/// <summary>
/// Saves the <paramref name="data"/> byte array to the <paramref name="outputFile"/>
/// </summary>
/// <param name="data">The stream as byte array</param>
/// <param name="outputFile">The output filename with path</param>
/// <returns></returns>
/// <exception cref="OfficeExtractor.Exceptions.OEFileIsCorrupt">Raised when the file is corrupt</exception>
internal string SaveByteArrayToFile(byte[] data, string outputFile)
{
// Because the data is stored in a stream we have no name for it so we
// have to check the magic bytes to see with what kind of file we are dealing
Logger.WriteToLog($"Saving byte array with length '{data.Length}' to file '{outputFile}'");
var extension = Path.GetExtension(outputFile);
if (string.IsNullOrEmpty(extension))
{
var fileType = FileTypeSelector.GetFileTypeFileInfo(data);
if (fileType != null && !string.IsNullOrEmpty(fileType.Extension))
{
outputFile += "." + fileType.Extension;
}
if (fileType != null)
{
extension = "." + fileType.Extension;
}
}
// Check if the output file already exists and if so make a new one
outputFile = FileManager.FileExistsMakeNew(outputFile);
if (extension != null)
{
switch (extension.ToUpperInvariant())
{
case ".XLS":
case ".XLT":
case ".XLW":
using (var memoryStream = new MemoryStream(data))
using (var compoundFile = new CompoundFile(memoryStream))
{
Excel.SetWorkbookVisibility(compoundFile.RootStorage);
compoundFile.Save(outputFile);
}
break;
case ".XLSB":
case ".XLSM":
case ".XLSX":
case ".XLTM":
case ".XLTX":
using (var memoryStream = new MemoryStream(data))
{
var file = Excel.SetWorkbookVisibility(memoryStream);
File.WriteAllBytes(outputFile, file.ToArray());
}
break;
default:
File.WriteAllBytes(outputFile, data);
break;
}
}
else
{
File.WriteAllBytes(outputFile, data);
}
return(outputFile);
}
/// <summary>
/// This method will extract and save the data from the given <paramref name="storage"/> node to the <paramref name="outputFolder"/>
/// </summary>
/// <param name="storage">The <see cref="CFStorage"/> node</param>
/// <param name="outputFolder">The outputFolder</param>
/// <param name="fileName">The fileName to use, null when the fileName is unknown</param>
/// <returns>Returns the name of the created file that or null if there was nothing to export within the given <paramref name="storage"/> node.</returns>
/// <exception cref="Exceptions.OEFileIsPasswordProtected">Raised when a WordDocument, WorkBook or PowerPoint Document stream is password protected</exception>
public string SaveFromStorageNode(CFStorage storage, string outputFolder, string fileName)
{
Logger.WriteToLog($"Saving CFStorage to output folder '{outputFolder}' with file name {fileName}");
if (storage.TryGetStream("CONTENTS", out var contents))
{
Logger.WriteToLog("CONTENTS stream found");
if (contents.Size <= 0)
{
Logger.WriteToLog("CONTENTS stream is empty");
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
const string delimiter = "%DocumentOle:";
var documentOleFileName = GetDelimitedStringFromData(delimiter, contents.GetData());
if (documentOleFileName != null)
{
if (!documentOleFileName.Equals(string.Empty))
{
fileName = Path.GetFileName(documentOleFileName);
}
contents.SetData(contents.GetData().Skip(delimiter.Length * 2 + documentOleFileName.Length).ToArray());
}
return(SaveByteArrayToFile(contents.GetData(), FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
if (storage.TryGetStream("Package", out var package))
{
Logger.WriteToLog("Package stream found");
if (package.Size <= 0)
{
Logger.WriteToLog("Package stream is empty");
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
return(SaveByteArrayToFile(package.GetData(), FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
if (storage.TryGetStream("EmbeddedOdf", out var embeddedOdf))
{
Logger.WriteToLog("EmbeddedOdf stream found");
// The embedded object is an Embedded ODF file
if (embeddedOdf.Size <= 0)
{
Logger.WriteToLog("EmbeddedOdf stream is empty");
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
return(SaveByteArrayToFile(embeddedOdf.GetData(), FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
if (storage.TryGetStream("\x0001Ole10Native", out _))
{
Logger.WriteToLog("Ole10Native stream found");
var ole10Native = new Ole10Native(storage);
Logger.WriteToLog($"Ole10Native stream format is '{ole10Native.Format}'");
if (ole10Native.Format == OleFormat.File)
{
return(SaveByteArrayToFile(ole10Native.NativeData,
FileManager.FileExistsMakeNew(Path.Combine(outputFolder, ole10Native.FileName))));
}
Logger.WriteToLog("Ole10Native is ignored");
return(null);
}
if (storage.TryGetStream("WordDocument", out _))
{
Logger.WriteToLog("WordDocument stream found");
// The embedded object is a Word file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded Word document.doc";
}
return(SaveStorageTreeToCompoundFile(storage, FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
if (storage.TryGetStream("Workbook", out _))
{
Logger.WriteToLog("Workbook stream found");
// The embedded object is an Excel file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded Excel document.xls";
}
Excel.SetWorkbookVisibility(storage);
return(SaveStorageTreeToCompoundFile(storage, FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
if (storage.TryGetStream("PowerPoint Document", out _))
{
Logger.WriteToLog("PowerPoint Document stream found");
// The embedded object is a PowerPoint file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded PowerPoint document.ppt";
}
return(SaveStorageTreeToCompoundFile(storage, FileManager.FileExistsMakeNew(Path.Combine(outputFolder, fileName))));
}
return(null);
}
