stream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
public void Verify_Base64_Images_CanBe_Processed()
{
var pdfDoc = new Document(PageSize.A4);
var pdfFilePath = TestUtils.GetOutputFileName();
var fileStream = new FileStream(pdfFilePath, FileMode.Create);
PdfWriter.GetInstance(pdfDoc, fileStream);
pdfDoc.AddAuthor(TestUtils.Author);
pdfDoc.Open();
FontFactory.Register(TestUtils.GetTahomaFontPath());
var html = "<img alt='' src='data:image/png;base64,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' />";
var styles = new StyleSheet();
PdfPCell pdfCell = new PdfPCell
{
Border = 0,
RunDirection = PdfWriter.RUN_DIRECTION_LTR
};
using (var reader = new StringReader(html))
{
var parsedHtmlElements = HtmlWorker.ParseToList(reader, styles);
foreach (IElement htmlElement in parsedHtmlElements)
{
pdfCell.AddElement(htmlElement);
}
}
var table1 = new PdfPTable(1);
table1.AddCell(pdfCell);
pdfDoc.Add(table1);
pdfDoc.Close();
fileStream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
public void Verify_Issue24_CanBe_Processed()
{
var pdfFilePath = TestUtils.GetOutputFileName();
var stream = new FileStream(pdfFilePath, FileMode.Create);
// create a StyleSheet
var styles = new StyleSheet();
// set the default font's properties
styles.LoadTagStyle(HtmlTags.BODY, "encoding", "Identity-H");
styles.LoadTagStyle(HtmlTags.BODY, HtmlTags.FONT, "Tahoma");
styles.LoadTagStyle(HtmlTags.BODY, "size", "16pt");
FontFactory.Register(TestUtils.GetTahomaFontPath());
var props = new Hashtable
{
{ "font_factory", new UnicodeFontProvider() } // Always use Unicode fonts --> It's defined here: https://github.com/VahidN/iTextSharp.LGPLv2.Core/blob/master/src/iTextSharp.LGPLv2.Core.FunctionalTests/HtmlWorkerTests.cs#L117
};
var document = new Document();
PdfWriter.GetInstance(document, stream);
document.AddAuthor(TestUtils.Author);
document.Open();
var objects = HtmlWorker.ParseToList(
new StringReader(@"<b>This is a test</b>
<br/>
<span style='color:blue;font-size:20pt;font-family:tahoma;font-style:italic;font-weight:bold'>
<b>Hi<b/>
</span>"),
styles,
props
);
foreach (IElement element in objects)
{
document.Add(element);
}
document.Close();
stream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
public void Verify_Html_To_Pdf_CanBeCreated()
{
var pdfFilePath = TestUtils.GetOutputFileName();
var stream = new FileStream(pdfFilePath, FileMode.Create);
// create a StyleSheet
var styles = new StyleSheet();
styles.LoadTagStyle("ul", "indent", "10");
styles.LoadTagStyle("li", "leading", "14");
styles.LoadStyle("country", "i", "");
styles.LoadStyle("country", "color", "#008080");
styles.LoadStyle("director", "b", "");
styles.LoadStyle("director", "color", "midnightblue");
var props = new Hashtable
{
{ "img_provider", new MyImageFactory() }
};
// step 1
var document = new Document();
// step 2
PdfWriter.GetInstance(document, stream);
// step 3
document.AddAuthor(TestUtils.Author);
document.Open();
// step 4
var objects = HtmlWorker.ParseToList(
new StringReader(createHtmlSnippet()),
styles,
props
);
foreach (IElement element in objects)
{
document.Add(element);
}
document.Close();
stream.Dispose();
TestUtils.VerifyPdfFileIsReadable(pdfFilePath);
}
protected FileDto CreatePdfPackage(string fileName, string htmlContent)
{
var file = new FileDto(fileName, MimeTypeNames.ApplicationPdf);
var filePath = Path.Combine(AppFolders.TempFileDownloadFolder, file.FileToken);
