public void DecompressionProducesExpectedOutput()
{
var decompressed = VbaCompression.Decompress(_expectedCompressedBytes);
Assert.Equal(_expectedDecompressedBytes.Length, decompressed.Length);
Assert.True(_expectedDecompressedBytes.SequenceEqual(decompressed));
}
public void GivenLargeByteSequenceWithLowCompressibilityCompressingAndDecompressionProducesSameInput()
{
var data = GetLargeByteSequenceWithLowCompressibility().ToArray();
var compressedData = VbaCompression.Compress(data);
var convertedData = VbaCompression.Decompress(compressedData);
Assert.True(data != convertedData);
Assert.True(data.LongLength == convertedData.LongLength);
Assert.True(data.SequenceEqual(convertedData));
}
/**
* Create VBA Project from file
*
* @param sPath path of macro enabled Excel, Powerpoint or Word file.
*/
public VBAProject(string sPath)
{
//Initialize pPath
pPath = sPath;
//Get unzip location
unzipTempPath = CreateUniqueTempDirectory();
/* EvilClippy brute forces the file open, It does this in a fairly unclean manner. We will assume the file,
* is named correctly (extension wise). This, although requiring more boiler plate will look a lot cleaner.
*/
// Get extension of file:
string ext = Path.GetExtension(sPath);
//Parse extension and create sOLEFileName
switch (ext)
{
case ".xlsm":
case ".xlam":
isOpenXML = true;
sOLEFileName = Path.Combine(unzipTempPath, "xl", "vbaProject.bin");
break;
case ".docm":
isOpenXML = true;
sOLEFileName = Path.Combine(unzipTempPath, "word", "vbaProject.bin");
break;
case ".pptm":
isOpenXML = true;
sOLEFileName = Path.Combine(unzipTempPath, "ppt", "vbaProject.bin"); //untested
break;
case ".xls":
case ".doc":
case ".ppt":
//Copy path to sOLEFileName to prevent overwriting (we'll overrite this file directly)
sOLEFileName = Path.Combine(unzipTempPath, Path.GetFileName(sPath));
File.Copy(sPath, sOLEFileName);
break;
default:
Console.WriteLine("ERROR: Could not open file " + sPath);
Console.WriteLine("Please make sure this file exists, has a valid extension and is of a valid type.");
Console.WriteLine();
break;
}
//Unzip to unzipTemoPath if isOpenXML. Otherwise create compound file.
if (isOpenXML)
{
ZipFile.ExtractToDirectory(sPath, unzipTempPath);
}
//Create Compound file from VBProject.bin or xls,doc,ppt file:
cf = new CompoundFile(sOLEFileName, CFSUpdateMode.Update, 0);
// Read relevant streams
switch (ext)
{
case ".xlsm":
case ".xlam":
case ".docm":
case ".pptm":
commonStorage = cf.RootStorage;
break;
case ".doc":
commonStorage = cf.RootStorage.GetStorage("Macros"); //Note you can also use `cf.RootStorage.TryGetStorage("Macros")` which returns null if not found.
break;
case ".ppt": //untested - don't know if this is the case for ppts
case ".xls":
commonStorage = cf.RootStorage.GetStorage("_VBA_PROJECT_CUR");
break;
}
vbaProjectStream = commonStorage.GetStorage("VBA").GetStream("_VBA_PROJECT").GetData();
//Contains VBA Project properties, including module names. Remove module names from this stream to hide them.
//PROJECT stream: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/cc848a02-6f87-49a4-ad93-6edb3103f593
//More information: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/d88cb9d8-a475-423d-b370-cc0caaf78628
projectStream = commonStorage.GetStream("project").GetData();
//PROJECTwm stream contains all names of all Modules.
//PROJECTwm stream: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/514acc65-ea7b-4813-aaf7-fabb1bca0ba2
//More Information: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/c458f2e6-f2cc-4c2d-96c7-91a3e63f2fe1
projectwmStream = commonStorage.GetStream("PROJECTwm").GetData();
//Contains names of visible modules and module source code text (compressed)
//VBA Storage: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/b693e0ba-489f-4ac8-ac9d-6387fb5779bb
//VBA StorageInfo: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/170f52a0-4cd6-4729-b51a-d08155cbd213
//Dir Stream: https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-ovba/672d276c-d83c-4452-993b-ca9eca3d8917
dirStream = VbaCompression.Decompress(commonStorage.GetStorage("VBA").GetStream("dir").GetData());
// Read project streams as string
string projectStreamString = System.Text.Encoding.UTF8.GetString(projectStream);
string projectwmStreamString = System.Text.Encoding.UTF8.GetString(projectwmStream);
Console.WriteLine(System.Text.Encoding.UTF8.GetString(projectStream));
Console.WriteLine(System.Text.Encoding.UTF8.GetString(projectwmStream));
// Find all VBA modules in current file
vbaModulesEx = ParseModulesFromDirStream(dirStream);
// Write streams to debug log (if verbosity enabled)
Console.WriteLine("Hex dump of original _VBA_PROJECT stream:\n" + Utils.HexDump(vbaProjectStream));
Console.WriteLine("Hex dump of original dir stream:\n" + Utils.HexDump(dirStream));
}