public void ThenSymbolsAreEncodedProperly()
{
this.File.Position = 0;
var doc = new Words.Document(this.File);
Assert.IsTrue(doc.GetText().Contains("строка"), "Wrong conversion");
}
public static void InsertTextInputFormField(string dataDir)
{
// ExStart:DocumentBuilderInsertTextInputFormField
Words.Document doc = new Words.Document();
DocumentBuilder builder = new DocumentBuilder(doc);
builder.InsertTextInput("TextInput", TextFormFieldType.Regular, "", "Hello", 0);
dataDir = dataDir + "DocumentBuilderInsertTextInputFormField_out.doc";
doc.Save(dataDir);
// ExEnd:DocumentBuilderInsertTextInputFormField
Console.WriteLine("\nText input form field using DocumentBuilder inserted successfully into a document.\nFile saved at " + dataDir);
}
// This method is called every time an email is sent.
private void Application_ItemSend(object Item, ref bool Cancel)
{
if (EnableAsposeWordsMetadata)
{
try
{
Outlook.MailItem thisEmail = Item as Outlook.MailItem;
bool flgPassword = false;
for (int i = thisEmail.Attachments.Count; i > 0; i--)
{
Outlook.Attachment attachment = thisEmail.Attachments[i];
// Save the attachment in the temp location.
int attachmentIndex = attachment.Index;
string tempPath = Path.GetTempPath();
string tempFileName = tempPath + attachment.FileName;
attachment.SaveAsFile(tempFileName);
// Check the file format for Microsoft Word documents.
FileFormatInfo info = FileFormatUtil.DetectFileFormat(tempFileName);
bool wordAttachment = false;
switch (info.LoadFormat)
{
case LoadFormat.Doc:
case LoadFormat.Dot:
case LoadFormat.Docx:
case LoadFormat.Docm:
case LoadFormat.Dotx:
case LoadFormat.Dotm:
case LoadFormat.FlatOpc:
case LoadFormat.Rtf:
case LoadFormat.WordML:
case LoadFormat.Html:
case LoadFormat.Mhtml:
case LoadFormat.Odt:
case LoadFormat.Ott:
case LoadFormat.DocPreWord60:
wordAttachment = true;
break;
}
// If a Word Attachment is found:
if (wordAttachment)
{
try
{
Aspose.Words.Document doc = new Words.Document(tempFileName);
// Remove if there is any protection on the document.
ProtectionType protection = doc.ProtectionType;
if (protection != ProtectionType.NoProtection)
{
doc.Unprotect();
}
// Remove all built-in and Custom Properties.
doc.CustomDocumentProperties.Clear();
doc.BuiltInDocumentProperties.Clear();
// Password will be removed if the document is password protected.
if (protection != ProtectionType.NoProtection)
{
doc.Protect(protection);
}
// Save the file back to temp location.
doc.Save(tempFileName);
// Replace the original attachment.
thisEmail.Attachments.Remove(attachmentIndex);
thisEmail.Attachments.Add(tempFileName, missing, attachmentIndex, missing);
}
catch (Words.IncorrectPasswordException ex)
{
flgPassword = true;
}
catch (Exception ex)
{
throw ex;
}
}
if (File.Exists(tempFileName))
{
File.Delete(tempFileName);
}
}
if (flgPassword)
{
MessageBox.Show("Password protected documents cannot be processed for Metadata cleaning");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Error");
Cancel = true;
}
}
}
/// <summary>
/// Converts source file into target format, saves resulted file to out file.
/// Returns null in case of multiple files (all saved into outFolder).
