uzlib - Deflate/Zlib-compatible LZ77 compression/decompression library

uzlib is a library which can decompress any valid Deflate, Zlib, and Gzip (further called just “Deflate”) bitstream, and compress data to Deflate- compatible bitstream, albeit with lower compression ratio than Zlib Deflate algorithm (very basic LZ77 compression algorithm is used instead, static Deflate Huffman tree encoding is used for bitstream).

uzlib aims for minimal code size and runtime memory requirements, and thus suitable for (deeply) embedded systems.

uzlib is based on:

  • tinf library by Joergen Ibsen (Deflate decompression)

  • Deflate Static Huffman tree routines by Simon Tatham

  • LZ77 compressor by Paul Sokolovsky

Library integrated and maintained by Paul Sokolovsky.

  1. 2014-2020 Paul Sokolovsky

uzlib library is licensed under Zlib license.

Decompressor features

Handling of input (compressed) stream:

  • Can reside (fully) in memory.

  • Can be received, byte by byte, from an application-defined callback function (which e.g. can read it from file or another I/O device).

  • Combination of the above: a chunk of input is buffered in memory, when buffer is exhausted, the application callback is called to refill it.

Handling of output (decompressed) stream:

  • In-memory decompression, where output stream fully resides in memory.

  • Streaming decompression, which allows to process arbitrary-sized streams (longer than available memory), but requires in-memory buffer for Deflate dictionary window.

  • Application specifies number of output bytes it wants to decompress, which can be as high as UINT_MAX to decompress everything into memory at once, or as low as 1 to decompress byte by byte, or any other value to decompress a chunk of that size.

Note that in regard to input stream handling, uzlib employs callback-based, “pull-style” design. The control flow looks as follows:

  1. Application requests uzlib to decompress given number of bytes.

  2. uzlib performs decompression.

  3. If more input is needed to decompress given number of bytes, uzlib calls back into application to provide more input bytes. (An implication of this is that uzlib will always return given number of output bytes, unless end of stream (or error) happens).

The original Zlib library instead features “push-style” design:

  1. An application prepares arbitrary number of input bytes in a buffer, and free space in output buffer, and calls Zlib with these buffers.

  2. Zlib tries to decode as much as possible input and produce as much as possible output. It returns back to the application if input buffer is exhausted, or output buffer is full, whatever happens first.

Currently, uzlib doesn’t support push-style operation a-la Zlib.

Compressor features

Compressor uses very basic implementation of LZ77 algorithm using hash table to find repeating substrings. The size of the hash table (on which compression efficiency depends), pointer to the hashtable memory, and the size of LZ77 dictionary should be configured in struct uzlib_comp.

Currently, compressor doesn’t support streaming operation, both input and output must reside in memory. Neither it supports incremental operation, entire input buffer is compressed at once with a single call to uzlib.

API and configuration

The API is defined in the file uzlib.h and should be largely self-describing. There are also examples implementing gzip-compatible compression and decompression applications in examples/ for further reference. (You may also refer to the original tinf README below for additional information, but it’s mostly provided for historical reference, and uzlib largely evolved beyond it).

There are some compile-time options for the library, defined in the file uzlib_conf.h. They can be altered directly in the file, or passed as the compiler options (-DXXX=YYY) when building library.

Binary sizes

To give an impression of code/data sizes of uzlib, the following figures are provided. Numbers for *.o files are code sizes of individual components (tinflate.o is decompressor, genlz77.o and defl_static.o - compressor), and TINF_DATA is the size of the corresponding data structure. These numbers are provided for different architectures, with default uzlib configuration, and with compilers/their options as specified.

gcc -m32 -Os
gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
2891 src/tinflate.o
381 src/genlz77.o
1685 src/defl_static.o

arm-none-eabi-gcc -mthumb -mcpu=cortex-m4 -Os
arm-none-eabi-gcc (GNU Tools for Arm Embedded Processors 9-2019-q4-major) 9.2.1 20191025 (release) [ARM/arm-9-branch revision 277599]
1620 src/tinflate.o
180 src/genlz77.o
1131 src/defl_static.o

Original tinf library README

For historical reference and to provide proper credit, the original ``tinf`` library README follows. NOTE: Many parts no longer apply to ``uzlib``. In particular, API is different, features supported are largely extended, and during decompression, there are checks to avoid erroneous/undefined behavior on incorrect Deflate bitstreams.

tinf - tiny inflate library

Version 1.00

Copyright (c) 2003 Joergen Ibsen


tinf is a small library implementing the decompression algorithm for the deflate compressed data format (called ‘inflate’). Deflate compression is used in e.g. zlib, gzip, zip and png.

