Further compression and png work.

This commit is contained in:
James Grogan 2022-11-23 15:41:33 +00:00
parent 318b481ccc
commit 9c8faa534b
34 changed files with 1164 additions and 203 deletions

View file

@ -0,0 +1,262 @@
#include "DeflateBlock.h"
#include "ByteUtils.h"
#include <algorithm>
#include <iostream>
DeflateBlock::DeflateBlock(BitStream* inputStream, BitStream* outputStream)
: mInputStream(inputStream),
mOutputStream(outputStream)
{
}
bool DeflateBlock::readNextCodeLengthSymbol(unsigned char& final_symbol)
{
unsigned working_index{0};
auto count = mCodeLengthMapping[working_index].first;
auto delta = count;
bool found{false};
unsigned char buffer{0};
unsigned char working_bits{0};
unsigned working_symbol{0};
while(!found)
{
auto valid = mInputStream->readNextNBits(delta, buffer);
working_bits = (working_bits << delta) | buffer;
for(const auto& entry : mCodeLengthMapping[working_index].second)
{
if (entry.first == working_bits)
{
found = true;
working_symbol = entry.second;
break;
}
}
if (!found)
{
working_index++;
if (working_index >= mCodeLengthMapping.size())
{
break;
}
auto new_count = mCodeLengthMapping[working_index].first;
delta = new_count - count;
count = new_count;
}
}
if (found)
{
final_symbol = working_symbol;
std::cout << "Found symbol " << working_symbol << " with bits " << ByteUtils::toString(working_bits) << std::endl;
std::cout << "At Byte offset " << mInputStream->getCurrentByteOffset() << " and bit offset " << mInputStream->getCurrentBitOffset() << std::endl;
return true;
}
else
{
std::cout << "SYMBOL NOT FOUND " << " with bits " << ByteUtils::toString(working_bits) << std::endl;
return false;
}
}
void DeflateBlock::setCodeLengthAlphabetLengths(const std::vector<unsigned char>& lengths)
{
mCodeLengthAlphabetLengths = lengths;
}
void DeflateBlock::setCodeLengthLength(unsigned length)
{
mHclen = length;
}
void DeflateBlock::setLiteralsTableLength(unsigned length)
{
mHlit = length;
}
void DeflateBlock::setDistanceTableLength(unsigned length)
{
mHdist = length;
}
void DeflateBlock::setIsFinalBlock(bool isFinal)
{
}
void DeflateBlock::flushToStream()
{
}
void DeflateBlock::readLiteralCodeLengths()
{
std::vector<unsigned> lengths;
unsigned char symbol{0};
while(lengths.size() < mHlit)
{
bool valid = readNextCodeLengthSymbol(symbol);
if (!valid)
{
std::cout << "Hit unknown symbol - bailing out" << std::endl;
break;
}
if (symbol < 16)
{
lengths.push_back(symbol);
}
else if(symbol == 16)
{
unsigned char num_reps{0};
mInputStream->readNextNBits(2, num_reps);
auto last_val = lengths[lengths.size()-1];
std::cout << "Got val 16 doing " << 3 + num_reps << std::endl;
for(unsigned idx=0; idx< 3 + num_reps; idx++)
{
lengths.push_back(last_val);
}
}
else if(symbol == 17)
{
unsigned char num_reps{0};
mInputStream->readNextNBits(3, num_reps);
std::cout << "Got val 17 doing " << 3 + num_reps << std::endl;
for(unsigned idx=0; idx< 3 + num_reps; idx++)
{
lengths.push_back(0);
}
}
else if(symbol == 18)
{
unsigned char num_reps{0};
mInputStream->readNextNBits(7, num_reps);
std::cout << "Got val 18 doing " << 11 + num_reps << std::endl;
for(unsigned idx=0; idx< 11 + num_reps; idx++)
{
lengths.push_back(0);
}
}
}
}
void DeflateBlock::buildCodeLengthMapping()
{
for(unsigned idx=1; idx<8; idx++)
{
std::vector<unsigned> entries;
for(unsigned jdx=0; jdx<mCodeLengthAlphabetLengths.size(); jdx++)
{
if (mCodeLengthAlphabetLengths[jdx] == idx)
{
entries.push_back(jdx);
}
}
if (entries.empty())
{
continue;
}
CodeLengthCountEntry count_entry{idx, {}};
std::sort(entries.begin(), entries.end());
unsigned char offset = 0x01 << idx - 1;
unsigned char count{0};
for (auto entry : entries)
{
count_entry.second.push_back(CodeLengthEntry{offset + count, entry});
count++;
}
mCodeLengthMapping.push_back(count_entry);
}
for (const auto& map_data : mCodeLengthMapping)
{
std::cout << "Map entry " << map_data.first << " has vals: " << std::endl;
for (const auto& entry : map_data.second)
{
std::cout << "Key " << ByteUtils::toString(entry.first) << " val: " << entry.second << std::endl;
}
}
}
void DeflateBlock::readDynamicHuffmanTable()
{
unsigned char h_lit{0};
mInputStream->readNextNBits(5, h_lit);
mHlit = h_lit + 257;
std::cout << "Got HLIT " << mHlit << std::endl;
unsigned char h_dist{0};
mInputStream->readNextNBits(5, h_dist);
mHdist = h_dist + 1;
std::cout << "Got HDIST " << mHdist << std::endl;
unsigned char h_clen{0};
mInputStream->readNextNBits(4, h_clen);
mHclen = h_clen + 4;
std::cout << "Got HCLEN " << mHclen << std::endl;
mCodeLengthAlphabetLengths = std::vector<unsigned char>(19, 0);
unsigned char buffer{0};
for(unsigned idx = 0; idx< mHclen; idx++)
{
mInputStream->readNextNBits(3, buffer);
mCodeLengthAlphabetLengths[CODE_LENGTH_ALPHABET_PERMUTATION[idx]] = buffer;
std::cout << "Got code length for " << CODE_LENGTH_ALPHABET_PERMUTATION[idx] << " of " << static_cast<unsigned>(buffer) << std::endl;
}
buildCodeLengthMapping();
readLiteralCodeLengths();
}
void DeflateBlock::readHeader()
{
auto working_byte = mInputStream->getCurrentByte();
std::cout << "Into process data "<< std::endl;
std::cout << mInputStream->logNextNBytes(9);
unsigned char final_block{0};
mInputStream->readNextNBits(1, final_block);
mInFinalBlock = bool(final_block);
if (mInFinalBlock)
{
std::cout << "Got final block" << std::endl;
}
mInputStream->readNextNBits(2, mCompressionType);
std::cout << "Compress type byte is: " << static_cast<unsigned>(mCompressionType) << std::endl;
if (mCompressionType == NO_COMPRESSION)
{
std::cout << "Got NO_COMPRESSION" << std::endl;
}
else if (mCompressionType == FIXED_HUFFMAN)
{
std::cout << "Got FIXED_HUFFMAN" << std::endl;
}
else if (mCompressionType == DYNAMIC_HUFFMAN)
{
std::cout << "Got DYNAMIC_HUFFMAN" << std::endl;
readDynamicHuffmanTable();
}
else if (mCompressionType == ERROR)
{
std::cout << "Got ERROR" << std::endl;
}
}