jspraul / miniz

Automatically exported from code.google.com/p/miniz
0 stars 0 forks source link

miniz

Automatically exported from code.google.com/p/miniz Readme copied using http://markitdown.medusis.com/

miniz.c is a lossless, high performance data compression library in a single source file that implements the zlib (RFC 1950) and Deflate (RFC 1951) compressed data format specification standards. It supports the most commonly used functions exported by the zlib library, but is a completely independent implementation so zlib's licensing requirements do not apply. miniz.c also contains simple to use functions for writing.PNG format image files and reading/writing/appending .ZIP format archives. miniz's compression speed has been tuned to be comparable to zlib's, and it also has a specialized real-time compressor function designed to compare well against fastlz/minilzo.

Here's a blog update on miniz bug fixes/enhancements, and the current status of zip64 support. The source to the zip64 variant of miniz was released as part of Valve's vogl codebase: http://richg42.blogspot.com/2014/03/zip64-version-of-miniz-library-released.html

Whenever I get the chance I'll be extracting it and releasing it separately.

[]()Features

I've created a wiki page comparing the speed and compression ratio of miniz.c verses various other open source codecs. miniz.c's compression ratio is very close to zlib's (sometimes better, or sometimes a small amount worse because it tends to output more blocks due to using less memory), and is typically (but not always) faster, even without any platform specific assembly language optimizations.

[]()Release History

The inflator in raw (non-zlib) mode is now usable on gzip or similar streams that have a bunch of bytes following the raw deflate data (problem discovered by rustyzip author williamw520). This version should never read beyond the last byte of the raw deflate data independent of how many bytes you pass into the input buffer.

The inflator now has a new failure status TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS (-4). Previously, if the inflator was starved of bytes and could not make progress (because the input buffer was empty and the caller did not set the TINFL_FLAG_HAS_MORE_INPUT flag - say on truncated or corrupted compressed data stream) it would append all 0's to the input and try to soldier on. This is scary behavior if the caller didn't know when to stop accepting output (because it didn't know how much uncompressed data was expected, or didn't enforce a sane maximum). v1.16 will instead return TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS immediately if it needs 1 or more bytes to make progress, the input buf is empty, and the caller has indicated that no more input is available. This is a "soft" failure, so you can call the inflator again with more input and it will try to continue, or you can give up and fail. This could be very useful in network streaming scenarios.

The other minor changes are documented at the top of miniz.c.

r4 STABLE - Oct 13, 2013: Merged over a few very minor bug fixes that I fixed in the zip64 branch. This is downloadable from hereand also in SVN head (as of 10/19/13).

ifndef MINIZ_NO_TIME guards around 

utime.h includes. Adding 

mz_free() function, so the caller can free heap blocks returned by miniz using whatever heap functions it has been configured to use, MSVC specific fixes to use "safe" variants of several functions (localtime_s, fopen_s, freopen_s).

static __forceinline is giving it fits, so I'm changing all usage of 

__forceinline to MZ_FORCEINLINE and forcing gcc to use

attribute((__always_inline__)) (and MSVC to use 

__forceinline). Also various fixes from fermtect for MinGW32: added #include <time.h>, 64-bit ftell/fseek fixes.

Eliminated a bunch of warnings when compiling with GCC 32-bit/64. Ran all examples, miniz.c, and tinfl.c through MSVC 2008's /analyze (static analysis) option and fixed all warnings (except for the silly "Use of the comma-operator in a tested expression.." analysis warning, which I purposely use to work around a MSVC compiler warning).

Created 32-bit and 64-bit Codeblocks projects/workspace. Built and tested Linux executables. The codeblocks workspace is compatible with Linux+Win32/x64. Added miniz_tester solution/project, which is a useful little app derived from LZHAM's tester app that I use as part of the regression test. Ran miniz.c and tinfl.c through another series of regression testing on ~500,000 files and archives. Modified example5.c so it purposely disables a bunch of high-level functionality (MINIZ_NO_STDIO, etc.). (Thanks to corysama for the MINIZ_NO_STDIO bug report.)

Fix ftell() usage in a few of the examples so they exit with an error on files which are too large (a limitation of the examples, not miniz itself). Fix fail logic handling in mz_zip_add_mem_to_archive_file_in_place() so it always calls mz_zip_writer_finalize_archive() and mz_zip_writer_end(), even if the file add fails.

[]()Details

miniz.c employs several proven techniques and approaches used in my much more powerful (but slower and more complex) lossless lzhamcodec, and in my jpeg-compressor project. Specifically, miniz.c's linear-time Huffman codelength generator, the single switch-statement approach used to implement the decompressor as a single function coroutine, and its very fast Huffman code symbol unpacker are all loosely based off the implementations I wrote for these earlier projects.

miniz.c v1.10 includes an optimized real-time compressor written specifically for compression level 1 (MZ_BEST_SPEED). miniz.c's level 1 compression ratio is around 5-9% higher than other real-time compressors, such as minilzofastlz, or liblzf. miniz.c's level 1 data consumption rate on a Core i7 3.2 GHz typically ranges between 70-120.5 MB/sec. Between levels 2-9, miniz.c is designed to compare favorably against zlib, where it typically has roughly equal or better performance.

miniz.c has been compiled and tested under Windows (x86 and x64, using MSVC 2005/2008 and GCC 4.5.0 using TDM-GCC x64), and under 32-bit Ubuntu Linux (Lucid Lynx using gcc). The code should be endian safe, but I have not tested it on a big endian platform yet. For maximum compatibility, unused or unnecessary features of miniz.c can be completely disabled via #define's - see the code for details.

miniz's PNG writer has been verified using the pngcheck tool.