/// </summary>
/// <param name="sourceFile">Source slides file to proceed.</param>
/// <param name="outFolder">Output folder.</param>
/// <param name="format">Output format.</param>
/// <returns>Result file path.</returns>
public string Conversion(
string sourceFile,
string outFolder,
SlidesConversionFormat format
)
{
using (var presentation = new Presentation(sourceFile))
{
var fileName = Path.GetFileNameWithoutExtension(sourceFile);
var outOneFile = Path.Combine(outFolder, $"{fileName}.{format}");
switch (format)
{
case SlidesConversionFormat.odp:
case SlidesConversionFormat.otp:
case SlidesConversionFormat.pptx:
case SlidesConversionFormat.pptm:
case SlidesConversionFormat.potx:
case SlidesConversionFormat.ppt:
case SlidesConversionFormat.pps:
case SlidesConversionFormat.ppsm:
case SlidesConversionFormat.pot:
case SlidesConversionFormat.potm:
case SlidesConversionFormat.pdf:
case SlidesConversionFormat.xps:
case SlidesConversionFormat.ppsx:
case SlidesConversionFormat.tiff:
case SlidesConversionFormat.html:
case SlidesConversionFormat.swf:
var slidesFormat = format.ToString().ParseEnum <SaveFormat>();
presentation.Save(outOneFile, slidesFormat);
return(outOneFile);
case SlidesConversionFormat.txt:
var lines = new List <string>();
foreach (var slide in presentation.Slides)
{
foreach (var shp in slide.Shapes)
{
if (shp is AutoShape ashp)
{
lines.Add(ashp.TextFrame.Text);
}
}
var notes = slide.NotesSlideManager.NotesSlide?.NotesTextFrame?.Text;
if (!string.IsNullOrEmpty(notes))
{
lines.Add(notes);
}
}
System.IO.File.WriteAllLines(outOneFile, lines);
return(outOneFile);
case SlidesConversionFormat.doc:
case SlidesConversionFormat.docx:
using (var stream = new MemoryStream())
{
presentation.Save(stream, SaveFormat.Html);
stream.Flush();
stream.Seek(0, SeekOrigin.Begin);
var doc = new Words.Document(stream);
switch (format)
{
case SlidesConversionFormat.doc:
doc.Save(outOneFile, Words.SaveFormat.Doc);
break;
case SlidesConversionFormat.docx:
doc.Save(outOneFile, Words.SaveFormat.Docx);
break;
default:
throw new ArgumentException($"Unknown format {format}");
}
}
return(outOneFile);
case SlidesConversionFormat.bmp:
case SlidesConversionFormat.jpeg:
case SlidesConversionFormat.png:
case SlidesConversionFormat.emf:
case SlidesConversionFormat.wmf:
case SlidesConversionFormat.gif:
case SlidesConversionFormat.exif:
case SlidesConversionFormat.ico:
ImageFormat GetImageFormat(SlidesConversionFormat f)
{
switch (format)
{
case SlidesConversionFormat.bmp:
return(ImageFormat.Bmp);
case SlidesConversionFormat.jpeg:
return(ImageFormat.Jpeg);
case SlidesConversionFormat.png:
return(ImageFormat.Png);
case SlidesConversionFormat.emf:
return(ImageFormat.Wmf);
case SlidesConversionFormat.wmf:
return(ImageFormat.Wmf);
case SlidesConversionFormat.gif:
return(ImageFormat.Gif);
case SlidesConversionFormat.exif:
return(ImageFormat.Emf);
case SlidesConversionFormat.ico:
return(ImageFormat.Icon);
default:
throw new ArgumentException($"Unknown format {format}");
}
}
///var size = presentation.SlideSize.Size;
for (var i = 0; i < presentation.Slides.Count; i++)
{
var slide = presentation.Slides[i];
var outFile = Path.Combine(outFolder, $"{i}.{format}");
using (var bitmap = slide.GetThumbnail(1, 1)) // (new Size((int)size.Width, (int)size.Height)))
bitmap.Save(outFile, GetImageFormat(format));
}
return(null);
case SlidesConversionFormat.svg:
var svgOptions = new SVGOptions
{
PicturesCompression = PicturesCompression.DocumentResolution
};
for (var i = 0; i < presentation.Slides.Count; i++)
{
var slide = presentation.Slides[i];
var outFile = Path.Combine(outFolder, $"{i}.{format}");
using (var stream = new FileStream(outFile, FileMode.CreateNew))
