aboutsummaryrefslogtreecommitdiff
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/PNGImage.cpp377
-rw-r--r--src/PNGImage.h60
-rw-r--r--src/debug.h3
-rw-r--r--src/image.cpp10
-rw-r--r--src/image.h51
-rw-r--r--src/image.hpp0
-rw-r--r--src/puff.cpp840
-rw-r--r--src/puff.h35
-rw-r--r--src/reader.cpp110
-rw-r--r--src/reader.h33
-rw-r--r--src/zlib.cpp356
-rw-r--r--src/zlib.h52
12 files changed, 1927 insertions, 0 deletions
diff --git a/src/PNGImage.cpp b/src/PNGImage.cpp
new file mode 100644
index 0000000..b7f28ce
--- /dev/null
+++ b/src/PNGImage.cpp
@@ -0,0 +1,377 @@
+#include "PNGImage.h"
+#include "zlib.h"
+#include "puff.h"
+#include <bitset>
+#include <cctype>
+#include <cstdint>
+#include <cstdio>
+#include <cstdlib>
+#include <cstring>
+#include <exception>
+#include <iostream>
+#include <array>
+#include <stdexcept>
+
+using std::cout, std::endl;
+
+PNGImage::PNGImage(std::string filename)
+ :reader(filename)
+ ,idatData()
+{
+ //cout << "Reader good" << endl;
+ REGISTER_CHUNK_READER(IHDR);
+ REGISTER_CHUNK_READER(iCCP);
+ REGISTER_CHUNK_READER(sRGB);
+ REGISTER_CHUNK_READER(eXIf);
+ REGISTER_CHUNK_READER(iDOT);
+ REGISTER_CHUNK_READER(pHYs);
+ REGISTER_CHUNK_READER(tIME);
+ REGISTER_CHUNK_READER(tEXt);
+ REGISTER_CHUNK_READER(IDAT);
+ REGISTER_CHUNK_READER(IEND);
+
+ //cout << "Chunk readers loaded" << endl;
+
+ char signature[8];
+ uint8_t expected[] = {137, 80, 78, 71, 13, 10, 26, 10};
+ reader.readBytes(signature, 8);
+ if(strncmp(signature, (char*)expected, 8) != 0)
+ cout << "UH OH" << endl;
+
+ idatData = nullptr;
+ idatDataSize = 0;
+
+ //cout << "PNG image initialised" << endl;
+}
+PNGImage::~PNGImage()
+{
+ free(idatData);
+ idatData = nullptr;
+ idatDataSize = 0;
+}
+
+bool PNGImage::readNextChunk()
+{
+ if(end)
+ return false;
+ uint32_t chunkSize = reader.readData<uint32_t>();
+
+ char chunkType[4];
+ reader.readBytes(chunkType, 4);
+ std::string chunkName(chunkType, 4);
+ cout << "-------------" << endl;
+ cout << "|Chunk: " << chunkName << "|" << endl;
+ cout << "-------------" << endl;
+
+ if(chunkReaders.count(chunkName) == 0)
+ {
+ cout << "Chunk reader not found!!!" << endl;
+ reader.skipBytes(chunkSize + 4);
+ if(islower(chunkType[0]))
+ {
+ cout << "\tAble to skip chunk" << endl;
+ return true;
+ }
+ cout << "\tFatal error" << endl;
+ return false;
+ }
+
+ void(PNGImage::*chunkReader)(uint32_t chunkSize) = chunkReaders.find(chunkName)->second;
+ (this->*chunkReader)(chunkSize);
+
+ reader.skipBytes(4); // CRC
+
+ return true;
+}
+
+DEFINE_CHUNK_READER(IHDR)
+{
+ width = reader.readData<uint32_t>();
+ height = reader.readData<uint32_t>();
+ bitDepth = reader.readData<uint8_t>();
+ colorType = reader.readData<uint8_t>();
+ compressionMethod = reader.readData<uint8_t>();
+ filterMethod = reader.readData<uint8_t>();
+ interlaceMethod = reader.readData<uint8_t>();
+ cout << "Width: " << width << ", Height: " << height << ", Bit depth: " << 0+bitDepth << ", Color type: " << 0+colorType << ", Compression method: " << 0+compressionMethod << ", Filter method: " << 0+filterMethod << ", Interlace method: " << 0+interlaceMethod << endl;
+
+ if(colorType != 2 && colorType != 6)
+ throw std::invalid_argument("Only color types 2 and 6 are supported");
+
+ switch(colorType)
+ {
+ case 2: colorValues = 3; break;
+ case 6: colorValues = 4; break;
+ }
+
+
+ bpp = colorValues * (bitDepth/8);
+
+ unsigned long imageDataSize = width * height * bpp;
+
+ cout << "Assigning " << imageDataSize << " bytes for image" << endl;
+
+ imageData = new uint8_t[imageDataSize];
+
+/*
+ Scanline<RGBPixel<uint8_t>>* lines = new Scanline<RGBPixel<uint8_t>> [height];
+ for(int i = 0; i < height; i++)
+ {
+ lines[i].pixels = new RGBPixel<uint8_t>[width];
+ }
+ imageData = (uint8_t*)lines;
+*/
+}
+
+DEFINE_CHUNK_READER(iCCP)
+{
+ cout << "!!! iCCP chunk reader not finished !!!" << endl;
+ std::string profileName;
+ char c = reader.readByte();
+ chunkSize--;
+ while(c != 0)
+ {
+ profileName.push_back(c);
+ c = reader.readByte();
+ chunkSize--;
+ }
+ cout << profileName << endl;
+ uint8_t compresssionMethod = reader.readByte();
+ chunkSize--;
+ cout << 0+compresssionMethod << endl;
+ uint8_t CMF = reader.readByte();
+ uint8_t CM = CMF & 0b00001111;
+ uint8_t CINFO = (CMF & 0b11110000) >> 4;
+ chunkSize--;
+ uint8_t FLG = reader.readByte();
+ bool check = (CMF * 256 + FLG)%31 == 0;
+ bool FDICT = FLG & 0b00100000;
+ uint8_t FLEVEL = FLG & 0b11000000;
+ chunkSize--;
+ cout << std::bitset<4>(CM) << ", " << std::bitset<4>(CINFO) << ", " << (check?"Valid":"Failed checksum") << ", " << (FDICT?"Dict is present":"No dict present") << ", " << std::bitset<2>(FLEVEL) << endl;
+ char compressedData[chunkSize - 4];
+ reader.readBytes(compressedData, chunkSize - 4);
+
+ const int compressedSize = chunkSize - 4;
+
+ uint32_t checkValue = reader.readData<uint32_t>();
+
+ //end = true;
+}
+
+DEFINE_CHUNK_READER(sRGB)
+{
+ renderingIntent = reader.readData<uint8_t>();
+ cout << "Rendering intent: " << 0+renderingIntent << endl;
+}
+
+DEFINE_CHUNK_READER(eXIf)
+{
+ char endian[4];
+ reader.readBytes(endian, 4);
+ for(int i = 0; i < 2; i++)
+ {
+ cout << endian[i];
+ }
+ for(int i = 2; i < 4; i++)
+ {
+ cout << " " << 0+endian[i];
+ }
+ cout << endl;
+ char rest[chunkSize - 4];
+ reader.readBytes(rest, chunkSize - 4);
+ cout << std::hex;
+ for(int i = 0; i < chunkSize - 4; i++)
+ {
+ cout << 0+rest[i] << " ";
+ }
+ cout << std::dec << endl;
+}
+
+DEFINE_CHUNK_READER(iDOT)
+{
+ cout << "!!! Ignoring iDOT !!!" << endl;
+ reader.skipBytes(chunkSize);
+}
+
+DEFINE_CHUNK_READER(pHYs)
+{
+ pixelsPerX = reader.readData<uint32_t>();
+ pixelsPerY = reader.readData<uint32_t>();
+ unit = reader.readData<uint8_t>();
+ cout << "Pixels per unit (x): " << pixelsPerX << ", Pixels per unit (y): " << pixelsPerY << ", unit: " << 0+unit << endl;
+}
+
+DEFINE_CHUNK_READER(tIME)
+{
+ year = reader.readData<uint16_t>();
+ month = reader.readData<uint8_t>();
+ day = reader.readData<uint8_t>();
+ hour = reader.readData<uint8_t>();
+ minute = reader.readData<uint8_t>();
+ second = reader.readData<uint8_t>();
+ cout << "Image last modified: " << 0+hour << ":" << 0+minute << ":" << 0+second << " " << 0+day << "-" << 0+month << "-" << 0+year << endl;
+}
+
+DEFINE_CHUNK_READER(tEXt)
+{
+
+ std::string keyword;
+ char c = reader.readByte();
+ chunkSize--;
+ while(c != 0)
+ {
+ keyword.push_back(c);
+ c = reader.readByte();
+ chunkSize--;
+ }
+ cout << keyword << endl;
+ std::string textString;
+ c = reader.readByte();
+ chunkSize--;
+ while(chunkSize > 0)
+ {
+ textString.push_back(c);
+ c = reader.readByte();
+ chunkSize--;
+ }
+ textString.push_back(c);
+ cout << textString << endl;
+}
+
+DEFINE_CHUNK_READER(IDAT)
+{
+ if(idatDataSize == 0)
+ {
+ uint8_t CMF = reader.readByte();
+ uint8_t CM = (CMF & 0b11110000) >> 4;
+ uint8_t CINFO = CMF & 0b00001111;
+ chunkSize--;
+ uint8_t FLG = reader.readByte();
+ bool check = (CMF * 256 + FLG)%31 == 0;
+ bool FDICT = FLG & 0b00000100;
+ uint8_t FLEVEL = FLG & 0b00000011;
+ chunkSize--;
+ cout << std::bitset<4>(CM) << ", " << std::bitset<4>(CINFO) << ", " << (check?"Valid":"Failed checksum") << ", " << (FDICT?"Dict is present":"No dict present") << ", " << std::bitset<2>(FLEVEL) << endl;
+ idatData = (uint8_t*)malloc(0);
+ }
+
+ idatData = (uint8_t *)realloc(idatData, idatDataSize + chunkSize);
