Screenshot of physical vice to represent compression

JCALG1 – Compression Library

This is a legacy project, a proprietary lossless compression algorithm based loosely on LZSS. It is written entirely in x86 assembly. It compresses very well, though is slow at compress (as fast as anything at decompression). It has a unique feature of being able to handle pre-compressed or high entropy data.

I later wrote C++ editions. If you want a proprietary compression algorithm that performs well for your company, email to see if we have anything to offer.

JCALG1 r5.xx
(c)1999-2011 Jeremy Collake

This software is provided as-is, without warranty of ANY KIND, either expressed or implied, including but not limited to the implied warranties of merchantability and/or fitness for a particular purpose. The author shall NOT be held liable for ANY damage to you, your computer, or to anyone or anything else, that may result from its use, or misuse. Basically, you use it at YOUR OWN RISK.

This library may be used freely in non-commercial applications. Commercial applications require a VERY MODEST licensing fee. Please contact for licensing.

First of all, I would like to credit Joergen Ibsen, author of apLib, for many suggestions and help with my development of JCALG1. You may notice some similarities between the JCALG1 decompressor and that of apLib. The reason for this is because I could not come up with a better encoding scheme than Joergen Ibsen’s. Therefore, I give him credit to developing much of the encoding scheme that JCALG1 uses.

JCALG1 is a small, open-source, LZSS derived compression library.


  • Coded in 100% 32bit x86 assembly language for maximum performance and minimum size.
  • Good compression ratio, typically much better than ZIP’s deflate.
  • Extremely small and fast decompressor.
  • Adjustable window size to allow for faster compression at the cost of compression ratio.
  • Decompression requires no memory, other than the destination buffer.
  • Easy integration with any application.
  • Free!

Canterbury Corpus compression results for JCALG1 with window size of 128Kb (c7):

File Original Size Compressed Size Bits Per Byte Weighted
bib 111,261 36594 2.63
book1 768,771 335310 3.49
book2 610,856 213402 2.79
geo 102,400 71022 5.55
news 377,109 142978 3.03
obj1 21,504 9982 3.71
obj2 246,814 77722 2.52
paper1 53,161 20014 3.01
paper2 82,199 32534 3.17
pic 513,216 58338 0.91
progc 39,611 14194 2.87
progl 71,646 17174 1.92
progp 49,379 11674 1.89
trans 93,695 19486 1.66
TOTAL: 3,141,622 1060424 2.80 2.70


This is where I tell you what you need to know, ever-so-briefly ;). I decided to document all functions in C, since most people are familar with it. All functions use the STDCALL calling convention (parameters pushed onto stack, called function adjusts stack before return).

DWORD JCALG1_Compress(
void *Source,
DWORD Length,
void *Destination,
DWORD WindowSize,
AllocFunc *pAlloc,
DeAllocFunc *pDealloc,
CallbackFunc *pCallback,
BOOLEAN bDisableChecksum);


  • Source is a pointer to the uncompressed data.
  • Length is the size of the uncompressed data.
  • Destination is a pointer to the destination of the compressed data. This buffer should be at least 4 bytes larger than the uncompressed data size. One can retrieve the minimum needed buffer size by issuing a call to JCALG1_GetNeededBufferSize.
  • WindowSize is a nonzero value up to the size of the file. The larger, the better the compression ratio but the slower the compression.
  • pAlloc is a pointer to a memory allocation function. See prototype below.
  • pDealloc is a pointer to a memory deallocation function. See prototype below.
  • pCallback is a pointer to a callback function which is called in every iteration of the main compression loop. See protoype below.
  • bDisableChecksum, if TRUE a checksum for the data is not computed/stored.

Returns: Size of the compressed data, or NULL if the data could not be compressed.

DWORD JCALG1_Decompress_Fast(
void *Source,
void *Destination);
DWORD JCALG1_Decompress_Small(
void *Source,
void *Destination);


  • Source is a pointer to the source data.
  • Destination is a pointer to the destination buffer for the uncompressed data.

Returns: Size of the uncompressed data or 0 if the block of data is not a valid JCALG1 compressed stream (failed header or checksum tests). As the function names attempt to infer, the ‘small’ version of this procedure is the one optimized for size while the ‘fast’ version is the one optimized for speed.

DWORD JCALG1_GetUncompressedSizeOfCompressedBlock(
void *pBlock)


  • pBlock is a pointer to a JCALG1 compressed block of memory.

Returns: Size of the uncompressed block, in bytes.


DWORD JCALG1_GetNeededBufferSize(
DWORD n Size);


  • n Size is size, in bytes, of a block of uncompressed data.

Returns: The minimum destination buffer size (currently nSize+4).


void JCALG1_GetInfo(
_JCALG1_Info *JCALG1_Info);



  • JCALG1_Info is a pointer to a _JCALG1_Info structure.
typedef struct __JCALG1_Info {
DWORD MajorRev;
DWORD MinorRev;
DWORD FastDecompressorSize;
DWORD SmallDecompressorSize;
} _JCALG1_Info;

BOOL CallbackFunc(
DWORD pCurrentSrc,
DWORD pCurrentDest);



  • pCurrentSrc is the relative offset of the current position in the uncompressed data.
  • pCurrentDest is the relative offset of the current position in the compressed data.

Return: FALSE to stop compression, TRUE to continue compression.


void *AllocFunc(
DWORD Size);


  • Size is the number of bytes requested.

Returns: Pointer to allocated memory block.


BOOL DeallocFunc(
void *pMemory)


  • pMemory is a pointer to the memory block to be deallocated.

Returns: FALSE if failure.