private static void EncodeTree(HuffmanTree tree, ushort[] buffer, ref int bitPosition)
{
if (tree.IsLeaf())
{
// put a 1 bit
buffer[bitPosition / 16] |= (ushort)(0x8000 >> (int)(bitPosition % 16));
bitPosition++;
// append the char code
AppendCode(buffer, (ushort)tree.Character, ref bitPosition);
return;
}
// if it's not a leaf, put a 0 bit
bitPosition++;
EncodeTree(tree.Left, buffer, ref bitPosition);
EncodeTree(tree.Right, buffer, ref bitPosition);
}
// reads and decodes `filename`, returns the decoded text
public static string Decode(string filename)
{
byte[] buffer;
try
{
buffer = File.ReadAllBytes(filename);
}
catch
{
throw;
}
if (buffer.Length == 0)
{
// empty file
return("");
}
// 1. read metadata
// first 6 bits
int fileFormatVersion = buffer[0] & 0xFC;
if (fileFormatVersion > FileFormatVersion)
{
throw new FormatException("File '" + filename + "' has a newer format (file format version " + fileFormatVersion.ToString("00") + "), or is invalid or corrupted.");
}
if (buffer.Length < 8)
{
// 4 bytes for metadata,
// at least 3 bytes for tree (a 1 bit and 16 bits for char code),
// and at least 1 byte for the text
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
int treeLengthInBytes = ((int)(buffer[0] & 0x03) << 16) | ((int)buffer[1] << 8) | ((int)buffer[2]);
int bitsOfLastByteOfTree = buffer[3] >> 4;
int bitsOfLastByteOfText = buffer[3] & 0x0F;
if (bitsOfLastByteOfTree < 1 || bitsOfLastByteOfTree > 8 || bitsOfLastByteOfText < 1 || bitsOfLastByteOfText > 8 || treeLengthInBytes < 3 || treeLengthInBytes > 147456 || buffer.Length < 5 + treeLengthInBytes)
{
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
// 2. decode the tree
HuffmanTree tree = new HuffmanTree();
// the first 32 bits are for metadata
int treeBitPosition = 32;
try
{
DecodeTree(tree, buffer, ref treeBitPosition, (3 + treeLengthInBytes) * 8 + bitsOfLastByteOfTree);
}
catch (FormatException)
{
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
// 3. decode the text
string decodedText = "";
// if the text entirely consists of a single character
if (tree.IsLeaf())
{
for (int i = 4 + treeLengthInBytes; i < buffer.Length; i++)
{
// the huffman code for the only character must be 0
if (buffer[i] != 0)
{
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
}
// length of the huffman code is 1,
// so the number of characters equals the number of bits of the encoded text
int numOfChars = (buffer.Length - treeLengthInBytes - 5) * 8 + bitsOfLastByteOfText;
for (int i = 0; i < numOfChars; i++)
{
decodedText += tree.Character;
}
return(decodedText);
}
HuffmanTree currentNode = tree;
for (int i = 4 + treeLengthInBytes; i < buffer.Length - 1; i++)
{
for (byte mask = 0x80; mask != 0; mask >>= 1)
{
if ((buffer[i] & mask) != 0)
{
// reached bit 1, go right
currentNode = currentNode.Right;
}
else
{
// reached bit 0, go left
currentNode = currentNode.Left;
}
if (currentNode == null)
{
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
if (currentNode.IsLeaf())
{
// reached a leaf, read a character
decodedText += currentNode.Character;
// go to the root
currentNode = tree;
}
}
}
// only some bits of the last byte (bitsOfLastByteOfEncodedText) are real data
for (byte mask = 0x80, count = 0; count < bitsOfLastByteOfText; mask >>= 1, count++)
{
if ((buffer[buffer.Length - 1] & mask) != 0)
{
// reached bit 1, go right
currentNode = currentNode.Right;
}
else
{
// reached bit 0, go left
currentNode = currentNode.Left;
}
if (currentNode == null)
{
throw new FormatException("File '" + filename + "' is invalid or corrupted.");
}
if (currentNode.IsLeaf())
{
// reached a leaf, read a character
decodedText += currentNode.Character;
// go to the root
currentNode = tree;
}
}
return(decodedText);
}