slide.WriteAsSvg(stream, svgOptions);
}
return(null);
default:
throw new ArgumentException($"Unknown format {format}");
}
}
}
/// <summary>
/// 将Office格式, PDF格式转换成HTML格式
/// </summary>
/// <param name="pathSrc"></param>
/// <param name="pathDst"></param>
/// <returns></returns>
public static string ConvertToHtml(this string pathSrc, string pathDst = null)
{
if (!System.IO.File.Exists(pathSrc))
{
throw new Exception($"指定文件 {pathSrc} 不存在!");
}
var extfilename = System.IO.Path.GetExtension(pathSrc);
if (string.IsNullOrWhiteSpace(pathDst))
{
pathDst = System.IO.Path.GetDirectoryName(pathSrc) + @"\Formated\" + System.IO.Path.GetFileName(pathSrc) + ".html";
}
switch (extfilename)
{
case ".doc":
case ".docx":
var docSrc = new Words.Document(pathSrc);
docSrc.Save(pathDst, Words.SaveFormat.Html);
break;
case ".xls":
case ".xlsx":
Cells.Workbook wb = new Cells.Workbook(pathSrc);
wb.Save(pathDst, new Cells.HtmlSaveOptions(Cells.SaveFormat.Html));
break;
case ".ppt":
case ".pptx":
using (var pres = new Slides.Presentation(pathSrc))
{
var htmlOpt = new Slides.Export.HtmlOptions
{
HtmlFormatter = Slides.Export.HtmlFormatter.CreateDocumentFormatter("", false)
};
pres.Save(pathDst, Slides.Export.SaveFormat.Html, htmlOpt);
}
break;
case ".pdf":
var pdfSrc = new Pdf.Document(pathSrc);
pdfSrc.Save(pathDst, Pdf.SaveFormat.Html);
//pdfSrc.Save(pathDst, new Pdf.HtmlSaveOptions
//{
// FixedLayout = true,
// RasterImagesSavingMode = Pdf.HtmlSaveOptions.RasterImagesSavingModes.AsExternalPngFilesReferencedViaSvg,
// FontSavingMode = Pdf.HtmlSaveOptions.FontSavingModes.SaveInAllFormats,
// // Split HTML output into pages
// SplitCssIntoPages = true,
// // Split css into pages
// SplitIntoPages = true
//});
break;
default:
throw new Exception("不支持的格式");
}
var html = System.IO.File.ReadAllText(pathDst);
html = Regex.Replace(html, "(Evaluation Only\\. Created with Aspose\\.(.+?)\\. Copyright \\d+-\\d+ Aspose Pty Ltd\\.)|(This document was truncated here because it was created in the Evaluation Mode\\.)", "");
System.IO.File.WriteAllText(pathDst, html);
return(pathDst);
}
// This method will be executed everytime an email is sent
private void Application_ItemSend(object Item, ref bool Cancel)
{
if (EnableAsposeWordsMetadata)
{
try
{
Outlook.MailItem thisEmail = Item as Outlook.MailItem;
bool flgPassword = false;
for (int i = thisEmail.Attachments.Count; i > 0; i--)
{
Outlook.Attachment attachment = thisEmail.Attachments[i];
// save attachment in temp location
int attachmentIndex = attachment.Index;
string tempPath = Path.GetTempPath();
string tempFileName = tempPath + attachment.FileName;
attachment.SaveAsFile(tempFileName);
// Check the file format for word documents
FileFormatInfo info = FileFormatUtil.DetectFileFormat(tempFileName);
bool WordAttachment = false;
switch (info.LoadFormat)
{
case LoadFormat.Doc:
case LoadFormat.Dot:
case LoadFormat.Docx:
case LoadFormat.Docm:
case LoadFormat.Dotx:
case LoadFormat.Dotm:
case LoadFormat.FlatOpc:
case LoadFormat.Rtf:
case LoadFormat.WordML:
case LoadFormat.Html:
case LoadFormat.Mhtml:
case LoadFormat.Odt:
case LoadFormat.Ott:
case LoadFormat.DocPreWord97:
WordAttachment = true;
break;
default:
WordAttachment = false;
break;
}
// If word Attachment is found
if (WordAttachment)
{
try
{
Aspose.Words.Document doc = new Words.Document(tempFileName);
// Remove if there is any protection on the document
ProtectionType protection = doc.ProtectionType;
if (protection != ProtectionType.NoProtection)
doc.Unprotect();
// Remove all built-in and Custom Properties
doc.CustomDocumentProperties.Clear();
doc.BuiltInDocumentProperties.Clear();
// Password will be removed if the document is password protected.
