1 |
root |
1.1 |
=head1 NAME |
2 |
|
|
|
3 |
|
|
Compress::LZV1 - extremely leight-weight Lev-Zimpel-Vogt compression |
4 |
|
|
|
5 |
|
|
=head1 SYNOPSIS |
6 |
|
|
|
7 |
|
|
use Compress::LZV1; |
8 |
|
|
|
9 |
|
|
$probably_compressed = compress $uncompressed_data; |
10 |
|
|
|
11 |
|
|
$original_data = decompress $probably_compressed; |
12 |
|
|
|
13 |
|
|
=head1 DESCRIPTION |
14 |
|
|
|
15 |
|
|
LZV1 is an extremely fast (not that much slower than a pure memcpy) |
16 |
|
|
compression algorithm. It is ideal for applications where you want to save |
17 |
|
|
I<some> space but not at the cost of speed. It is ideal for repetitive |
18 |
|
|
data as well. The module is self-contained and very small (no large |
19 |
|
|
library to be pulled in). |
20 |
|
|
|
21 |
|
|
I have no idea wether any patents in any countries apply to this |
22 |
|
|
algorithm, but at the moment it is believed that it is free from any |
23 |
|
|
patents. |
24 |
|
|
|
25 |
|
|
=head1 FUNCTIONS |
26 |
|
|
|
27 |
|
|
=head2 $compressed = compress $uncompressed |
28 |
|
|
|
29 |
|
|
Try to compress the given string as quickly and as much as possible. In |
30 |
|
|
the worst case, the string can enlarge by at most a single byte. Empty |
31 |
|
|
strings yield empty strings. The uncompressed data string must be smaller |
32 |
|
|
than 16MB (1<<24). |
33 |
|
|
|
34 |
|
|
The compressed data is (currently) in one of two forms: |
35 |
|
|
|
36 |
|
|
* a literal 'U', followed by the original, uncompressed data |
37 |
|
|
|
38 |
|
|
* a literal 'L', followed by three bytes (big-endian) uncompressed length, followed by the actual LZV1 data |
39 |
|
|
|
40 |
|
|
This could be improved for small (uncompressible strings) to no overhead |
41 |
|
|
on average, but I'm too lazy to do that for just one byte savings ;) |
42 |
|
|
|
43 |
|
|
=head2 $decompressed = decompress $compressed |
44 |
|
|
|
45 |
|
|
Uncompress the string (compressed by C<compress>) and return the original |
46 |
|
|
data. Decompression errors can result in either broken data (there is no |
47 |
|
|
checksum kept) or a runtime error. |
48 |
|
|
|
49 |
|
|
=head1 SEE ALSO |
50 |
|
|
|
51 |
|
|
Other Compress::* modules, especially Compress::LZO (for speed) and |
52 |
|
|
Compress::Zlib. |
53 |
|
|
|
54 |
|
|
=head1 AUTHOR |
55 |
|
|
|
56 |
|
|
This perl extension was written by Marc Lehmann <pcg@goof.com> (See |
57 |
|
|
also http://www.goof.com/pcg/marc/). The original lzv1 code was written |
58 |
|
|
by Hermann Vogt and put under the GPL. (There is also a i386 assembler |
59 |
|
|
version that is not used in this module). |
60 |
|
|
|
61 |
|
|
The lzv1 code was accompanied by the following comment: |
62 |
|
|
|
63 |
|
|
=over 4 |
64 |
|
|
|
65 |
|
|
The method presented here is faster and compresses better than lzrw1 and |
66 |
|
|
lzrw1-a. I named it lzv for "Lev-Zimpel-Vogt". It uses ideas introduced |
67 |
|
|
by Ross Williams in his algorithm lzrw1 [R. N. Williams (1991): "An |
68 |
|
|
Extremly Fast ZIV-Lempel Data Compression Algorithm", Proceedings IEEE |
69 |
|
|
Data Compression Conference, Snowbird, Utah, 362-371] and by Fiala and |
70 |
|
|
Green in their algorithm a1 [E. R. Fiala, D. H. Greene (1989): "Data |
71 |
|
|
Compression with Finite Windows", Communications of the ACM, 4, 490-505]. |
72 |
|
|
Because lzv differs strongly from both, I hope there will be no patent |
73 |
|
|
problems. The hashing-method has been stolen from Jean-loup Gailly's |
74 |
|
|
(patent free) gzip. |
75 |
|
|
|
76 |
|
|
=back |
77 |
|
|
|
78 |
|
|
=cut |
79 |
|
|
|
80 |
|
|
package Compress::LZV1; |
81 |
|
|
|
82 |
|
|
require Exporter; |
83 |
|
|
require DynaLoader; |
84 |
|
|
|
85 |
|
|
$VERSION = 0.04; |
86 |
|
|
@ISA = qw/Exporter DynaLoader/; |
87 |
|
|
@EXPORT = qw(compress decompress); |
88 |
|
|
bootstrap Compress::LZV1 $VERSION; |
89 |
|
|
|
90 |
|
|
1; |
91 |
|
|
|
92 |
|
|
|
93 |
|
|
|
94 |
|
|
|
95 |
|
|
|