if (!Directory.Exists(AppFolders.TempFileDownloadFolder))
{
Directory.CreateDirectory(AppFolders.TempFileDownloadFolder);
}
var pdfDoc = new Document(PageSize.A4, 5f, 5f, 5f, 5f);
var fileStream = new FileStream(filePath, FileMode.Create);
PdfWriter.GetInstance(pdfDoc, fileStream);
pdfDoc.Open();
var styles = new StyleSheet();
PdfPCell pdfCell = new PdfPCell
{
Border = 0,
RunDirection = PdfWriter.RUN_DIRECTION_LTR
};
using (var reader = new StringReader(htmlContent))
{
var parsedHtmlElements = HtmlWorker.ParseToList(reader, styles);
foreach (IElement htmlElement in parsedHtmlElements)
{
pdfCell.AddElement(htmlElement);
}
}
var table1 = new PdfPTable(1);
table1.AddCell(pdfCell);
pdfDoc.Add(table1);
pdfDoc.Close();
fileStream.Dispose();
return(file);
}
public byte[] GeneratePdfFromHtml(string html)
{
var pdfDoc = new Document(PageSize.A4, 10f, 10f, 10f, 0f);
var htmlparser = new HtmlWorker(pdfDoc);
using (MemoryStream memoryStream = new MemoryStream())
{
PdfWriter writer = PdfWriter.GetInstance(pdfDoc, memoryStream);
pdfDoc.Open();
StringReader sr = new StringReader(html);
htmlparser.Parse(sr);
pdfDoc.Close();
return(memoryStream.ToArray());
//byte[] bytes = memoryStream.ToArray();
//memoryStream.Close();
}
}
private void CreatePdfNormalText(string html, string path)
{
TextReader reader = new StringReader(html);
Document document = new Document(PageSize.A4, 30, 30, 30, 30);
PdfWriter writer = PdfWriter.GetInstance(document, new FileStream("my.pdf", FileMode.Create));
/*
* Sprawdzić jak z Downloadem dla roznych systemow np. dla telefonu
*/
HtmlWorker worker = new HtmlWorker(document);
document.Open();
var bigFont = FontFactory.GetFont(BaseFont.TIMES_ROMAN, BaseFont.CP1257, 18, Font.BOLD);
var para = new Paragraph(html, bigFont);
document.Add(para);
document.Close();
}
public byte[] GeneratePdfGromHtml(string html)
{
var pdfDoc = new Document(PageSize.A4, 10f, 10f, 10f, 0f);
var htmlParser = new HtmlWorker(pdfDoc);
using (var memoryStream = new MemoryStream())
{
var writer = PdfWriter.GetInstance(pdfDoc, memoryStream);
pdfDoc.Open();
using (var stringReader = new StringReader(html))
{
htmlParser.Parse(stringReader);
}
pdfDoc.Close();
return(memoryStream.ToArray());
}
}
public byte[] Generate(string htmlString)
{
Document pdfDoc = new Document(PageSize.A4, 10f, 10f, 10f, 0f);
var htmlparser = new HtmlWorker(pdfDoc);
using (MemoryStream memoryStream = new MemoryStream())
{
PdfWriter writer = PdfWriter.GetInstance(pdfDoc, memoryStream);
pdfDoc.Open();
using (var sr = new StringReader(htmlString))
{
htmlparser.Parse(sr);
}
pdfDoc.Close();
byte[] bytes = memoryStream.ToArray();
return(bytes);
}
}
public byte[] CreateCertificateInPdf(string html)
{
Document pdfDocument = new Document(PageSize.A4);
HtmlWorker worker = new HtmlWorker(pdfDocument);
using (var mem = new MemoryStream())
{
PdfWriter writer = PdfWriter.GetInstance(pdfDocument, mem);
pdfDocument.Open();
using (var stringReader = new StringReader(html))
{
worker.Parse(stringReader);
}
pdfDocument.Close();
return(mem.ToArray());
}
}
public byte[] CreatePdf(OrderViewModel order, List <CartProduct> cList, string email)
{
var html = BuildEmailContent(order, cList, email);
byte[] bytes = null;
StringReader sr = new StringReader(html.ToString());
Document pdfDoc = new Document(PageSize.A4, 10f, 10f, 10f, 0f);
HtmlWorker htmlparser = new HtmlWorker(pdfDoc);
