LZC - new fastr LZ coder compressor -

[Nania Francesco]

LZC v.0.01
LZCode file compressor/archiver
By Nania Francesco Antonio(Italy)
Copyright © 2007 . All rights reserved.

To compress: lzc c mode in out
To decompress: lzc d input
options:
1= 128 MB to compress 2MB to decompress
2= 256 MB to compress 4MB to decompress
3= 512 MB to compress 8MB to decompress
4= 1 GB to compress 16MB to decompress
Total Memory = Merory mode + 100 MB
(LZ:ROLZ:LZP:LZMA) vs LZC
LZMA = best POSITION for best LENGTH;
LZ:ROLZ:LZP = POSITION + LENGTH;
LZC= BEST CODE for best POSITION & LENGHT

[Bulat Ziganshin]

like me, you forget to include download url

as i already said in rissuin, i fell nyself like i started arms race

[encode]

Link:
LZC.zip (9 KB)



However, I'm not really understand what is LZC? Actually, COMPRESS uses the LZC algorithm (LZW with flat binary codes).

[Bulat Ziganshin]

thanks, encode! i also interested in knowuing details, but it may be hard for Nania to describe it. may be you can show some (pseudo)code?

[Nania Francesco]

unfortunately my English is not optimal! LZC as you have been able to see is one my invention that I want guardian (All rights reserved) and I cannot reveal a pseudo-code to you! however considered approval and that it is Encode that Bulat is very great computer programmers I can only say if you are interested to one production with of the program? All why in the first place this is they puts into effect it simple bit encoder !!
int decode()
{
int y;if (bitto==0) ogge=getc(in);y=(ogge>>(7-bitto))&1;
bitto++;if (bitto==8) bitto=0;
return y;
}
void encode(int y)
{
ogge+=(y<<(7-bitto));
bitto++;
if (bitto==8){putc(ogge,out);ogge=0;tot++;bitto=0;}
}
I only say you that theoretically in enwik8 and enwik9 30-32 MB could be arrived with the computer used from Matt to these turns out to you ENWIK8 20 sec ENWIK9 245-250 MB in 200 sec If you want to co-produce this Encode program has my email contact me!

[encode]

Probably, LZC indeed uses LZC algo. Just try to "compress" A10.jpg, and see how it inflates. The worst case with LZW is that file can be inflated by ~125...150%. However, LZW shouldn't use such amounts of memory. Maybe the catch in implementation.

LZC is well known modification of LZW. IMHO, this algo is crap in all terms - compression and speed even with best and very clever implementation. LZP with byte aligned output can provide much higher compression and speed.

Anyway, some info about LZW/LZC:

The output of the LZW consists of only the codes of phrases.

Initially, dictionary contains all one-byte phrases - from 0 to 255, so we able to code any byte from input.

During compression/decompression, coder adds a new phrases to the dictionary. Example of the dictionary:
0 - 0
...
255 - 255
256 - "ab"
257 - "ba"
258 - "abc"
...
At each step, encoder firstly checks, is dictionary contains current phrase? If it does, encoder moves to the next byte, and again checks... until new phrase will be found. After it adds this new phrase to the dictionary, and outputs code corresponding to the longest phrase found.

The baseline LZW uses fixed length codes. For example, if we have dictionary with 65536 entries it will use an 16-bit codes.

The LZC, modification of the LZW uses flat binary codes. Like I said, initially, LZW contains only 256 phrases, after each step dictionary will grow up. For example, if we have, 256 phrases we can encode a codewords using 8-bits, if we have 512 or less phrases we can encode a codewords using 9-bits, and so on. In other words, the code size will grow up according the the actual dictionary size.

Another thing, is called LZW-ARI. Instead of coding codewords with flat binary codes, we can use a flat arithmetic coder - all codes have the same probabliliry, but the alphabet size is increased according to the dictionary size.

In addition, here we have a few ways, what to do if dictionary becomes full:
+ Froze the dictionary (As with original LZW)
+ Reset the dictionary
+ Froze the dictionary while we get descent compression, after reset the dictionary and drop the special RESET codeword, so, decoder will also resets the dictionary syncronously.

Note that, some time ago I implemented all variations of things described above, including LZW-ARI. The conclusion - it's not worth it. In 80's LZW was very popular, but today...

