/// <summary>
/// When a Excel document is embedded in for example a Word document the Workbook
/// is set to hidden. Don't know why Microsoft does this but they do. To solve this
/// problem we seek the WINDOW1 record in the BOF record of the stream. In there a
/// gbit structure is located. The first bit in this structure controls the visibility
/// of the workbook, so we check if this bit is set to 1 (hidden) and if so set it to 0.
/// Normally a Workbook stream only contains one WINDOW record but when it is embedded
/// it will contain 2 or more records.
/// </summary>
/// <param name="rootStorage">The <see cref="CFStorage">Root storage</see> of a <see cref="CompoundFile"/></param>
/// <exception cref="OEFileIsCorrupt">Raised when the <paramref name="rootStorage"/> does not have a Workbook stream</exception>
public static void SetWorkbookVisibility(CFStorage rootStorage)
{
if (!rootStorage.ExistsStream("WorkBook"))
{
throw new OEFileIsCorrupt("Could not check workbook visibility because the WorkBook stream is not present");
}
try
{
var stream = rootStorage.GetStream("WorkBook") as CFStream;
if (stream == null)
{
return;
}
var bytes = stream.GetData();
using (var memoryStream = new MemoryStream(bytes))
using (var binaryReader = new BinaryReader(memoryStream))
{
// Get the record type, at the beginning of the stream this should always be the BOF
var recordType = binaryReader.ReadUInt16();
var recordLength = binaryReader.ReadUInt16();
// Something seems to be wrong, we would expect a BOF but for some reason it isn't
if (recordType != 0x809)
{
throw new OEFileIsCorrupt("The file is corrupt");
}
binaryReader.BaseStream.Position += recordLength;
while (binaryReader.BaseStream.Position < binaryReader.BaseStream.Length)
{
recordType = binaryReader.ReadUInt16();
recordLength = binaryReader.ReadUInt16();
// Window1 record (0x3D)
if (recordType == 0x3D)
{
// ReSharper disable UnusedVariable
var xWn = binaryReader.ReadUInt16();
var yWn = binaryReader.ReadUInt16();
var dxWn = binaryReader.ReadUInt16();
var dyWn = binaryReader.ReadUInt16();
// ReSharper restore UnusedVariable
// The grbit contains the bit that hides the sheet
var grbit = binaryReader.ReadBytes(2);
var bitArray = new BitArray(grbit);
// When the bit is set then unset it (bitArray.Get(0) == true)
if (bitArray.Get(0))
{
bitArray.Set(0, false);
// Copy the byte back into the stream, 2 positions back so that we overwrite the old bytes
bitArray.CopyTo(bytes, (int)binaryReader.BaseStream.Position - 2);
}
break;
}
binaryReader.BaseStream.Position += recordLength;
}
}
stream.SetData(bytes);
}
catch (Exception exception)
{
throw new OEFileIsCorrupt(
"Could not check workbook visibility because the file seems to be corrupt", exception);
}
}
/// <summary>
/// This method will extract and save the data from the given <see cref="storage"/> node to the <see cref="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>
/// <exception cref="OEFileIsPasswordProtected">Raised when a WordDocument, WorkBook or PowerPoint Document stream is password protected</exception>
public static string SaveFromStorageNode(CFStorage storage, string outputFolder, string fileName)
{
if (storage.ExistsStream("CONTENTS"))
{
var contents = storage.GetStream("CONTENTS");
if (contents.Size <= 0)
{
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
return(SaveByteArrayToFile(contents.GetData(), Path.Combine(outputFolder, fileName)));
}
if (storage.ExistsStream("Package"))
{
var package = storage.GetStream("Package");
if (package.Size <= 0)
{
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
return(SaveByteArrayToFile(package.GetData(), Path.Combine(outputFolder, fileName)));
}
if (storage.ExistsStream("EmbeddedOdf"))
{
// The embedded object is an Embedded ODF file
var package = storage.GetStream("EmbeddedOdf");
if (package.Size <= 0)
{
return(null);
}
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = DefaultEmbeddedObjectName;
}
return(SaveByteArrayToFile(package.GetData(), Path.Combine(outputFolder, fileName)));
}
if (storage.ExistsStream("\x0001Ole10Native"))
{
var ole10Native = new Ole10Native(storage);
return(ole10Native.Format == OleFormat.File
? SaveByteArrayToFile(ole10Native.NativeData, Path.Combine(outputFolder, ole10Native.FileName))
: null);
}
if (storage.ExistsStream("WordDocument"))
{
// The embedded object is a Word file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded Word document.doc";
}
return(SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName)));
}
if (storage.ExistsStream("Workbook"))
{
// The embedded object is an Excel file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded Excel document.xls";
}
Excel.SetWorkbookVisibility(storage);
return(SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName)));
}
if (storage.ExistsStream("PowerPoint Document"))
{
// The embedded object is a PowerPoint file
if (string.IsNullOrWhiteSpace(fileName))
{
fileName = "Embedded PowerPoint document.ppt";
}
return(SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName)));
}
return(null);
}
/// <summary>