+ reader.readBytes((char *)&idatData[idatDataSize], chunkSize);
+ idatDataSize += chunkSize;
+
+ /*
+ unsigned long compressedSize = chunkSize - 4;
+
+ unsigned long imageDataSize = height * (width * 3 + 1);
+ cout << zlib.decodeData((uint8_t*)compressedData, compressedSize, imageData, imageDataSize) << endl;
+ //cout << (int)puff((unsigned char*)imageData, &imageDataSize, (const unsigned char*)compressedData, &compressedSize) << endl;
+ */
+
+ //uint32_t checkValue = reader.readData<uint32_t>();
+
+ //end = true;
+}
+
+uint8_t paethPredictor(uint8_t a, uint8_t b, uint8_t c)
+{
+ int p = a + b - c;
+ int pa = abs(p - a);
+ int pb = abs(p - b);
+ int pc = abs(p - c);
+ if (pa <= pb && pa <= pc)
+ return a;
+ else if (pb <= pc)
+ return b;
+ else
+ return c;
+}
+
+DEFINE_CHUNK_READER(IEND)
+{
+ unsigned long imageDataSize = height * (width * bpp + 1);
+ uint8_t* pngImageData = new uint8_t[imageDataSize];
+ cout << "My inflate " << zlib.decodeData(idatData, idatDataSize, pngImageData, imageDataSize) << endl;
+ end = true;
+ reader.close();
+
+
+ FILE* fd = fopen("tmp.bmp", "w");
+ char magic[] = "BM";
+ fwrite(magic, sizeof(char), 2, fd);
+ uint32_t fileSize = 14 + 12 + width*height*/*(bitDepth/8)*/8*3;
+ fwrite(&fileSize, sizeof(uint32_t), 1, fd);
+ char zero[] = "\0\0\0\0";
+ fwrite(zero, sizeof(char), 4, fd);
+ uint32_t offset = 26;
+ fwrite(&offset, sizeof(uint32_t), 1, fd);
+ uint32_t headerSize = 12;
+ fwrite(&headerSize, sizeof(uint32_t), 1, fd);
+ uint16_t width = this->width;
+ uint16_t height = this->height;
+ uint16_t colorPlanes = 1;
+ uint16_t bitsPerPixel = /*bitDepth*/8*3;
+ fwrite(&width, sizeof(uint16_t), 1, fd);
+ fwrite(&height, sizeof(uint16_t), 1, fd);
+ fwrite(&colorPlanes, sizeof(uint16_t), 1, fd);
+ fwrite(&bitsPerPixel, sizeof(uint16_t), 1, fd);
+
+#define imageDataIndex(x, y) imageData[y*width*bpp + x]
+#define pngImageDataIndex(x, y) pngImageData[y*(width*bpp + 1) + x + 1]
+#define filterByte(y) pngImageDataIndex(-1, y)
+
+ for(int y = 0; y < height; y++)
+ {
+ for(int x = 0; x < width*bpp; x++)
+ {
+ if(filterByte(y) == 0)
+ {
+ imageDataIndex(x, y) = pngImageDataIndex(x, y);
+ }
+ else if(filterByte(y) == 1)
+ {
+ uint8_t sub = pngImageDataIndex(x, y);
+ uint8_t raw = (x>=bpp)?imageDataIndex((x-bpp), y):0;
+ imageDataIndex(x, y) = sub + raw;
+ }
+ else if(filterByte(y) == 2)
+ {
+ uint8_t up = pngImageDataIndex(x, y);
+ uint8_t prior = (y>=1)?imageDataIndex(x, (y-1)):0;
+ imageDataIndex(x, y) = up + prior;
+ }
+ else if(filterByte(y) == 4)
+ {
+ uint8_t a = (x>=bpp)?imageDataIndex((x-bpp), y):0;
+ uint8_t b = (y>=1)?imageDataIndex(x, (y-1)):0;
+ uint8_t c = (x>=bpp && y>=1)?imageDataIndex((x-bpp), (y-1)):0;
+ uint8_t paeth = pngImageDataIndex(x, y);
+ uint8_t predictor = paethPredictor(a, b, c);
+ imageDataIndex(x, y) = paeth + predictor;
+ }
+ else
+ {
+ cout << "No method for filter type: " << (int)filterByte(y) << ", row: " << y << endl;
+ throw "uh oh";
+ }
+ }
+ }
+
+#undef imageDataIndex
+#undef pngImageDataIndex
+#undef filterByte
+
+ for(int y = height-1; y >= 0; y--)
+ {
+ for(int x = 0; x < width; x++)
+ {
+ Pixel<uint8_t> pixel = getPixel<uint8_t>(x, y);
+ fwrite(&pixel.b, bitDepth/8, 1, fd);
+ fwrite(&pixel.g, bitDepth/8, 1, fd);
+ fwrite(&pixel.r, bitDepth/8, 1, fd);
+ }
+ }
+
+ delete [] pngImageData;
+ fclose(fd);
+}
diff --git a/src/PNGImage.h b/src/PNGImage.h
new file mode 100644
index 0000000..23c32a0
--- /dev/null
+++ b/src/PNGImage.h
@@ -0,0 +1,60 @@
+#pragma once
+
+#include "reader.h"
+#include "image.h"
+#include "zlib.h"
+#include <cstddef>
+#include <cstdint>
+#include <map>
+#include <string>
+#include <vector>
+
+#define CHUNK_READER(X) void X(uint32_t chunkSize)
+#define REGISTER_CHUNK_READER(X) chunkReaders.insert({#X, &PNGImage::X})
+#define DEFINE_CHUNK_READER(X) void PNGImage::X(uint32_t chunkSize)
+
+class PNGImage : Image
+{
+private:
+ ZLibInflator zlib;
+ uint8_t* idatData;
+ unsigned long idatDataSize;
+public:
+ PNGImage(std::string filename);
+ ~PNGImage();
+
+ // sRGB
+ uint8_t renderingIntent;
+
+ // pHYs
+ uint32_t pixelsPerX;
+ uint32_t pixelsPerY;
+ uint8_t unit;
+
+ // tIME
+ uint16_t year;
+ uint8_t month;
+ uint8_t day;
+ uint8_t hour;
+ uint8_t minute;
+ uint8_t second;
+
+ bool readNextChunk();
+
+private:
+ std::map<std::string, void(PNGImage::*)(uint32_t chunkSize)> chunkReaders;
+ CHUNK_READER(IHDR);
+ CHUNK_READER(iCCP);
+ CHUNK_READER(sRGB);
+ CHUNK_READER(eXIf);
+ CHUNK_READER(iDOT);
+ CHUNK_READER(pHYs);
+ CHUNK_READER(tIME);
+ CHUNK_READER(tEXt);
+ CHUNK_READER(IDAT);
+ CHUNK_READER(IEND);
+
+ bool end = false;
+
+ Reader reader;
+};
diff --git a/src/debug.h b/src/debug.h
new file mode 100644
index 0000000..356b6cb
--- /dev/null
+++ b/src/debug.h
@@ -0,0 +1,3 @@
+#pragma once
+
+#define DEBUG(X)
diff --git a/src/image.cpp b/src/image.cpp
new file mode 100644
index 0000000..35941e5
--- /dev/null
+++ b/src/image.cpp
@@ -0,0 +1,10 @@
+#include "image.h"
+
+Image::~Image()
+{
+ if(width != 0 && height != 0)
+ {
+ delete[] imageData;
+ }
+}
+
diff --git a/src/image.h b/src/image.h
new file mode 100644
index 0000000..f4071ea
--- /dev/null
+++ b/src/image.h
@@ -0,0 +1,51 @@
+#pragma once
+
+#include <cstddef>
+#include <cstdint>
+
+template <typename T>
+struct Pixel
+{
+ T r, g, b, a;
+};
+
+
+class Image
+{
+protected:
+ uint8_t* imageData;
+ uint8_t colorValues;
+ uint8_t bpp;
+public:
+ Image() = default;
+ ~Image();
+
+ uint32_t width = 0;
+ uint32_t height = 0;
+ uint8_t bitDepth;
+ uint8_t colorType;
+ uint8_t compressionMethod;
+ uint8_t filterMethod;
+ uint8_t interlaceMethod;
+
+ template <typename T>
+ Pixel<T> getPixel(unsigned int x, unsigned int y);
+};
+
+
+template <typename T>
+Pixel<T> Image::getPixel(unsigned int x, unsigned int y)
+{
+ Pixel<T> pixel;
+
+ pixel.r = (T)imageData[y * width * colorValues * bitDepth/8 + x * colorValues * bitDepth/8];
+ pixel.g = (T)imageData[y * width * colorValues * bitDepth/8 + x * colorValues * bitDepth/8 + 1];
+ pixel.b = (T)imageData[y * width * colorValues * bitDepth/8 + x * colorValues * bitDepth/8 + 2];
+
+ if(colorValues == 4)
+ pixel.a = imageData[y * width * colorValues * bitDepth/8 + x * colorValues * bitDepth/8 + 3];
+ else
+ pixel.a = 0;
+
+ return pixel;
+}
diff --git a/src/image.hpp b/src/image.hpp
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/src/image.hpp
diff --git a/src/puff.cpp b/src/puff.cpp
new file mode 100644
index 0000000..c6c90d7
--- /dev/null
+++ b/src/puff.cpp
@@ -0,0 +1,840 @@
+/*
+ * puff.c
+ * Copyright (C) 2002-2013 Mark Adler
+ * For conditions of distribution and use, see copyright notice in puff.h
+ * version 2.3, 21 Jan 2013
+ *
+ * puff.c is a simple inflate written to be an unambiguous way to specify the
+ * deflate format. It is not written for speed but rather simplicity. As a
+ * side benefit, this code might actually be useful when small code is more
+ * important than speed, such as bootstrap applications. For typical deflate
+ * data, zlib's inflate() is about four times as fast as puff(). zlib's
+ * inflate compiles to around 20K on my machine, whereas puff.c compiles to
+ * around 4K on my machine (a PowerPC using GNU cc). If the faster decode()
+ * function here is used, then puff() is only twice as slow as zlib's
+ * inflate().