using (MemoryStream memoryStream = new MemoryStream())
{
PdfWriter writer = PdfWriter.GetInstance(pdfDoc, memoryStream);
pdfDoc.Open();
htmlparser.Parse(sr);
pdfDoc.Close();
bytes = memoryStream.ToArray();
memoryStream.Close();
}
return(bytes);
}
public byte[] GeneratePdfFromHtl(string html)
{
Document pdf = new Document(PageSize.A4, 10f, 10f, 10f, 0f);
HtmlWorker htmlParser = new HtmlWorker(pdf);
using (MemoryStream memoryStream = new MemoryStream())
{
PdfWriter writer = PdfWriter.GetInstance(pdf, memoryStream);
pdf.Open();
using (StringReader streamReader = new StringReader(html))
{
htmlParser.Parse(streamReader);
}
pdf.Close();
byte[] bytes = memoryStream.ToArray();
return(bytes);
}
}
private byte[] GetPDF(string pHTML)
{
byte[] bPDF = null;
MemoryStream ms = new MemoryStream();
TextReader txtReader = new StringReader(pHTML);
Document doc = new Document(PageSize.Letter, 30, 30, 30, 30);
PdfWriter oPdfWriter = PdfWriter.GetInstance(doc, ms);
doc.Open();
var xmlWorker = HtmlWorker.ParseToList(txtReader, new StyleSheet());
foreach (Paragraph item in xmlWorker)
{
if (item.Content.Contains("Semana:"))
{
doc.NewPage();
}
doc.Add(item);
}
doc.Close();
bPDF = ms.ToArray();
return(bPDF);
}
public byte[] ExportPDF(string html)
{
StringReader stringReader = new StringReader(html);
Document pdfDoc = new Document(PageSize.A4);
HtmlWorker htmlparser = new HtmlWorker(pdfDoc);
MemoryStream memoryStream = new MemoryStream();
PdfWriter writer = PdfWriter.GetInstance(pdfDoc, memoryStream);
pdfDoc.Open();
htmlparser.Parse(stringReader);
pdfDoc.Close();
byte[] bytes = memoryStream.ToArray();
memoryStream.Close();
return(bytes != null ? bytes : new byte[] { });
}
static async Task Main(string[] args)
{
Console.WriteLine("Введите ссылку: ");
var input = Console.ReadLine()?.Trim();
var filesProvider = new FilesProvider();
if (input != string.Empty)
{
var indexHandler = new IndexerHandler();
var worker = new HtmlWorker(input);
await filesProvider.WriteAllData(worker.GetPages());
indexHandler.IndexWords(worker.HtmlInfos);
await filesProvider.WriteInvertedList(indexHandler.IndexedWords);
var tfIdfInfo = indexHandler.IndexedWords
.Select(x => new { x.Key, Value = x.Value.Distinct() })
.Select(x =>
new WordFrequencyInfo
{
Word = x.Key,
IDF = Math.Log10(worker.HtmlInfos.Count / (double)x.Value.Count()),
TF = x.Value.ToDictionary(q => q, q =>
{
var docContentInfo = worker.HtmlInfos.FirstOrDefault(z => z.Level == q)?.Content;
var wordsCount =
(double)docContentInfo.WordsInfo.Count(z => z.LemmatizedWord == x.Key);
return(wordsCount / docContentInfo.WordsCount);
})
});
await filesProvider.WriteTfIdf(tfIdfInfo);
}
else
{
var indexedWords = await filesProvider.ReadInvertedList();
Console.WriteLine("Введите запрос: ");
var searchQuery = Console.ReadLine();
//searchQuery = "Android OR fb OR Биробиджан";
var query = QueryWorker.ParseQuery(searchQuery);
var searchResult = indexedWords.Where(query.Compile()).ToList();
var words = searchQuery.Split(' ');
var result = QueryWorker.GetDocNumbers(words, searchResult);
Console.WriteLine("Результаты:");
foreach (var item in result)
{
Console.Write(item + " ");
}
}
var sourceTfIdfList = await filesProvider.ReadTfIdfList();
var searchQueryTfIdf = filesProvider.GetSearchQueryTfIdf(sourceTfIdfList, new[] { "ГАБДУЛИНА" }).ToList();
filesProvider.GetRelevantDocs(await filesProvider.ReadTfIdfList(), searchQueryTfIdf);
Console.WriteLine();
}