[encode]

void PutBit(int bit) /* Output one bit (bit = 0,1) */
{
static unsigned int buffer = 0, mask = 128;

if (bit) buffer |= mask;
if ((mask >>= 1) == 0) {
if (putc(buffer, outfile) == EOF) Error("Write Error");
buffer = 0; mask = 128; codesize++;
}
}

void FlushBitBuffer(void) /* Send remaining bits */
{
int i;

for (i = 0; i < 7; i++) PutBit(0);
}

int GetBit(void) /* Get one bit (0 or 1) */
{
static unsigned int buffer, mask = 0;

if ((mask >>= 1) == 0) {
buffer = getc(infile); mask = 128;
}
return ((buffer & mask) != 0);
}

[LovePimple]

Thanks Nania and Ilia!

Mirror: Download

[Bulat Ziganshin]

unfortunately my English is not optimal!

unlike your encoding

if you want to keep this in secret, lets do. 10 years ago i thought in the same direction, then i realized that noone is really interested in kidnapping my ideas

[encode]

Today, it's impossible to invent something really new. A couple of VERY smart people already invented all things that can be invented. All new stuff (including LZPM) is just varitions of already invented things.

[Bulat Ziganshin]

if you are interested to one production with of the program?

i dont really understand your question but guess you are asking why we are so interested. the answer is that i try to uderstand each new idea, because the new methods typically come in long way through the different people

[Bulat Ziganshin]

encode
he doesn't say anything about lzw, so it seems that his LZC method has nothing common with lz78 family - ?????? ????????? ?????????? ????

[Bulat Ziganshin]

int decode()

it is very inefficient ones. try smth like this:

unsigned word, buf[SIZE],*p=buf;

encode(int bit)
{
if word>(unsigned)INT_MAX
{
*p++ = word*2+bit;
if (p==buf+SIZE) write(outfile,buf,SIZE),p=buf;
word = 1;
}
else word = word*2+bit;
}

you can find efficient byte and bit codecs in my tornado

[Nania Francesco]

I do not understand why created you these problems! If you want to collaborate with me to the program I put you to coded disposition source and you with your acquaintances could carry it to highest levels! I cannot put it open source!

[LovePimple]

Quick test...

Test Machine = AMD Sempron 2400+


MC SFC test:

LZC v0.01 (c 1)

A10.jpg > 1,031,362
AcroRd32.exe > 2,111,485
english.dic > 1,463,803
FlashMX.pdf > 4,766,274
FP.LOG > 2,356,353
MSO97.DLL > 2,611,910
ohs.doc > 1,332,228
rafale.bmp > 1,230,813
vcfiu.hlp > 1,337,130
world95.txt > 1,229,658

Total = 19,471,016 bytes


LZC v0.01 (c 2)

A10.jpg > 1,098,426
AcroRd32.exe > 2,144,834
english.dic > 1,485,304
FlashMX.pdf > 5,109,660
FP.LOG > 2,140,932
MSO97.DLL > 2,640,523
ohs.doc > 1,344,601
rafale.bmp > 1,192,465
vcfiu.hlp > 1,334,946
world95.txt > 1,126,751

Total = 19,618,442 bytes


LZC v0.01 (c 3)

A10.jpg > 1,163,805
AcroRd32.exe > 2,176,164
english.dic > 1,508,219
FlashMX.pdf > 5,415,506
FP.LOG > 1,980,300
MSO97.DLL > 2,664,911
ohs.doc > 1,358,091
rafale.bmp > 1,173,044
vcfiu.hlp > 1,332,897
world95.txt > 1,039,116

Total = 19,812,053 bytes


LZC v0.01 (c 4)

A10.jpg > 1,165,182
AcroRd32.exe > 2,195,158
english.dic > 1,529,820
FlashMX.pdf > 5,649,854
FP.LOG > 1,857,545
MSO97.DLL > 2,672,681
ohs.doc > 1,359,406
rafale.bmp > 1,171,400
vcfiu.hlp > 1,335,544
world95.txt > 977,330

Total = 19,913,920 bytes



ENWIK8:

LZC v0.01 (c 1)
Compressed Size = 45,940,785 bytes
Compression Time = 00:00:56.328

LZC v0.01 (c 2)
Compressed Size = 43,940,908 bytes
Compression Time = 00:00:55.187

LZC v0.01 (c 3)
Compressed Size = 42,085,698 bytes
Compression Time = 00:00:54.609

LZC v0.01 (c 4)
Compressed Size = 40,312,925 bytes
Compression Time = 00:00:54.156

[Nania Francesco]

Thanks Love Pimple!

[LovePimple]

Good work Nania!

It would be nice if you could make changes to prevent "a10.jpg" from inflating for the next version.