if (protection != ProtectionType.NoProtection)
doc.Protect(protection);
// Save the file back to temp location
doc.Save(tempFileName);
// Replace the original attachment
thisEmail.Attachments.Remove(attachmentIndex);
thisEmail.Attachments.Add(tempFileName, missing, attachmentIndex, missing);
}
catch (Words.IncorrectPasswordException ex)
{
flgPassword = true;
}
catch (Exception ex)
{
throw ex;
}
}
// Delete file from temp folder
if (File.Exists(tempFileName))
File.Delete(tempFileName);
}
if (flgPassword)
{
MessageBox.Show("Password protected documents cannot be processed for Metadata cleaning");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Error");
Cancel = true;
}
}
}
public string Conversion(
string sourceFile,
string outFolder,
string format
)
{
using (var presentation = new Presentation(sourceFile))
{
var fileName = Path.GetFileNameWithoutExtension(sourceFile);
var outOneFile = Path.Combine(outFolder, $"{fileName}.{format}");
switch (format)
{
case "odp":
case "otp":
case "pptx":
case "pptm":
case "potx":
case "ppt":
case "pps":
case "ppsm":
case "pot":
case "potm":
case "pdf":
case "xps":
case "ppsx":
case "tiff":
case "html":
case "swf":
var slidesFormat = format.ToString().ParseEnum <SaveFormat>();
presentation.Save(outOneFile, slidesFormat);
return(outOneFile);
case "txt":
var lines = new List <string>();
foreach (var slide in presentation.Slides)
{
foreach (var shp in slide.Shapes)
{
if (shp.Placeholder != null)
{
lines.Add(((AutoShape)shp).TextFrame.Text);
}
}
var notes = slide.NotesSlideManager.NotesSlide?.NotesTextFrame?.Text;
if (!string.IsNullOrEmpty(notes))
{
lines.Add(notes);
}
}
System.IO.File.WriteAllLines(outOneFile, lines);
return(outOneFile);
case "doc":
case "docx":
using (var stream = new MemoryStream())
{
presentation.Save(stream, SaveFormat.Html);
stream.Flush();
stream.Seek(0, SeekOrigin.Begin);
var doc = new Words.Document(stream);
switch (format)
{
case "doc":
doc.Save(outOneFile, Words.SaveFormat.Doc);
break;
case "docx":
doc.Save(outOneFile, Words.SaveFormat.Docx);
break;
default:
throw new ArgumentException($"Unknown format {format}");
}
}
return(outOneFile);
case "bmp":
case "jpeg":
case "png":
case "emf":
case "wmf":
case "gif":
case "exif":
case "ico":
ImageFormat GetImageFormat(string f)
{
switch (format)
{
case "bmp":
return(ImageFormat.Bmp);
case "jpeg":
return(ImageFormat.Jpeg);
case "png":
return(ImageFormat.Png);
case "emf":
return(ImageFormat.Wmf);
case "wmf":
return(ImageFormat.Wmf);
case "gif":
return(ImageFormat.Gif);
case "exif":
return(ImageFormat.Emf);
case "ico":
return(ImageFormat.Icon);
default:
throw new ArgumentException($"Unknown format {format}");
}
}
///var size = presentation.SlideSize.Size;
for (var i = 0; i < presentation.Slides.Count; i++)
{
var slide = presentation.Slides[i];
var outFile = Path.Combine(outFolder, $"{i}.{format}");
using (var bitmap = slide.GetThumbnail(1, 1)) // (new Size((int)size.Width, (int)size.Height)))
bitmap.Save(outFile, GetImageFormat(format));
}
return(null);
case "svg":
var svgOptions = new SVGOptions
{
PicturesCompression = PicturesCompression.DocumentResolution
};
for (var i = 0; i < presentation.Slides.Count; i++)
{
var slide = presentation.Slides[i];
var outFile = Path.Combine(outFolder, $"{i}.{format}");
using (var stream = new FileStream(outFile, FileMode.CreateNew))
slide.WriteAsSvg(stream, svgOptions);
}
return(null);
default:
throw new ArgumentException($"Unknown format {format}");
}
}
}