I wrote it because I needed a small in-memory zlib decompressor for a self- extracting archive, and the zlib library added 15k to my program. The tinf code added only 2k.

Naturally the size difference is insignificant in most cases. Also, the zlib library has many more features, is more secure, and mostly faster. But if you have a project that calls for a small and simple deflate decompressor, give it a try :-)

While the implementation should be fairly compliant, it does assume it is given valid compressed data, and that there is sufficient space for the decompressed data.

Simple wrappers for decompressing zlib streams and gzip’ed data in memory are supplied.

tgunzip, an example command-line gzip decompressor in C, is included.

The inflate algorithm and data format are from ‘DEFLATE Compressed Data Format Specification version 1.3’ (RFC 1951).

The zlib data format is from ‘ZLIB Compressed Data Format Specification version 3.3’ (RFC 1950).

The gzip data format is from ‘GZIP file format specification version 4.3’ (RFC 1952).

Ideas for future versions:

  • the fixed Huffman trees could be built by tinf_decode_trees() using a small table

  • memory for the TINF_DATA object should be passed, to avoid using more than 1k of stack space

  • wrappers for unpacking zip archives and png images

  • implement more in x86 assembler

  • more sanity checks

  • in tinf_uncompress, the (entry value of) destLen and sourceLen are not used

  • blocking of some sort, so everything does not have to be in memory

  • optional table-based huffman decoder


void tinf_init();

Initialise the global uninitialised data used by the decompression code. This function must be called once before any calls to the decompression functions.

int tinf_uncompress(void *dest,
                    unsigned int *destLen,
                    const void *source,
                    unsigned int sourceLen);

Decompress data in deflate compressed format from source[] to dest[]. destLen is set to the length of the decompressed data. Returns TINF_OK on success, and TINF_DATA_ERROR on error.

int tinf_gzip_uncompress(void *dest,
                         unsigned int *destLen,
                         const void *source,
                         unsigned int sourceLen);

Decompress data in gzip compressed format from source[] to dest[]. destLen is set to the length of the decompressed data. Returns TINF_OK on success, and TINF_DATA_ERROR on error.

int tinf_zlib_uncompress(void *dest,
                         unsigned int *destLen,
                         const void *source,
                         unsigned int sourceLen);

Decompress data in zlib compressed format from source[] to dest[]. destLen is set to the length of the decompressed data. Returns TINF_OK on success, and TINF_DATA_ERROR on error.

unsigned int tinf_adler32(const void *data,
                          unsigned int length);

Computes the Adler-32 checksum of length bytes starting at data. Used by tinf_zlib_uncompress().

unsigned int tinf_crc32(const void *data,
                        unsigned int length);

Computes the CRC32 checksum of length bytes starting at data. Used by tinf_gzip_uncompress().

Source Code

The source code is ANSI C, and assumes that int is 32-bit. It has been tested on the x86 platform under Windows and Linux.

The decompression functions should be endian-neutral, and also reentrant and thread-safe (not tested).

In src/nasm there are 32-bit x86 assembler (386+) versions of some of the files.

Makefiles (GNU Make style) for a number of compilers are included.

Frequently Asked Questions

Q: Is it really free? Can I use it in my commercial ExpenZip software?

A: It’s open-source software, available under the zlib license (see

later), which means you can use it for free – even in commercial products. If you do, please be kind enough to add an acknowledgement.

Q: Did you just strip stuff from the zlib source to make it smaller?

A: No, tinf was written from scratch, using the RFC documentation of

the formats it supports.

Q: What do you mean by: ‘the zlib library .. is more secure’?

A: The zlib decompression code checks the compressed data for validity

while decompressing, so even on corrupted data it will not crash. The tinf code assumes it is given valid compressed data.

Q: I’m a Delphi programmer, can I use tinf?

A: Sure, the object files produced by both Borland C and Watcom C should

be linkable with Delphi.

Q: Will tinf work on UltraSTRANGE machines running WhackOS?

A: I have no idea .. please try it out and let me know!

Q: Why are all the makefiles in GNU Make style?

A: I’m used to GNU Make, and it has a number of features that are missing

in some of the other Make utilities.

Q: This is the first release, how can there be frequently asked questions?

A: Ok, ok .. I made the questions up ;-)


tinf - tiny inflate library

Copyright (c) 2003 Joergen Ibsen

This software is provided ‘as-is’, without any express or implied warranty. In no event will the authors 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.