+ *
+ * All dynamically allocated memory comes from the stack. The stack required
+ * is less than 2K bytes. This code is compatible with 16-bit int's and
+ * assumes that long's are at least 32 bits. puff.c uses the short data type,
+ * assumed to be 16 bits, for arrays in order to conserve memory. The code
+ * works whether integers are stored big endian or little endian.
+ *
+ * In the comments below are "Format notes" that describe the inflate process
+ * and document some of the less obvious aspects of the format. This source
+ * code is meant to supplement RFC 1951, which formally describes the deflate
+ * format:
+ *
+ * http://www.zlib.org/rfc-deflate.html
+ */
+
+/*
+ * Change history:
+ *
+ * 1.0 10 Feb 2002 - First version
+ * 1.1 17 Feb 2002 - Clarifications of some comments and notes
+ * - Update puff() dest and source pointers on negative
+ * errors to facilitate debugging deflators
+ * - Remove longest from struct huffman -- not needed
+ * - Simplify offs[] index in construct()
+ * - Add input size and checking, using longjmp() to
+ * maintain easy readability
+ * - Use short data type for large arrays
+ * - Use pointers instead of long to specify source and
+ * destination sizes to avoid arbitrary 4 GB limits
+ * 1.2 17 Mar 2002 - Add faster version of decode(), doubles speed (!),
+ * but leave simple version for readabilty
+ * - Make sure invalid distances detected if pointers
+ * are 16 bits
+ * - Fix fixed codes table error
+ * - Provide a scanning mode for determining size of
+ * uncompressed data
+ * 1.3 20 Mar 2002 - Go back to lengths for puff() parameters [Gailly]
+ * - Add a puff.h file for the interface
+ * - Add braces in puff() for else do [Gailly]
+ * - Use indexes instead of pointers for readability
+ * 1.4 31 Mar 2002 - Simplify construct() code set check
+ * - Fix some comments
+ * - Add FIXLCODES #define
+ * 1.5 6 Apr 2002 - Minor comment fixes
+ * 1.6 7 Aug 2002 - Minor format changes
+ * 1.7 3 Mar 2003 - Added test code for distribution
+ * - Added zlib-like license
+ * 1.8 9 Jan 2004 - Added some comments on no distance codes case
+ * 1.9 21 Feb 2008 - Fix bug on 16-bit integer architectures [Pohland]
+ * - Catch missing end-of-block symbol error
+ * 2.0 25 Jul 2008 - Add #define to permit distance too far back
+ * - Add option in TEST code for puff to write the data
+ * - Add option in TEST code to skip input bytes
+ * - Allow TEST code to read from piped stdin
+ * 2.1 4 Apr 2010 - Avoid variable initialization for happier compilers
+ * - Avoid unsigned comparisons for even happier compilers
+ * 2.2 25 Apr 2010 - Fix bug in variable initializations [Oberhumer]
+ * - Add const where appropriate [Oberhumer]
+ * - Split if's and ?'s for coverage testing
+ * - Break out test code to separate file
+ * - Move NIL to puff.h
+ * - Allow incomplete code only if single code length is 1
+ * - Add full code coverage test to Makefile
+ * 2.3 21 Jan 2013 - Check for invalid code length codes in dynamic blocks
+ */
+
+#include <setjmp.h> /* for setjmp(), longjmp(), and jmp_buf */
+#include "puff.h" /* prototype for puff() */
+
+#define local static /* for local function definitions */
+
+/*
+ * Maximums for allocations and loops. It is not useful to change these --
+ * they are fixed by the deflate format.
+ */
+#define MAXBITS 15 /* maximum bits in a code */
+#define MAXLCODES 286 /* maximum number of literal/length codes */
+#define MAXDCODES 30 /* maximum number of distance codes */
+#define MAXCODES (MAXLCODES+MAXDCODES) /* maximum codes lengths to read */
+#define FIXLCODES 288 /* number of fixed literal/length codes */
+
+/* input and output state */
+struct state {
+ /* output state */
+ unsigned char *out; /* output buffer */
+ unsigned long outlen; /* available space at out */
+ unsigned long outcnt; /* bytes written to out so far */
+
+ /* input state */
+ const unsigned char *in; /* input buffer */
+ unsigned long inlen; /* available input at in */
+ unsigned long incnt; /* bytes read so far */
+ int bitbuf; /* bit buffer */
+ int bitcnt; /* number of bits in bit buffer */
+
+ /* input limit error return state for bits() and decode() */
+ jmp_buf env;
+};
+
+/*
+ * Return need bits from the input stream. This always leaves less than
+ * eight bits in the buffer. bits() works properly for need == 0.
+ *
+ * Format notes:
+ *
+ * - Bits are stored in bytes from the least significant bit to the most
+ * significant bit. Therefore bits are dropped from the bottom of the bit
+ * buffer, using shift right, and new bytes are appended to the top of the
+ * bit buffer, using shift left.
+ */
+local int bits(struct state *s, int need)
+{
+ long val; /* bit accumulator (can use up to 20 bits) */
+
+ /* load at least need bits into val */
+ val = s->bitbuf;
+ while (s->bitcnt < need) {
+ if (s->incnt == s->inlen)
+ longjmp(s->env, 1); /* out of input */
+ val |= (long)(s->in[s->incnt++]) << s->bitcnt; /* load eight bits */
+ s->bitcnt += 8;
+ }
+
+ /* drop need bits and update buffer, always zero to seven bits left */
+ s->bitbuf = (int)(val >> need);
+ s->bitcnt -= need;
+
+ /* return need bits, zeroing the bits above that */
+ return (int)(val & ((1L << need) - 1));
+}
+
+/*
+ * Process a stored block.
+ *
+ * Format notes:
+ *
+ * - After the two-bit stored block type (00), the stored block length and
+ * stored bytes are byte-aligned for fast copying. Therefore any leftover
+ * bits in the byte that has the last bit of the type, as many as seven, are
+ * discarded. The value of the discarded bits are not defined and should not
+ * be checked against any expectation.
+ *
+ * - The second inverted copy of the stored block length does not have to be
+ * checked, but it's probably a good idea to do so anyway.
+ *
+ * - A stored block can have zero length. This is sometimes used to byte-align
+ * subsets of the compressed data for random access or partial recovery.
