public static void CloneBlobHeap(IMetadataStream stream, IBinaryStreamWriter writer,
IndexBlobAction indexAction)
{
if (!stream.CanRead)
{
throw new ArgumentException("Stream to clone must be readable.");
}
var reader = stream.CreateReader();
// Only copy first byte to output stream if we have already written something to the output stream.
byte b = reader.ReadByte();
if (writer.Length != 1)
{
writer.WriteByte(b);
}
// Perform linear sweep of the raw data.
// Note: This might result in incorrect blob being indexed if garbage data was injected in the heap.
// This is okay as long as we still copy all the data, including the garbage data.
// The only side-effect we get is that blobs that did appear in the original stream might
// be duplicated in the new stream. This is an acceptable side-effect, as the purpose of this
// import function is to only preserve existing data, and not necessarily make sure that we use
// the most efficient storage mechanism.
uint index = 1;
while (index < stream.GetPhysicalSize())
{
ulong startOffset = reader.Offset;
if (!reader.TryReadCompressedUInt32(out uint dataLength))
{
break;
}
uint headerLength = (uint)(reader.Offset - startOffset);
reader.Offset -= headerLength;
if (dataLength > 0)
{
indexAction(index, writer.Length);
}
// Copy over raw data of blob to output stream.
// This is important since technically it is possible to encode the same blob in multiple ways.
var buffer = new byte[headerLength + dataLength];
int actualLength = reader.ReadBytes(buffer, 0, buffer.Length);
writer.WriteBytes(buffer, 0, actualLength);
// Move to next blob.
index += (uint)actualLength;
}
}