/// This method will extract and save the data from the given <see cref="storage"/> node to the <see cref="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>
/// <exception cref="OEFileIsPasswordProtected">Raised when a WordDocument, WorkBook or PowerPoint Document stream is password protected</exception>
public static string SaveFromStorageNode(CFStorage storage, string outputFolder, string fileName)
{
if (storage.ExistsStream("CONTENTS"))
{
var contents = storage.GetStream("CONTENTS");
if (contents.Size <= 0) return null;
if (string.IsNullOrWhiteSpace(fileName)) fileName = DefaultEmbeddedObjectName;
return SaveByteArrayToFile(contents.GetData(), Path.Combine(outputFolder, fileName));
}
if (storage.ExistsStream("Package"))
{
var package = storage.GetStream("Package");
if (package.Size <= 0) return null;
if (string.IsNullOrWhiteSpace(fileName)) fileName = DefaultEmbeddedObjectName;
return SaveByteArrayToFile(package.GetData(), Path.Combine(outputFolder, fileName));
}
if (storage.ExistsStream("EmbeddedOdf"))
{
// The embedded object is an Embedded ODF file
var package = storage.GetStream("EmbeddedOdf");
if (package.Size <= 0) return null;
if (string.IsNullOrWhiteSpace(fileName)) fileName = DefaultEmbeddedObjectName;
return SaveByteArrayToFile(package.GetData(), Path.Combine(outputFolder, fileName));
}
if (storage.ExistsStream("\x0001Ole10Native"))
{
var ole10Native = new Ole10Native(storage);
return ole10Native.Format == OleFormat.File
? SaveByteArrayToFile(ole10Native.NativeData, Path.Combine(outputFolder, ole10Native.FileName))
: null;
}
if (storage.ExistsStream("WordDocument"))
{
// The embedded object is a Word file
if (string.IsNullOrWhiteSpace(fileName)) fileName = "Embedded Word document.doc";
return SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName));
}
if (storage.ExistsStream("Workbook"))
{
// The embedded object is an Excel file
if (string.IsNullOrWhiteSpace(fileName)) fileName = "Embedded Excel document.xls";
Excel.SetWorkbookVisibility(storage);
return SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName));
}
if (storage.ExistsStream("PowerPoint Document"))
{
// The embedded object is a PowerPoint file
if (string.IsNullOrWhiteSpace(fileName)) fileName = "Embedded PowerPoint document.ppt";
return SaveStorageTreeToCompoundFile(storage, Path.Combine(outputFolder, fileName));
}
return null;
}
/// <summary>
/// When a Excel document is embedded in for example a Word document the Workbook
/// is set to hidden. Don't know why Microsoft does this but they do. To solve this
/// problem we seek the WINDOW1 record in the BOF record of the stream. In there a
/// gbit structure is located. The first bit in this structure controls the visibility
/// of the workbook, so we check if this bit is set to 1 (hidden) en is so set it to 0.
/// Normally a Workbook stream only contains one WINDOW record but when it is embedded
/// it will contain 2 or more records.
/// </summary>
/// <param name="rootStorage">The <see cref="CFStorage">Root storage</see> of a <see cref="CompoundFile"/></param>
/// <exception cref="OEFileIsCorrupt">Raised when the <paramref name="rootStorage"/> does not have a Workbook stream</exception>
public static void SetWorkbookVisibility(CFStorage rootStorage)
{
if (!rootStorage.ExistsStream("WorkBook"))
throw new OEFileIsCorrupt("Could not check workbook visibility because the WorkBook stream is not present");
try
{
var stream = rootStorage.GetStream("WorkBook") as CFStream;
if (stream == null) return;
var bytes = stream.GetData();
using (var memoryStream = new MemoryStream(bytes))
using (var binaryReader = new BinaryReader(memoryStream))
{
// Get the record type, at the beginning of the stream this should always be the BOF
var recordType = binaryReader.ReadUInt16();
var recordLength = binaryReader.ReadUInt16();
// Something seems to be wrong, we would expect a BOF but for some reason it isn't
if (recordType != 0x809)
throw new OEFileIsCorrupt("The file is corrupt");
binaryReader.BaseStream.Position += recordLength;
while (binaryReader.BaseStream.Position < binaryReader.BaseStream.Length)
{
recordType = binaryReader.ReadUInt16();
recordLength = binaryReader.ReadUInt16();
// Window1 record (0x3D)
if (recordType == 0x3D)
{
// ReSharper disable UnusedVariable
var xWn = binaryReader.ReadUInt16();
var yWn = binaryReader.ReadUInt16();
var dxWn = binaryReader.ReadUInt16();
var dyWn = binaryReader.ReadUInt16();
// ReSharper restore UnusedVariable
// The grbit contains the bit that hides the sheet
var grbit = binaryReader.ReadBytes(2);
var bitArray = new BitArray(grbit);
// When the bit is set then unset it (bitArray.Get(0) == true)
if (bitArray.Get(0))
{
bitArray.Set(0, false);
// Copy the byte back into the stream, 2 positions back so that we overwrite the old bytes
bitArray.CopyTo(bytes, (int)binaryReader.BaseStream.Position - 2);
}
break;
}
binaryReader.BaseStream.Position += recordLength;
}
}
stream.SetData(bytes);
}
catch (Exception exception)
{
throw new OEFileIsCorrupt(
"Could not check workbook visibility because the file seems to be corrupt", exception);
}
}