+ */
+local int stored(struct state *s)
+{
+ unsigned len; /* length of stored block */
+
+ /* discard leftover bits from current byte (assumes s->bitcnt < 8) */
+ s->bitbuf = 0;
+ s->bitcnt = 0;
+
+ /* get length and check against its one's complement */
+ if (s->incnt + 4 > s->inlen)
+ return 2; /* not enough input */
+ len = s->in[s->incnt++];
+ len |= s->in[s->incnt++] << 8;
+ if (s->in[s->incnt++] != (~len & 0xff) ||
+ s->in[s->incnt++] != ((~len >> 8) & 0xff))
+ return -2; /* didn't match complement! */
+
+ /* copy len bytes from in to out */
+ if (s->incnt + len > s->inlen)
+ return 2; /* not enough input */
+ if (s->out != NIL) {
+ if (s->outcnt + len > s->outlen)
+ return 1; /* not enough output space */
+ while (len--)
+ s->out[s->outcnt++] = s->in[s->incnt++];
+ }
+ else { /* just scanning */
+ s->outcnt += len;
+ s->incnt += len;
+ }
+
+ /* done with a valid stored block */
+ return 0;
+}
+
+/*
+ * Huffman code decoding tables. count[1..MAXBITS] is the number of symbols of
+ * each length, which for a canonical code are stepped through in order.
+ * symbol[] are the symbol values in canonical order, where the number of
+ * entries is the sum of the counts in count[]. The decoding process can be
+ * seen in the function decode() below.
+ */
+struct huffman {
+ short *count; /* number of symbols of each length */
+ short *symbol; /* canonically ordered symbols */
+};
+
+/*
+ * Decode a code from the stream s using huffman table h. Return the symbol or
+ * a negative value if there is an error. If all of the lengths are zero, i.e.
+ * an empty code, or if the code is incomplete and an invalid code is received,
+ * then -10 is returned after reading MAXBITS bits.
+ *
+ * Format notes:
+ *
+ * - The codes as stored in the compressed data are bit-reversed relative to
+ * a simple integer ordering of codes of the same lengths. Hence below the
+ * bits are pulled from the compressed data one at a time and used to
+ * build the code value reversed from what is in the stream in order to
+ * permit simple integer comparisons for decoding. A table-based decoding
+ * scheme (as used in zlib) does not need to do this reversal.
+ *
+ * - The first code for the shortest length is all zeros. Subsequent codes of
+ * the same length are simply integer increments of the previous code. When
+ * moving up a length, a zero bit is appended to the code. For a complete
+ * code, the last code of the longest length will be all ones.
+ *
+ * - Incomplete codes are handled by this decoder, since they are permitted
+ * in the deflate format. See the format notes for fixed() and dynamic().
+ */
+#ifdef SLOW
+local int decode(struct state *s, const struct huffman *h)
+{
+ int len; /* current number of bits in code */
+ int code; /* len bits being decoded */
+ int first; /* first code of length len */
+ int count; /* number of codes of length len */
+ int index; /* index of first code of length len in symbol table */
+
+ code = first = index = 0;
+ for (len = 1; len <= MAXBITS; len++) {
+ code |= bits(s, 1); /* get next bit */
+ count = h->count[len];
+ if (code - count < first) /* if length len, return symbol */
+ return h->symbol[index + (code - first)];
+ index += count; /* else update for next length */
+ first += count;
+ first <<= 1;
+ code <<= 1;
+ }
+ return -10; /* ran out of codes */
+}
+
+/*
+ * A faster version of decode() for real applications of this code. It's not
+ * as readable, but it makes puff() twice as fast. And it only makes the code
+ * a few percent larger.
+ */
+#else /* !SLOW */
+local int decode(struct state *s, const struct huffman *h)
+{
+ int len; /* current number of bits in code */
+ int code; /* len bits being decoded */
+ int first; /* first code of length len */
+ int count; /* number of codes of length len */
+ int index; /* index of first code of length len in symbol table */
+ int bitbuf; /* bits from stream */
+ int left; /* bits left in next or left to process */
+ short *next; /* next number of codes */
+
+ bitbuf = s->bitbuf;
+ left = s->bitcnt;
+ code = first = index = 0;
+ len = 1;
+ next = h->count + 1;
+ while (1) {
+ while (left--) {
+ code |= bitbuf & 1;
+ bitbuf >>= 1;
+ count = *next++;
+ if (code - count < first) { /* if length len, return symbol */
+ s->bitbuf = bitbuf;
+ s->bitcnt = (s->bitcnt - len) & 7;
+ return h->symbol[index + (code - first)];
+ }
+ index += count; /* else update for next length */
+ first += count;
+ first <<= 1;
+ code <<= 1;
+ len++;
+ }
+ left = (MAXBITS+1) - len;
+ if (left == 0)
+ break;
+ if (s->incnt == s->inlen)
+ longjmp(s->env, 1); /* out of input */
+ bitbuf = s->in[s->incnt++];
+ if (left > 8)
+ left = 8;
+ }
+ return -10; /* ran out of codes */
+}
+#endif /* SLOW */
+
+/*
+ * Given the list of code lengths length[0..n-1] representing a canonical
+ * Huffman code for n symbols, construct the tables required to decode those
+ * codes. Those tables are the number of codes of each length, and the symbols
+ * sorted by length, retaining their original order within each length. The
+ * return value is zero for a complete code set, negative for an over-
+ * subscribed code set, and positive for an incomplete code set. The tables
+ * can be used if the return value is zero or positive, but they cannot be used
+ * if the return value is negative. If the return value is zero, it is not
+ * possible for decode() using that table to return an error--any stream of
+ * enough bits will resolve to a symbol. If the return value is positive, then
+ * it is possible for decode() using that table to return an error for received
+ * codes past the end of the incomplete lengths.
+ *
+ * Not used by decode(), but used for error checking, h->count[0] is the number
+ * of the n symbols not in the code. So n - h->count[0] is the number of
+ * codes. This is useful for checking for incomplete codes that have more than
+ * one symbol, which is an error in a dynamic block.
+ *
+ * Assumption: for all i in 0..n-1, 0 <= length[i] <= MAXBITS
+ * This is assured by the construction of the length arrays in dynamic() and
+ * fixed() and is not verified by construct().
+ *
+ * Format notes:
+ *
+ * - Permitted and expected examples of incomplete codes are one of the fixed
+ * codes and any code with a single symbol which in deflate is coded as one
+ * bit instead of zero bits. See the format notes for fixed() and dynamic().
+ *
+ * - Within a given code length, the symbols are kept in ascending order for
+ * the code bits definition.
+ */
+local int construct(struct huffman *h, const short *length, int n)
+{
+ int symbol; /* current symbol when stepping through length[] */
+ int len; /* current length when stepping through h->count[] */
+ int left; /* number of possible codes left of current length */
+ short offs[MAXBITS+1]; /* offsets in symbol table for each length */
+
+ /* count number of codes of each length */
+ for (len = 0; len <= MAXBITS; len++)
+ h->count[len] = 0;
+ for (symbol = 0; symbol < n; symbol++)
+ (h->count[length[symbol]])++; /* assumes lengths are within bounds */
+ if (h->count[0] == n) /* no codes! */
+ return 0; /* complete, but decode() will fail */
+
+ /* check for an over-subscribed or incomplete set of lengths */
+ left = 1; /* one possible code of zero length */
+ for (len = 1; len <= MAXBITS; len++) {
+ left <<= 1; /* one more bit, double codes left */
+ left -= h->count[len]; /* deduct count from possible codes */
+ if (left < 0)
+ return left; /* over-subscribed--return negative */
+ } /* left > 0 means incomplete */
+
+ /* generate offsets into symbol table for each length for sorting */
+ offs[1] = 0;
+ for (len = 1; len < MAXBITS; len++)
+ offs[len + 1] = offs[len] + h->count[len];
+
+ /*
+ * put symbols in table sorted by length, by symbol order within each
+ * length
+ */
+ for (symbol = 0; symbol < n; symbol++)
+ if (length[symbol] != 0)
+ h->symbol[offs[length[symbol]]++] = symbol;
+
+ /* return zero for complete set, positive for incomplete set */
+ return left;
+}
+
+/*
+ * Decode literal/length and distance codes until an end-of-block code.
+ *
+ * Format notes:
+ *
+ * - Compressed data that is after the block type if fixed or after the code
+ * description if dynamic is a combination of literals and length/distance
+ * pairs terminated by and end-of-block code. Literals are simply Huffman
+ * coded bytes. A length/distance pair is a coded length followed by a
+ * coded distance to represent a string that occurs earlier in the
+ * uncompressed data that occurs again at the current location.
+ *
+ * - Literals, lengths, and the end-of-block code are combined into a single
+ * code of up to 286 symbols. They are 256 literals (0..255), 29 length
+ * symbols (257..285), and the end-of-block symbol (256).
+ *
+ * - There are 256 possible lengths (3..258), and so 29 symbols are not enough
+ * to represent all of those. Lengths 3..10 and 258 are in fact represented
+ * by just a length symbol. Lengths 11..257 are represented as a symbol and
+ * some number of extra bits that are added as an integer to the base length
+ * of the length symbol. The number of extra bits is determined by the base
+ * length symbol. These are in the static arrays below, lens[] for the base
+ * lengths and lext[] for the corresponding number of extra bits.
