private static List<RemainingBytes> SearchLocalFileForSignaturesBasedOnSize(CompleteSignature sig, MemoryMappedViewAccessor accessor, List<RemainingBytes> remainingByteList, int sigSize, long fileSize, List<BlockSignature> signaturesToReuse)
{
var windowSize = sigSize;
var newRemainingBytes = new List<RemainingBytes>();
var sigDict = GenerateBlockDict(sig);
var buffer = new byte[sigSize];
var offset = 0L;
foreach (var byteRange in remainingByteList)
{
var byteRangeSize = byteRange.EndOffset - byteRange.BeginOffset + 1;
// if byte range is large... and signature size is small (what values???) then dont check.
// We could end up with LOADS of tiny sig matching where ideally we'd use a larger new sig block.
// The exception is when the sig size exactly matches the byterange size... then we allow it to check if the sig will match
// in practice this allows small (1-2 byte sigs) to match the byte ranges.
if (byteRangeSize > 1000 && sigSize > 100 || byteRangeSize == sigSize)
{
// if byteRange is smaller than the key we're using, then there cannot be a match so add
// it to the newRemainingBytes list
if (byteRange.EndOffset - byteRange.BeginOffset + 1 >= windowSize)
{
// search this byterange for all possible keys.
offset = byteRange.BeginOffset;
var generateFreshSig = true;
var bytesRead = 0L;
RollingSignature? currentSig = null;
long oldEndOffset = byteRange.BeginOffset;
do
{
if (generateFreshSig)
{
bytesRead = accessor.ReadArray(offset, buffer, 0, windowSize);
currentSig = CreateRollingSignature(buffer, (int)bytesRead);
}
else
{
// roll existing sig.
var previousByte = accessor.ReadByte(offset - 1);
var nextByte = accessor.ReadByte(offset + windowSize - 1); // Need bounds checking?
currentSig = RollSignature(windowSize, previousByte, nextByte, currentSig.Value);
}
if (sigDict.ContainsKey(currentSig.Value))
{
// populate buffer. Potential waste of IO here.
bytesRead = accessor.ReadArray(offset, buffer, 0, windowSize);
// check md5 sig.
var md5Sig = CreateMD5Signature(buffer, (int)bytesRead);
var sigsForCurrentRollingSig = sigDict[currentSig.Value];
// have a matching md5? If so, we have a match.
var matchingSigs =
sigsForCurrentRollingSig.Where(s => s.MD5Signature.SequenceEqual(md5Sig))
.Select(n => n)
.ToList();
if (matchingSigs.Any())
{
// need to add any byte ranges between oldEndOffset and offset as bytes remaining (ie not part of any sig).
// This is for catching any bytes BEFORE the sig match we've just found.
if (oldEndOffset != offset)
{
newRemainingBytes.Add(new RemainingBytes()
{
BeginOffset = oldEndOffset,
EndOffset = offset - 1
});
}
var matchingSig = matchingSigs[0];
// when storing which existing sig to use, make sure we know the offset in the NEW file it should appear.
matchingSig.Offset = offset;
signaturesToReuse.Add(matchingSig);
offset += windowSize;
generateFreshSig = true;
oldEndOffset = offset;
}
else
{
offset++;
generateFreshSig = false;
}
}
else
{
// no match. Just increment offset and generate rolling sig.
offset++;
generateFreshSig = false;
}
} while (offset + windowSize <= byteRange.EndOffset + 1);
// add remaining bytes to newRemainingBytes list
// Possible to have single byte at end with offset at very last byte.
if (offset <= byteRange.EndOffset)
{
newRemainingBytes.Add(new RemainingBytes()
{
BeginOffset = oldEndOffset,
EndOffset = byteRange.EndOffset
});
}
// if last sig
}
else
{
newRemainingBytes.Add(byteRange);
}
}
else
{
newRemainingBytes.Add(byteRange);
}
}
return newRemainingBytes;
}
private void LoadHeaderAndStreams(PeHeaderReader pe, MemoryMappedViewAccessor mm)
{
var clrDataDir = pe.DataDirectories[PeHeaderReader.Image_Directory_Entry_Type.COM_DESCRIPTOR];
if (Marshal.SizeOf(typeof(IMAGE_COR20_HEADER)) != clrDataDir.Size)
throw new Exception("Size wrong.");
mm.Read<IMAGE_COR20_HEADER>(pe.GetFileOffset(clrDataDir.VirtualAddress), out mHeader);
if (mHeader.cb != clrDataDir.Size)
throw new Exception("Size wrong.");
var metaLoc = pe.GetFileOffset(mHeader.MetaData.VirtualAddress);
mm.Read<MetaDataHeaderPart1>(metaLoc, out mMetaHeader);
var versionBytes = new byte[mMetaHeader.VersionLength];
metaLoc += Marshal.SizeOf(typeof(MetaDataHeaderPart1));
mm.ReadArray<byte>(metaLoc, versionBytes, 0, versionBytes.Length);
int versionSize = 0;
while (versionSize < versionBytes.Length && versionBytes[versionSize] != 0)
versionSize++;
mVersionName = Encoding.ASCII.GetString(versionBytes, 0, versionSize);
metaLoc += mMetaHeader.VersionLength;
mMetaDataFlags = mm.ReadUInt16(metaLoc);
metaLoc += 2;
uint numberOfMetaStreams = mm.ReadUInt16(metaLoc);
metaLoc += 2;
for (int i = 0; i < numberOfMetaStreams; i++)
{
MetaDataStream mds;
mm.Read<MetaDataStream>(metaLoc, out mds);
metaLoc += Marshal.SizeOf(typeof(MetaDataStream));
byte b;
StringBuilder sb = new StringBuilder();
while ((b = mm.ReadByte(metaLoc++)) != 0)
{
sb.Append((char)b);
}
metaLoc += 3;
metaLoc &= ~3;
mMetaStreams.Add(sb.ToString(), mds);
}
}