[Bulat Ziganshin]

I do not understand why created you these problems! If you want to collaborate with me to the program I put you to coded disposition source and you with your acquaintances could carry it to highest levels! I cannot put it open source!

no problems. i tried to help you with better bit coder, if you interested. if you dont want to talk about lzc - its also ok

[Nania Francesco]

Thanks for the aid Bulat indeed! If you have interest my offer is always valid !
contact me email : miadebora@hotmail.it

[Uwe Herklotz]

LZC= BEST CODE for best POSITION & LENGHT

I checked the results posted by LovePimple and I think we must have different definitions of "BEST".

[Nania Francesco]

LZC v.0.02 Copyright © 2007 . All rights reserved.
INSTRUCTIONS
*- better compression; more faster in compression (30-60%)

[LovePimple]

LZC v.0.02 Copyright © 2007 . All rights reserved.
INSTRUCTIONS
*- better compression; more faster in compression (30-60%)

Excellent! Can you upload it to www.mytempdir.com for us?

[Nania Francesco]

LZC 0.02
link:
http://www.mytempdir.com/1324552

[LovePimple]

Thank You Nania!

Mirror: Download

[Nania Francesco]

I now hope not to receive too many critics! but I will very still improve I hope! Hi Love Pimple!

[LovePimple]

Quick test...

Test Machine = AMD Sempron 2400+


MC SFC test:

LZC v0.02 (c 1)

A10.jpg > 941,077
AcroRd32.exe > 2,107,258
english.dic > 1,457,323
FlashMX.pdf > 4,413,711
FP.LOG > 2,345,553
MSO97.DLL > 2,605,475
ohs.doc > 1,329,302
rafale.bmp > 1,228,207
vcfiu.hlp > 1,336,992
world95.txt > 1,221,845

Total = 18,986,743 bytes


LZC v0.02 (c 2)

A10.jpg > 941,077
AcroRd32.exe > 2,084,200
english.dic > 1,432,492
FlashMX.pdf > 4,413,711
FP.LOG > 2,065,918
MSO97.DLL > 2,564,459
ohs.doc > 1,308,825
rafale.bmp > 1,153,992
vcfiu.hlp > 1,298,403
world95.txt > 1,087,131

Total = 18,350,208 bytes


LZC v0.02 (c 3)

A10.jpg > 941,077
AcroRd32.exe > 2,081,897
english.dic > 1,429,189
FlashMX.pdf > 4,413,711
FP.LOG > 1,879,234
MSO97.DLL > 2,545,498
ohs.doc > 1,300,824
rafale.bmp > 1,116,936
vcfiu.hlp > 1,277,312
world95.txt > 986,338

Total = 17,972,016 bytes


LZC v0.02 (c 4)

A10.jpg > 941,077
AcroRd32.exe > 2,084,574
english.dic > 1,438,347
FlashMX.pdf > 4,413,711
FP.LOG > 1,749,127
MSO97.DLL > 2,533,721
ohs.doc > 1,296,347
rafale.bmp > 1,107,043
vcfiu.hlp > 1,270,536
world95.txt > 921,471

Total = 17,755,954 bytes


ENWIK8:

LZC v0.02 (c 1)
Compressed Size = 45,639,272 bytes
Compression Time = 00:00:31.093

LZC v0.02 (c 2)
Compressed Size = 42,353,016 bytes
Compression Time = 00:00:33.093

LZC v0.02 (c 3)
Compressed Size = 39,882,915 bytes
Compression Time = 00:00:35.156

LZC v0.02 (c 4)
Compressed Size = 37,906,466 bytes
Compression Time = 00:00:38.203


thortest.tar:

LZC v0.02 (c 1)
Compressed Size = 354,321,859 bytes
Compression Time = 00:03:43.546

[encode]

Link:
lzc002.zip (9 KB)

[Nania Francesco]

Thanks Encode!

[Vacon]

Hello everyone,

I now hope not to receive too many critics! but I will very still improve

This is my first posting here and I decided to register *not* to educate anyone, but to support the hard work that most posters sometimes suffer from.
You seem to be irritated by some comments that deal with your compressors, but I *really believe* that the comments you get shall help you to improve your efforts! So *please* do not take them as offense, but as support!
I, and as I read nearly anyone commenting your postings, like your work, and consider it worthy to deal with and help you to improve.

Best regards!

[Nania Francesco]

LZC v.0.03

Excuse me! corrected bugs on LZC v.0.02!