+ *
+ * - The reason that 258 gets its own symbol is that the longest length is used
+ * often in highly redundant files. Note that 258 can also be coded as the
+ * base value 227 plus the maximum extra value of 31. While a good deflate
+ * should never do this, it is not an error, and should be decoded properly.
+ *
+ * - If a length is decoded, including its extra bits if any, then it is
+ * followed a distance code. There are up to 30 distance symbols. Again
+ * there are many more possible distances (1..32768), so extra bits are added
+ * to a base value represented by the symbol. The distances 1..4 get their
+ * own symbol, but the rest require extra bits. The base distances and
+ * corresponding number of extra bits are below in the static arrays dist[]
+ * and dext[].
+ *
+ * - Literal bytes are simply written to the output. A length/distance pair is
+ * an instruction to copy previously uncompressed bytes to the output. The
+ * copy is from distance bytes back in the output stream, copying for length
+ * bytes.
+ *
+ * - Distances pointing before the beginning of the output data are not
+ * permitted.
+ *
+ * - Overlapped copies, where the length is greater than the distance, are
+ * allowed and common. For example, a distance of one and a length of 258
+ * simply copies the last byte 258 times. A distance of four and a length of
+ * twelve copies the last four bytes three times. A simple forward copy
+ * ignoring whether the length is greater than the distance or not implements
+ * this correctly. You should not use memcpy() since its behavior is not
+ * defined for overlapped arrays. You should not use memmove() or bcopy()
+ * since though their behavior -is- defined for overlapping arrays, it is
+ * defined to do the wrong thing in this case.
+ */
+local int codes(struct state *s,
+ const struct huffman *lencode,
+ const struct huffman *distcode)
+{
+ int symbol; /* decoded symbol */
+ int len; /* length for copy */
+ unsigned dist; /* distance for copy */
+ static const short lens[29] = { /* Size base for length codes 257..285 */
+ 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
+ 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258};
+ static const short lext[29] = { /* Extra bits for length codes 257..285 */
+ 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
+ 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0};
+ static const short dists[30] = { /* Offset base for distance codes 0..29 */
+ 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
+ 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
+ 8193, 12289, 16385, 24577};
+ static const short dext[30] = { /* Extra bits for distance codes 0..29 */
+ 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
+ 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
+ 12, 12, 13, 13};
+
+ /* decode literals and length/distance pairs */
+ do {
+ symbol = decode(s, lencode);
+ if (symbol < 0)
+ return symbol; /* invalid symbol */
+ if (symbol < 256) { /* literal: symbol is the byte */
+ /* write out the literal */
+ if (s->out != NIL) {
+ if (s->outcnt == s->outlen)
+ return 1;
+ s->out[s->outcnt] = symbol;
+ }
+ s->outcnt++;
+ }
+ else if (symbol > 256) { /* length */
+ /* get and compute length */
+ symbol -= 257;
+ if (symbol >= 29)
+ return -10; /* invalid fixed code */
+ len = lens[symbol] + bits(s, lext[symbol]);
+
+ /* get and check distance */
+ symbol = decode(s, distcode);
+ if (symbol < 0)
+ return symbol; /* invalid symbol */
+ dist = dists[symbol] + bits(s, dext[symbol]);
+#ifndef INFLATE_ALLOW_INVALID_DISTANCE_TOOFAR_ARRR
+ if (dist > s->outcnt)
+ return -11; /* distance too far back */
+#endif
+
+ /* copy length bytes from distance bytes back */
+ if (s->out != NIL) {
+ if (s->outcnt + len > s->outlen)
+ return 1;
+ while (len--) {
+ s->out[s->outcnt] =
+#ifdef INFLATE_ALLOW_INVALID_DISTANCE_TOOFAR_ARRR
+ dist > s->outcnt ?
+ 0 :
+#endif
+ s->out[s->outcnt - dist];
+ s->outcnt++;
+ }
+ }
+ else
+ s->outcnt += len;
+ }
+ } while (symbol != 256); /* end of block symbol */
+
+ /* done with a valid fixed or dynamic block */
+ return 0;
+}
+
+/*
+ * Process a fixed codes block.
+ *
+ * Format notes:
+ *
+ * - This block type can be useful for compressing small amounts of data for
+ * which the size of the code descriptions in a dynamic block exceeds the
+ * benefit of custom codes for that block. For fixed codes, no bits are
+ * spent on code descriptions. Instead the code lengths for literal/length
+ * codes and distance codes are fixed. The specific lengths for each symbol
+ * can be seen in the "for" loops below.
+ *
+ * - The literal/length code is complete, but has two symbols that are invalid
+ * and should result in an error if received. This cannot be implemented
+ * simply as an incomplete code since those two symbols are in the "middle"
+ * of the code. They are eight bits long and the longest literal/length\
+ * code is nine bits. Therefore the code must be constructed with those
+ * symbols, and the invalid symbols must be detected after decoding.
+ *
+ * - The fixed distance codes also have two invalid symbols that should result
+ * in an error if received. Since all of the distance codes are the same
+ * length, this can be implemented as an incomplete code. Then the invalid
+ * codes are detected while decoding.
+ */
+local int fixed(struct state *s)
+{
+ static int virgin = 1;
+ static short lencnt[MAXBITS+1], lensym[FIXLCODES];
+ static short distcnt[MAXBITS+1], distsym[MAXDCODES];
+ static struct huffman lencode, distcode;
+
+ /* build fixed huffman tables if first call (may not be thread safe) */
+ if (virgin) {
+ int symbol;
+ short lengths[FIXLCODES];
+
+ /* construct lencode and distcode */
+ lencode.count = lencnt;
+ lencode.symbol = lensym;
+ distcode.count = distcnt;
+ distcode.symbol = distsym;
+
+ /* literal/length table */
+ for (symbol = 0; symbol < 144; symbol++)
+ lengths[symbol] = 8;
+ for (; symbol < 256; symbol++)
+ lengths[symbol] = 9;
+ for (; symbol < 280; symbol++)
+ lengths[symbol] = 7;
+ for (; symbol < FIXLCODES; symbol++)
+ lengths[symbol] = 8;
+ construct(&lencode, lengths, FIXLCODES);
+
+ /* distance table */
+ for (symbol = 0; symbol < MAXDCODES; symbol++)
+ lengths[symbol] = 5;
+ construct(&distcode, lengths, MAXDCODES);
+
+ /* do this just once */
+ virgin = 0;
+ }
+
+ /* decode data until end-of-block code */
+ return codes(s, &lencode, &distcode);
+}
+
+/*
+ * Process a dynamic codes block.
+ *
+ * Format notes:
+ *
+ * - A dynamic block starts with a description of the literal/length and
+ * distance codes for that block. New dynamic blocks allow the compressor to
+ * rapidly adapt to changing data with new codes optimized for that data.
+ *
+ * - The codes used by the deflate format are "canonical", which means that
+ * the actual bits of the codes are generated in an unambiguous way simply
+ * from the number of bits in each code. Therefore the code descriptions
+ * are simply a list of code lengths for each symbol.
+ *
+ * - The code lengths are stored in order for the symbols, so lengths are
+ * provided for each of the literal/length symbols, and for each of the
+ * distance symbols.
+ *
+ * - If a symbol is not used in the block, this is represented by a zero as
+ * as the code length. This does not mean a zero-length code, but rather
+ * that no code should be created for this symbol. There is no way in the
+ * deflate format to represent a zero-length code.
+ *
+ * - The maximum number of bits in a code is 15, so the possible lengths for
+ * any code are 1..15.
+ *
+ * - The fact that a length of zero is not permitted for a code has an
+ * interesting consequence. Normally if only one symbol is used for a given
+ * code, then in fact that code could be represented with zero bits. However
+ * in deflate, that code has to be at least one bit. So for example, if
+ * only a single distance base symbol appears in a block, then it will be
+ * represented by a single code of length one, in particular one 0 bit. This
+ * is an incomplete code, since if a 1 bit is received, it has no meaning,
+ * and should result in an error. So incomplete distance codes of one symbol
+ * should be permitted, and the receipt of invalid codes should be handled.
+ *
+ * - It is also possible to have a single literal/length code, but that code
+ * must be the end-of-block code, since every dynamic block has one. This
+ * is not the most efficient way to create an empty block (an empty fixed
+ * block is fewer bits), but it is allowed by the format. So incomplete
+ * literal/length codes of one symbol should also be permitted.
+ *
+ * - If there are only literal codes and no lengths, then there are no distance
+ * codes. This is represented by one distance code with zero bits.
+ *
+ * - The list of up to 286 length/literal lengths and up to 30 distance lengths
+ * are themselves compressed using Huffman codes and run-length encoding. In
+ * the list of code lengths, a 0 symbol means no code, a 1..15 symbol means
+ * that length, and the symbols 16, 17, and 18 are run-length instructions.
+ * Each of 16, 17, and 18 are follwed by extra bits to define the length of
+ * the run. 16 copies the last length 3 to 6 times. 17 represents 3 to 10
+ * zero lengths, and 18 represents 11 to 138 zero lengths. Unused symbols
+ * are common, hence the special coding for zero lengths.