private void SaveChunk(string outputDir, string chunkName, string fileName, ISlide[] chunkSlides, SlidesConversionFormat format, CancellationToken cancellationToken = default(CancellationToken))
{
if (chunkSlides.Length == 0)
{
return;
}
cancellationToken.ThrowIfCancellationRequested();
var presentation = chunkSlides[0].Presentation;
switch (format)
{
case SlidesConversionFormat.pptx:
case SlidesConversionFormat.pptm:
case SlidesConversionFormat.ppsx:
case SlidesConversionFormat.ppsm:
case SlidesConversionFormat.potx:
case SlidesConversionFormat.potm:
case SlidesConversionFormat.pps:
case SlidesConversionFormat.ppt:
case SlidesConversionFormat.pot:
case SlidesConversionFormat.odp:
case SlidesConversionFormat.otp:
using (var splitPresentation = new Presentation())
{
while (splitPresentation.Slides.Count > 0)
{
splitPresentation.Slides.RemoveAt(0);
}
foreach (var slide in chunkSlides)
{
cancellationToken.ThrowIfCancellationRequested();
splitPresentation.Slides.AddClone(slide);
}
splitPresentation.Save(Path.Combine(outputDir, $"{fileName}_{chunkName}.{format}"), format.ToString().ParseEnum <SaveFormat>());
}
break;
case SlidesConversionFormat.pdf:
case SlidesConversionFormat.xps:
case SlidesConversionFormat.tiff:
case SlidesConversionFormat.html:
case SlidesConversionFormat.swf:
presentation.Save(Path.Combine(outputDir, $"{fileName}_{chunkName}.{format}"),
chunkSlides.Select(s => s.SlideNumber).ToArray(),
format.ToString().ParseEnum <SaveFormat>());
break;
case SlidesConversionFormat.doc:
case SlidesConversionFormat.docx:
using (var stream = new MemoryStream())
{
presentation.Save(stream, chunkSlides.Select(s => s.SlideNumber).ToArray(), SaveFormat.Html);
stream.Flush();
stream.Seek(0, SeekOrigin.Begin);
var doc = new Words.Document(stream);
var wordFormat = format == SlidesConversionFormat.doc
? Words.SaveFormat.Doc
: Words.SaveFormat.Docx;
doc.Save(Path.Combine(outputDir, $"{fileName}_{chunkName}.{format}"), wordFormat);
}
break;
case SlidesConversionFormat.txt:
var lines = new List <string>();
foreach (var slide in chunkSlides)
{
cancellationToken.ThrowIfCancellationRequested();
foreach (var shp in slide.Shapes)
{
if (shp is AutoShape ashp)
{
lines.Add(ashp.TextFrame.Text);
}
}
var notes = slide.NotesSlideManager.NotesSlide?.NotesTextFrame?.Text;
if (!string.IsNullOrEmpty(notes))
{
lines.Add(notes);
}
}
System.IO.File.WriteAllLines(Path.Combine(outputDir, $"{fileName}_{chunkName}.{format}"), lines);
break;
case SlidesConversionFormat.bmp:
case SlidesConversionFormat.jpeg:
case SlidesConversionFormat.png:
case SlidesConversionFormat.emf:
case SlidesConversionFormat.wmf:
case SlidesConversionFormat.gif:
case SlidesConversionFormat.exif:
case SlidesConversionFormat.ico:
foreach (var slide in chunkSlides)
{
cancellationToken.ThrowIfCancellationRequested();
var outFile = Path.Combine(outputDir, $"{fileName}_{slide.SlideNumber:D2}.{format}");
using (var bitmap = slide.GetThumbnail(1, 1))
{
bitmap.Save(outFile, GetImageFormat(format));
}
}
break;
case SlidesConversionFormat.svg:
var svgOptions = new SVGOptions
{
PicturesCompression = PicturesCompression.DocumentResolution
};
foreach (var slide in chunkSlides)
{
cancellationToken.ThrowIfCancellationRequested();
var outFile = Path.Combine(outputDir, $"{fileName}_{slide.SlideNumber:D2}.{format}");
using (var stream = new FileStream(outFile, FileMode.CreateNew))
{
slide.WriteAsSvg(stream, svgOptions);
}
}
break;
default:
throw new ArgumentException($"Unknown format {format}");
}
}