If you use miniz, I would greatly appreciate it if you sent me an email with any feedback, or info on how you're using it in practice.

[]()Implemented zlib Functions

miniz.c implements the following standard zlib functions:

deflateInit, 

deflateInit2, 

deflateReset, 

deflate, 

deflateEnd, 

deflateBound

compress, 

compress2, 

compressBound

inflateInit, 

inflateInit2, 

inflate, 

inflateEnd

uncompress

alloc_func, 

free_func

z_stream

crc32, 

adler32, 

zError, 

zlib_version

Note, miniz.c actually prefixes its zlib-like functions and macros with the 

mz_ or 

MZ_ prefixes, and it uses a set of macros to redefine the zlib symbols to miniz's internal symbols. These optional remapping macros can be disabled by defining the 

MINIZ_NO_ZLIB_COMPATIBLE_NAMESmacro, allowing miniz.c to be mixed with zlib in the same compilation unit.

[]()Simple Examples

Include miniz.c and call one of these (zlib-style) helper functions for:

Memory to Memory Compression

  // Returns Z_OK on success, or one of the error codes from deflate() on failure.
  int compress(BytepDest, uLong pDest_len,constByte*pSource, uLong source_len);

  // Like compress() but with more control, level may range from 0 (storing) to 9 (max. compression)
  int compress2(BytepDest, uLong pDest_len,constByte*pSource, uLong source_len,int level);

Memory to Memory Decompression

  // Returns Z_OK on success, or one of the error codes from inflate() on failure.
  int uncompress(BytepDest, uLong pDest_len,constByte*pSource, uLong source_len);

Extract a single file from a ZIP archive:

  // Returns pointer to extracted file, or NULL on failure.
  voidmz_zip_extract_archive_file_to_heap(constcharpZip_filename,constcharpArchive_name,
    size_t
pSize, mz_uint zip_flags);

Add a single file to an existing ZIP archive, or create a new archive, with optional file comment:

  mz_bool mz_zip_add_mem_to_archive_file_in_place(constcharpZip_filename,constcharpArchive_name,
    constvoidpBuf, size_t buf_size,constvoidpComment, mz_uint16 comment_size, mz_uint level_and_flags);

Write a .PNG format image to a file in memory:

  voidtdefl_write_image_to_png_file_in_memory(constvoidpImage,int w,int h,int num_chans, size_t *pLen_out);

These are just basic examples. For more control, miniz.c has many helper functions to compress zlib and raw deflate streams to/from files, memory, the heap, or user supplied callbacks. For maximum control, miniz.c also includes a set of low-level "tinfl" (decompression) and "tdefl" (compression) coroutine functions that read and write to user supplied buffers and do not use the heap in any way -- see the top of the file for more details. (Internally, all high level compression related functions are implemented in terms of tinfl and tdefl.)

[]()Known Problems

[]()Special Thanks

Thanks to Alex Evans for the PNG writer function. Also, thanks to Paul Holden and Thorsten Scheuermann for feedback and testing, Matt Pritchard for all his encouragement, and Sean Barrett's various public domain libraries for inspiration (and encouraging me to write miniz.c in C, which was much more enjoyable and less painful than I thought it would be considering I've been programming in C++ for so long).

Thanks to Bruce Dawson <bruced@valvesoftware.com> for reporting a problem with the level_and_flags archive API parameter (which is fixed in v1.12) and general feedback, and Janez Zemva <janezz55@gmail.com> for indirectly encouraging me into writing more examples.

[]()Patents

I was recently asked if miniz avoids patent issues. miniz purposely uses the same core algorithms as the ones used by zlib. The compressor uses vanilla hash chaining as described here. Also see the gzip FAQ. In my opinion, if miniz falls prey to a patent attack then zlib/gzip are likely to be at serious risk too.

[]()Building the Examples Under Linux

v1.15 includes a CMakeLists.txt file to use with cmake:

   cmake .
   make clean
   make

This will build the x64 release examples by default on x64 systems. Optionally, you can use ccmake (package cmake-curses-gui) to configure the project after running cmake ("ccmake ."). I've built and tested v1.15 with tcc 0.9.26 (Tiny C Compiler), clang v3.3, and gcc 4.6.3.

[]()Example Projects

I've included a Visual Studio 2008 solution (examples.sln) with 6 examples:

For any questions or problems with this code please contact Rich Geldreich at <richgel99 at gmail.com>. Here's my twitter page.