+ *
+ * - The symbols for 0..18 are Huffman coded, and so that code must be
+ * described first. This is simply a sequence of up to 19 three-bit values
+ * representing no code (0) or the code length for that symbol (1..7).
+ *
+ * - A dynamic block starts with three fixed-size counts from which is computed
+ * the number of literal/length code lengths, the number of distance code
+ * lengths, and the number of code length code lengths (ok, you come up with
+ * a better name!) in the code descriptions. For the literal/length and
+ * distance codes, lengths after those provided are considered zero, i.e. no
+ * code. The code length code lengths are received in a permuted order (see
+ * the order[] array below) to make a short code length code length list more
+ * likely. As it turns out, very short and very long codes are less likely
+ * to be seen in a dynamic code description, hence what may appear initially
+ * to be a peculiar ordering.
+ *
+ * - Given the number of literal/length code lengths (nlen) and distance code
+ * lengths (ndist), then they are treated as one long list of nlen + ndist
+ * code lengths. Therefore run-length coding can and often does cross the
+ * boundary between the two sets of lengths.
+ *
+ * - So to summarize, the code description at the start of a dynamic block is
+ * three counts for the number of code lengths for the literal/length codes,
+ * the distance codes, and the code length codes. This is followed by the
+ * code length code lengths, three bits each. This is used to construct the
+ * code length code which is used to read the remainder of the lengths. Then
+ * the literal/length code lengths and distance lengths are read as a single
+ * set of lengths using the code length codes. Codes are constructed from
+ * the resulting two sets of lengths, and then finally you can start
+ * decoding actual compressed data in the block.
+ *
+ * - For reference, a "typical" size for the code description in a dynamic
+ * block is around 80 bytes.
+ */
+local int dynamic(struct state *s)
+{
+ int nlen, ndist, ncode; /* number of lengths in descriptor */
+ int index; /* index of lengths[] */
+ int err; /* construct() return value */
+ short lengths[MAXCODES]; /* descriptor code lengths */
+ short lencnt[MAXBITS+1], lensym[MAXLCODES]; /* lencode memory */
+ short distcnt[MAXBITS+1], distsym[MAXDCODES]; /* distcode memory */
+ struct huffman lencode, distcode; /* length and distance codes */
+ static const short order[19] = /* permutation of code length codes */
+ {16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
+
+ /* construct lencode and distcode */
+ lencode.count = lencnt;
+ lencode.symbol = lensym;
+ distcode.count = distcnt;
+ distcode.symbol = distsym;
+
+ /* get number of lengths in each table, check lengths */
+ nlen = bits(s, 5) + 257;
+ ndist = bits(s, 5) + 1;
+ ncode = bits(s, 4) + 4;
+ if (nlen > MAXLCODES || ndist > MAXDCODES)
+ return -3; /* bad counts */
+
+ /* read code length code lengths (really), missing lengths are zero */
+ for (index = 0; index < ncode; index++)
+ lengths[order[index]] = bits(s, 3);
+ for (; index < 19; index++)
+ lengths[order[index]] = 0;
+
+ /* build huffman table for code lengths codes (use lencode temporarily) */
+ err = construct(&lencode, lengths, 19);
+ if (err != 0) /* require complete code set here */
+ return -4;
+
+ /* read length/literal and distance code length tables */
+ index = 0;
+ while (index < nlen + ndist) {
+ int symbol; /* decoded value */
+ int len; /* last length to repeat */
+
+ symbol = decode(s, &lencode);
+ if (symbol < 0)
+ return symbol; /* invalid symbol */
+ if (symbol < 16) /* length in 0..15 */
+ lengths[index++] = symbol;
+ else { /* repeat instruction */
+ len = 0; /* assume repeating zeros */
+ if (symbol == 16) { /* repeat last length 3..6 times */
+ if (index == 0)
+ return -5; /* no last length! */
+ len = lengths[index - 1]; /* last length */
+ symbol = 3 + bits(s, 2);
+ }
+ else if (symbol == 17) /* repeat zero 3..10 times */
+ symbol = 3 + bits(s, 3);
+ else /* == 18, repeat zero 11..138 times */
+ symbol = 11 + bits(s, 7);
+ if (index + symbol > nlen + ndist)
+ return -6; /* too many lengths! */
+ while (symbol--) /* repeat last or zero symbol times */
+ lengths[index++] = len;
+ }
+ }
+
+ /* check for end-of-block code -- there better be one! */
+ if (lengths[256] == 0)
+ return -9;
+
+ /* build huffman table for literal/length codes */
+ err = construct(&lencode, lengths, nlen);
+ if (err && (err < 0 || nlen != lencode.count[0] + lencode.count[1]))
+ return -7; /* incomplete code ok only for single length 1 code */
+
+ /* build huffman table for distance codes */
+ err = construct(&distcode, lengths + nlen, ndist);
+ if (err && (err < 0 || ndist != distcode.count[0] + distcode.count[1]))
+ return -8; /* incomplete code ok only for single length 1 code */
+
+ /* decode data until end-of-block code */
+ return codes(s, &lencode, &distcode);
+}
+
+/*
+ * Inflate source to dest. On return, destlen and sourcelen are updated to the
+ * size of the uncompressed data and the size of the deflate data respectively.
+ * On success, the return value of puff() is zero. If there is an error in the
+ * source data, i.e. it is not in the deflate format, then a negative value is
+ * returned. If there is not enough input available or there is not enough
+ * output space, then a positive error is returned. In that case, destlen and
+ * sourcelen are not updated to facilitate retrying from the beginning with the
+ * provision of more input data or more output space. In the case of invalid
+ * inflate data (a negative error), the dest and source pointers are updated to
+ * facilitate the debugging of deflators.
+ *
+ * puff() also has a mode to determine the size of the uncompressed output with
+ * no output written. For this dest must be (unsigned char *)0. In this case,
+ * the input value of *destlen is ignored, and on return *destlen is set to the
+ * size of the uncompressed output.
+ *
+ * The return codes are:
+ *
+ * 2: available inflate data did not terminate
+ * 1: output space exhausted before completing inflate
+ * 0: successful inflate
+ * -1: invalid block type (type == 3)
+ * -2: stored block length did not match one's complement
+ * -3: dynamic block code description: too many length or distance codes
+ * -4: dynamic block code description: code lengths codes incomplete
+ * -5: dynamic block code description: repeat lengths with no first length
+ * -6: dynamic block code description: repeat more than specified lengths
+ * -7: dynamic block code description: invalid literal/length code lengths
+ * -8: dynamic block code description: invalid distance code lengths
+ * -9: dynamic block code description: missing end-of-block code
+ * -10: invalid literal/length or distance code in fixed or dynamic block
+ * -11: distance is too far back in fixed or dynamic block
+ *
+ * Format notes:
+ *
+ * - Three bits are read for each block to determine the kind of block and
+ * whether or not it is the last block. Then the block is decoded and the
+ * process repeated if it was not the last block.
+ *
+ * - The leftover bits in the last byte of the deflate data after the last
+ * block (if it was a fixed or dynamic block) are undefined and have no
+ * expected values to check.
+ */
+int puff(unsigned char *dest, /* pointer to destination pointer */
+ unsigned long *destlen, /* amount of output space */
+ const unsigned char *source, /* pointer to source data pointer */
+ unsigned long *sourcelen) /* amount of input available */
+{
+ struct state s; /* input/output state */
+ int last, type; /* block information */
+ int err; /* return value */
+
+ /* initialize output state */
+ s.out = dest;
+ s.outlen = *destlen; /* ignored if dest is NIL */
+ s.outcnt = 0;
+
+ /* initialize input state */
+ s.in = source;
+ s.inlen = *sourcelen;
+ s.incnt = 0;
+ s.bitbuf = 0;
+ s.bitcnt = 0;
+
+ /* return if bits() or decode() tries to read past available input */
+ if (setjmp(s.env) != 0) /* if came back here via longjmp() */
+ err = 2; /* then skip do-loop, return error */
+ else {
+ /* process blocks until last block or error */
+ do {
+ last = bits(&s, 1); /* one if last block */
+ type = bits(&s, 2); /* block type 0..3 */
+ err = type == 0 ?
+ stored(&s) :
+ (type == 1 ?
+ fixed(&s) :
+ (type == 2 ?
+ dynamic(&s) :
+ -1)); /* type == 3, invalid */
+ if (err != 0)
+ break; /* return with error */
+ } while (!last);
+ }
+
+ /* update the lengths and return */
+ if (err <= 0) {
+ *destlen = s.outcnt;
+ *sourcelen = s.incnt;
+ }
+ return err;
+}
diff --git a/src/puff.h b/src/puff.h
new file mode 100644
index 0000000..e23a245
--- /dev/null
+++ b/src/puff.h
@@ -0,0 +1,35 @@
+/* puff.h
+ Copyright (C) 2002-2013 Mark Adler, all rights reserved
+ version 2.3, 21 Jan 2013
+
+ This software is provided 'as-is', without any express or implied
+ warranty. In no event will the author be held liable for any damages
+ arising from the use of this software.
