// Huffman Tree Encoding:
// - Max 14-bit symbol length = One nibble transmitted per symbol
// - 0 means no code
// - 15 is our RLE start symbol
// - An RLE run would go as 3 nibbles: <Symbol> <RLE Command: 15> <Length of run>
// - The lengh of run is transmitted as a nibble.
// - This takes a minimum of 2 nibbles so 3 is the shortest run worth encoding.
// - Therefore the range of the length of the run symbol is 3..18.
// We don't try to do tree elision. The entire chunking thing is entirely an encoder-side
// concept. We can emit a New Tree command whenever we want, so a more serious encoder could take
// much greater efforts to be fancier with when to emit a new huffman tree.
void EncodeHuffTree(HuffmanEncoder huffman)
{
int lastEmittedLength = -1;
for (int i = 0; i < NUM_SYMBOLS;)
{
if (i < NUM_SYMBOLS - 4 &&
huffman.Symbols[i].Length == lastEmittedLength &&
huffman.Symbols[i + 1].Length == lastEmittedLength &&
huffman.Symbols[i + 2].Length == lastEmittedLength)
{
Output.WriteBits(15, 4); // Emit an RLE token
int runLength = 3; // Determine run length
for (; runLength < 18 && i + runLength < NUM_SYMBOLS; runLength++)
{
if (huffman.Symbols[i + runLength].Length != lastEmittedLength)
{
break;
}
}
Output.WriteBits(runLength - 3, 4);
i += runLength;
}
else
{
int length = huffman.Symbols[i].Length;
lastEmittedLength = length;
Output.WriteBits(length, 4);
i++;
}
}
}
void EncodeChunk()
{
var huff = new HuffmanEncoder(SymbolFreqs, 14);
EncodeHuffTree(huff);
for (int i = 0; i < ChunkOffset; i++)
{
int symbol = ChunkBuffer[i];
var sym = huff.Symbols[symbol];
Output.WriteBits(sym.Code, sym.Length);
if (symbol >= SYM_MATCH)
{
var match = MatchEncodes.Dequeue();
Output.WriteBits(match.Distance, match.DistanceBits);
EncodeMatchLength(match.Length);
}
}
}