+
+ Permission is granted to anyone to use this software for any purpose,
+ including commercial applications, and to alter it and redistribute it
+ freely, subject to the following restrictions:
+
+ 1. The origin of this software must not be misrepresented; you must not
+ claim that you wrote the original software. If you use this software
+ in a product, an acknowledgment in the product documentation would be
+ appreciated but is not required.
+ 2. Altered source versions must be plainly marked as such, and must not be
+ misrepresented as being the original software.
+ 3. This notice may not be removed or altered from any source distribution.
+
+ Mark Adler madler@alumni.caltech.edu
+ */
+
+
+/*
+ * See puff.c for purpose and usage.
+ */
+#ifndef NIL
+# define NIL ((unsigned char *)0) /* for no output option */
+#endif
+
+int puff(unsigned char *dest, /* pointer to destination pointer */
+ unsigned long *destlen, /* amount of output space */
+ const unsigned char *source, /* pointer to source data pointer */
+ unsigned long *sourcelen); /* amount of input available */
diff --git a/src/reader.cpp b/src/reader.cpp
new file mode 100644
index 0000000..59b30c0
--- /dev/null
+++ b/src/reader.cpp
@@ -0,0 +1,110 @@
+#include "reader.h"
+
+#include "debug.h"
+
+#include <algorithm>
+#include <cstddef>
+#include <cstdint>
+#include <cstring>
+#include <iostream>
+
+Reader::Reader(std::string filename)
+{
+ file = fopen(filename.c_str(), "rb");
+ refreshBuffer();
+ ready = true;
+}
+Reader::~Reader()
+{
+ if(ready)
+ fclose(file);
+}
+
+char Reader::readByte()
+{
+ if(pos == BUFFER_SIZE)
+ refreshBuffer();
+ return buffer[pos++];
+}
+
+void Reader::refreshBuffer()
+{
+ fread(buffer, sizeof(buffer), 1, file);
+ pos = 0;
+}
+
+template<> uint8_t Reader::readData<uint8_t>()
+{
+ return readByte();
+}
+
+template<> uint16_t Reader::readData<uint16_t>()
+{
+ uint16_t num = 0;
+ for(int i = 0; i < 2; i++)
+ {
+ num += readByte() << (8 * (1-i));
+ }
+ return num;
+}
+
+template<> uint32_t Reader::readData<uint32_t>()
+{
+ uint32_t num = 0;
+ for(int i = 0; i < 4; i++)
+ {
+ uint8_t byte = readByte();
+ DEBUG(std::cout << std::hex << 0+byte << " ");
+ num += byte << (8 * (3-i));
+ }
+ DEBUG(std::cout << std::dec << std::endl);
+ return num;
+}
+
+template<> uint64_t Reader::readData<uint64_t>()
+{
+ uint64_t num = 0;
+ for(int i = 0; i < 8; i++)
+ {
+ num += readByte() << (8 * (7-i));
+ }
+ return num;
+}
+
+void Reader::readBytes(char* out, size_t len)
+{
+ while(len > 0)
+ {
+ size_t bytesToRead = std::min(len, BUFFER_SIZE - pos);
+ if(bytesToRead == 0)
+ {
+ refreshBuffer();
+ continue;
+ }
+ memcpy(out, buffer + pos, bytesToRead);
+ out += bytesToRead;
+ len -= bytesToRead;
+ pos += bytesToRead;
+ }
+}
+
+void Reader::skipBytes(size_t len)
+{
+ while(len > 0)
+ {
+ size_t bytesToRead = std::min(len, BUFFER_SIZE - pos);
+ if(bytesToRead == 0)
+ {
+ refreshBuffer();
+ continue;
+ }
+ len -= bytesToRead;
+ pos += bytesToRead;
+ }
+}
+
+void Reader::close()
+{
+ fclose(file);
+ ready = false;
+}
diff --git a/src/reader.h b/src/reader.h
new file mode 100644
index 0000000..ab01092
--- /dev/null
+++ b/src/reader.h
@@ -0,0 +1,33 @@
+#pragma once
+
+#include <cstddef>
+#include <cstdio>
+#include <string>
+
+#define BUFFER_SIZE 1024
+
+class Reader
+{
+public:
+ //Bytes are big endian
+ Reader(std::string file);
+ ~Reader();
+
+ template <typename T>
+ T readData();
+
+ char readByte();
+
+ void readBytes(char* out, size_t len);
+
+ void skipBytes(size_t len);
+
+ void close();
+private:
+ char buffer[BUFFER_SIZE];
+ size_t pos;
+ FILE* file;
+ bool ready = false;
+
+ void refreshBuffer();
+};
diff --git a/src/zlib.cpp b/src/zlib.cpp
new file mode 100644
index 0000000..c53502c
--- /dev/null
+++ b/src/zlib.cpp
@@ -0,0 +1,356 @@
+#include "zlib.h"
+
+#include <bitset>
+#include <cstdint>
+#include <cstring>
+#include <iomanip>
+#include <iostream>
+#include <stdexcept>
+
+#define MAXBITS 15
+
+using std::cout, std::endl;
+
+void ZLibInflator::calculateCodes(uint8_t* lengths, uint16_t* codesOut, int codeCount)
+{
+ const int biggestLen = 15;
+
+ uint16_t lenCounts[biggestLen + 1];
+ memset(lenCounts, 0, (biggestLen + 1)*sizeof(uint16_t));
+ for(int i = 0; i < codeCount; i++)
+ lenCounts[lengths[i]]++;
+
+
+ lenCounts[0] = 0;
+
+ uint16_t nextCodes[biggestLen + 1];
+ uint16_t code = 0;
+ for(int bits = 1; bits < biggestLen + 1; bits++)
+ {
+ code = (code + lenCounts[bits - 1]) << 1;
+ nextCodes[bits] = code;
+ }
+
+ for(int i = 0; i < codeCount; i++)
+ {
+ uint8_t len = lengths[i];
+ if(len == 0)
+ continue;
+ codesOut[i] = nextCodes[len]++;
+ }
+}
+
+void ZLibInflator::buildHuffmanTree(uint8_t* lengths, uint16_t* codes, int codeCount)
+{
+ buildHuffmanTree(lengths, codes, codeCount, &tree);
+}
+
+
+
+void ZLibInflator::buildHuffmanTree(uint8_t* lengths, uint16_t* codes, int codeCount, HuffmanTree* tree)
+{
+ for(uint16_t i = 0; i < codeCount; i++)
+ {
+ uint16_t code = codes[i];
+ uint8_t len = lengths[i];
+ if(len == 0)
+ continue;
+ HuffmanTree* current = tree;
+ for(int j = 0; j < len; j++)
+ {
+ bool right = code & (0b1 << (len - 1 - j));
+ if(right)
+ {
+ if(current->right != nullptr)
+ current = current->right;
+ else
+ {
+ current->right = new HuffmanTree();
+ current = current->right;
+ }
+ }
+ else
+ {
+ if(current->left != nullptr)
+ current = current->left;
+ else
+ {
+ current->left = new HuffmanTree();
+ current = current->left;
+ }
+ }
+ }
+ current->val = i;
+ }
+}
+
+void ZLibInflator::buildStaticHuffmanTree()
+{
+ if(staticTree)
+ return;
+ cout << "Building static tree" << endl;
+ buildStaticHuffmanTree(&tree, &distTree);
+ staticTree = true;
+ haveTree = true;
+}
+void ZLibInflator::buildStaticHuffmanTree(HuffmanTree* treeOut, HuffmanTree* distTreeOut)
+{
+ const int codeCount = 288;
+ uint8_t lens[codeCount];
+ uint16_t codes[codeCount];
+ for(int i = 0; i < codeCount; i++)
+ {
+ if(i < 144)
+ lens[i] = 8;
+ else if(i < 256)
+ lens[i] = 9;
+ else if(i < 280)
+ lens[i] = 7;
+ else if(i < 288)
+ lens[i] = 8;
+ }
+ calculateCodes(lens, codes, codeCount);
+ buildHuffmanTree(lens, codes, codeCount, treeOut);
+
+ uint8_t nDistCodes = 32;
+ uint8_t distCodeLens[nDistCodes];
+ uint16_t distCodes[nDistCodes];
+
+ for(int i = 0; i < nDistCodes; i++)
+ distCodeLens[i] = i;
+
+ calculateCodes(distCodeLens, distCodes, nDistCodes);
+ buildHuffmanTree(distCodeLens, distCodes, nDistCodes, distTreeOut);
+}
+
+void ZLibInflator::buildDynamicHuffmanTree(StreamData* stream)
+{
+ if(haveTree)
+ {
+ tree.free();
+ distTree.free();
+ }
+ buildDynamicHuffmanTree(stream, &tree, &distTree);
+ haveTree = true;
+}
+void ZLibInflator::buildDynamicHuffmanTree(StreamData* stream, HuffmanTree* treeOut, HuffmanTree* distTreeOut)
+{
+ unsigned int nLitCodes = nextBits(stream, 5) + 257;
+ uint16_t litCodes[nLitCodes];
+
+ unsigned int nDistCodes = nextBits(stream, 5) + 1;
+ uint16_t distCodes[nDistCodes];
+
+ uint8_t codeLens[nLitCodes + nDistCodes];
+
+ uint8_t nLenCodes = nextBits(stream, 4) + 4;
+ uint8_t lenCodeLens[19];
+ memset(lenCodeLens, 0, sizeof(uint8_t)*19);
+ uint16_t lenCodes[19];
+
+ const static uint8_t lenCodeOrder[] = { 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 };
+
+ for(int i = 0; i < nLenCodes; i++)
+ lenCodeLens[lenCodeOrder[i]] = nextBits(stream, 3);
+
+ calculateCodes(lenCodeLens, lenCodes, 19);
+ HuffmanTree lenCodeTree;
+ buildHuffmanTree(lenCodeLens, lenCodes, 19, &lenCodeTree);
+
+ int i = 0;
+ uint8_t extraBits[] = {2, 3, 7};
+ uint8_t repStarts[] = {3, 3, 11};
+
+ while(i < nLitCodes + nDistCodes)
+ {
+ uint16_t code = getNextCode(stream, &lenCodeTree);
+ if(code < 16)
+ codeLens[i++] = (uint8_t)code;
+ else if(code < 19)
+ {
+ code -= 16;
+ int reps = repStarts[code] + nextBits(stream, extraBits[code]);
+ uint8_t len = 0;
+ if(code == 0)
+ {
+ if(i == 0)
+ cout << "Trying to repeat non existent value in dynamic huffman code creation" << endl;
+ else
+ len = codeLens[i - 1];
+ }
+ if(i + reps > nLitCodes + nDistCodes)
+ cout << "other big errror oh no " << i << " " << 0+reps << endl;
+ for(int j = 0; j < reps; j++)
+ {
+ codeLens[i++] = len;
+ }
+ }
+ else
+ {
+ cout << "big error oh no" << endl;
+ }
+ }
+
+ calculateCodes(codeLens, litCodes, nLitCodes);
+ buildHuffmanTree(codeLens, litCodes, nLitCodes, treeOut);
+
+ calculateCodes(&codeLens[nLitCodes], distCodes, nDistCodes);
+ buildHuffmanTree(&codeLens[nLitCodes], distCodes, nDistCodes, distTreeOut);
+}
+
+void HuffmanTree::free()
+{
+ if(left != nullptr)
+ {
+ delete left;
+ left = nullptr;
+ }
+ if(right != nullptr)
+ {
+ delete right;
+ right = nullptr;
+ }
+ val = 0xFFFF;
+}
+HuffmanTree::~HuffmanTree()
+{
+ free();
+}
+
+uint16_t ZLibInflator::getNextCode(StreamData* stream)
+{
+ return getNextCode(stream, &tree);
+}
+
+uint16_t ZLibInflator::getNextCode(StreamData* stream, HuffmanTree* tree)
+{
+ while(tree->val == 0xFFFF)
+ {
+ bool right = nextBit(stream);
+
+ if(tree->left == nullptr && !right)
+ cout << "bad left" << endl;
+ if(tree->right == nullptr && right)
+ cout << "bad right" << endl;
+ tree = (right)?tree->right:tree->left;
+ }
+ return tree->val;
+}
+
+bool ZLibInflator::nextBit(StreamData* stream)
+{
+ long bit = stream->pos % 8;
+ long byte = stream->pos / 8;
+ if(byte >= stream->length)
+ {
+ cout << byte << " " << stream->length << endl;
+ throw std::out_of_range("Ran out of compressed data");
+ }
+ stream->pos++;
+ //cout << ((stream->data[byte] & (0b1 << bit))?"1":"0");
+ return (stream->data[byte] & (0b1 << bit))?0b1:0b0;
+}
+
+uint16_t ZLibInflator::nextBits(StreamData* stream, int bits)
+{
+ if(bits > 16)
+ cout << "Too many bits(" << bits << ")!!!" << endl;
+ uint16_t out = 0;
+ for(int i = 0; i < bits; i++)
+ {
+ out |= nextBit(stream) << i;
+ }
+ return out;
+}
+
+int ZLibInflator::decodeData(uint8_t* data, unsigned long length, uint8_t* out, unsigned long outLength)
+{
+ staticTree = false;
+ const unsigned int lenStart[] = { /* Size base for length codes 257..285 */
+ 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
+ 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258};
+ const unsigned int lenExtra[] = { /* Extra bits for length codes 257..285 */
+ 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
+ 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0};
+
+
+ const unsigned int distStart[] = { /* Offset base for distance codes 0..29 */
+ 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
+ 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
+ 8193, 12289, 16385, 24577};
+ const unsigned int distExtra[] = { /* Extra bits for distance codes 0..29 */
+ 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
+ 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
+ 12, 12, 13, 13};
+
+
+ StreamData stream = { data, length, 0};
+ long outPos = 0;
+
+ bool final;
+ do
+ {
+
+ final = nextBit(&stream);
+ int method = nextBits(&stream, 2);
+
+ //cout << (final?"Final chunk!\n":"") << "Compression method: " << method << endl;
+
+ //cout << outPos << " " << outLength << endl;
+
+ if(method == 1)
+ buildStaticHuffmanTree();
+ else if(method == 2)
+ buildDynamicHuffmanTree(&stream);
+ else if(method == 0)
+ {
+ while(stream.pos%8 != 0)
+ stream.pos++;
+ uint16_t LEN = nextBits(&stream, 16);
+ uint16_t NLEN = nextBits(&stream, 16);
+ NLEN++;
+ if(LEN + NLEN != 0)
+ throw std::invalid_argument("NLEN and LEN don't match");
+ for(int i = 0; i < LEN; i++)
+ out[outPos++] = nextBits(&stream, 8);
+ continue;
+ }
+ else
+ {
+ cout << "Reserved???" << endl;
+ return -1;
+ }
+
+ uint16_t code;
+ do
+ {
+ code = getNextCode(&stream);
+ if(outPos > outLength && code != 256)
+ {
+ throw std::out_of_range("No more space left in image (normal)");
+ }
+ if(code < 256)
+ out[outPos++] = (uint8_t)code;
+ else if(code > 256)
+ {
+ unsigned int len = lenStart[code-257] + (int)nextBits(&stream, lenExtra[code-257]);
+ unsigned int distCode = getNextCode(&stream, &distTree);
+
+ unsigned int dist = distStart[distCode] + (int)nextBits(&stream, distExtra[distCode]);
+ if(outPos + len > outLength)
+ {
+ throw std::out_of_range("No more space left in image (RLE error)");
+ }
+ for(int i = 0; i < len; i++)
+ {
+ out[outPos] = out[outPos - dist];
+ outPos++;
+ }
+ }
+ }
+ while(code != 256);
+ }
+ while(!final);
+
+ return 0;
+}
diff --git a/src/zlib.h b/src/zlib.h
new file mode 100644
index 0000000..d9d6b44
--- /dev/null
+++ b/src/zlib.h
@@ -0,0 +1,52 @@
+#pragma once
+
+#include <cstdint>
+#include <string>
+
+struct HuffmanTree
+{
+ uint16_t val = 0xFFFF;
+ HuffmanTree* left = nullptr; // 0
+ HuffmanTree* right = nullptr; // 1
+ HuffmanTree() = default;
+ ~HuffmanTree();
+ void free();
+};
+
+struct StreamData
+{
+ uint8_t* data;
+ unsigned long length;
+ unsigned long pos;
+};
+
+class ZLibInflator
+{
+private:
+ HuffmanTree tree;
+ HuffmanTree distTree;
+ bool staticTree = false;
+ bool haveTree = false;
+public:
+ ZLibInflator() = default;
+ ~ZLibInflator() = default;
+
+ static bool nextBit(StreamData* stream);
+ static uint16_t nextBits(StreamData* stream, int bits); // Max 16 bits
+
+ void calculateCodes(uint8_t* lengths, uint16_t* codesOut, int codeCount);
+
+ void buildHuffmanTree(uint8_t* lengths, uint16_t* codes, int codeCount);
+ void buildHuffmanTree(uint8_t* lengths, uint16_t* codes, int codeCount, HuffmanTree* treeOut);
+
+ void buildStaticHuffmanTree();
+ void buildStaticHuffmanTree(HuffmanTree* treeOut, HuffmanTree* distTreeOut);
+
+ void buildDynamicHuffmanTree(StreamData* stream);
+ void buildDynamicHuffmanTree(StreamData* stream, HuffmanTree* treeOut, HuffmanTree* distTreeOut);
+
+ uint16_t getNextCode(StreamData* stream);
+ uint16_t getNextCode(StreamData* stream, HuffmanTree* tree);
+
+ int decodeData(uint8_t* data, unsigned long length, uint8_t* out, unsigned long outLength);
+};
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-11 22:20:08 +0000