package EBook::Tools::LZSS;
use warnings; use strict; use utf8;
use English qw( -no_match_vars );
use version 0.74; our $VERSION = qv("0.4.8");
# Perl Critic overrides:
## no critic (Package variable)
# RequireBriefOpen seems to be way too brief to be useful
## no critic (RequireBriefOpen)
# Double-sigils are needed for lexical filehandles in clear print statements
## no critic (Double-sigil dereference)
=head1 NAME
EBook::Tools::LZSS - Lempel-Ziv-Storer-Szymanski compression and decompression
=head1 SYNOPSIS
use EBook::Tools::LZSS;
my $lzss = EBook::Tools::LZSS->new(lengthbits => 3,
offsetbits => 14,
windowinit => 'the man');
my $textref = $lzss->uncompress(\$data);
=cut
require Exporter;
use base qw(Exporter);
our @EXPORT_OK;
@EXPORT_OK = qw (
);
our %EXPORT_TAGS = ('all' => [@EXPORT_OK]);
use Bit::Vector;
use Carp;
use EBook::Tools qw(:all);
use Encode;
use File::Basename qw(dirname fileparse);
use File::Path; # Exports 'mkpath' and 'rmtree'
binmode(STDERR,":utf8");
use constant ENCODED => 0;
use constant UNCODED => 1;
####################################################
########## CONSTRUCTOR AND INITIALIZATION ##########
####################################################
my %rwfields = (
'lengthbits' => 'integer',
'offsetbits' => 'integer',
'windowinit' => 'string',
'windowsize' => 'integer',
'windowstart' => 'integer',
'maxuncoded' => 'integer',
'screwybits' => 'boolean',
'verbose' => 'integer',
);
my %rofields = (
);
my %privatefields = (
);
# A simple 'use fields' will not work here: use takes place inside
# BEGIN {}, so the @...fields variables won't exist.
require fields;
fields->import(
keys(%rwfields),keys(%rofields),keys(%privatefields)
);
=head1 CONSTRUCTOR AND INITIALIZATION
=head2 C<new(%args)>
Instantiates a new EBook::Tools::LZSS object.
=head3 Arguments
All arguments are optional, but must be identical between compression
and decompression for the result to be valid.
=over
=item * C<lengthbits>
The number of bits used to encode the length of a LZSS reference. If
not specified defaults to 4 bits.
The eBookwise .IMP format typically compresses with 3 length bits.
Note that the actual length of the LZSS reference in bytes is greater
than the value stored in the length bits. The actual number of bytes
returned is the decoded length bits value plus C<maxuncoded> plus 1,
=item * C<offsetbits>
The number of bits used to encode the offset to a LZSS reference.
This also determines the size of the sliding window of reference data.
If not specified, it defaults to 12 bits (4096-byte window).
The eBookwise .IMP format typically compresses with 14 offset bits
(16384-byte window).
=item * C<windowinit>
A string used to initalize the sliding window. If specified, this
string MUST be the same length as the window size, or the subroutine
will croak. If not specified, the window will be initialized with
spaces.
=item * C<windowstart>
The first byte position in the window that will be overwritten by
decoded text. If not specified, defaults to 0.
=item * C<maxuncoded>
The maximum number of uncoded bytes (?). This currently isn't used
for that purpose, but determines the actual length of a LZSS reference.
=item * C<screwybits>
If set to true and the number of offset bits is greater than 8, then
the offset bits will be read first in a chunk of 8 for the least
significant bits, and then the remaining bits will be read and use as
the most significant bits. This seems to be necessary for
compatibility with Michael Dipperstein's LZSS C compression library
but does not hold true for IMP e-book LZSS decompression.
=item * C<verbose>
If set to true, compression and uncompression will provide additional
status feedback on STDOUT.
=back
=cut
sub new ## no critic (Always unpack @_ first)
{
my $self = shift;
my $class = ref($self) || $self;
my %args = @_;
my $subname = (caller(0))[3];
debug(2,"DEBUG[",$subname,"]");
my %valid_args = (
'lengthbits' => 1,
'offsetbits' => 1,
'windowinit' => 1,
'windowstart' => 1,
'maxuncoded' => 1,
'screwybits' => 1,
'verbose' => 1,
);
foreach my $arg (keys %args)
{
croak($subname,"(): invalid argument '",$arg,"'")
if(!$valid_args{$arg});
}
$self = fields::new($class);
$self->{lengthbits} = $args{lengthbits} || 4;
$self->{offsetbits} = $args{offsetbits} || 12;
$self->{windowsize} = 1 << $self->{offsetbits};
$self->{windowinit} = $args{windowinit} || ' ' x $self->{windowsize};
if(defined $args{windowstart})
{
$self->{windowstart} = $args{windowstart};
}
else { $self->{windowstart} = 0; }
$self->{maxuncoded} = $args{maxuncoded} || 2;
$self->{screwybits} = $args{screwybits};
$self->{verbose} = $args{verbose} || 0;
if($self->{windowinit}
and length($self->{windowinit}) != $self->{windowsize})
{
croak($subname,"(): \n window initialization string size (",
length($self->{windowinit}),
" does not match specified number of offset bits (",
$self->{windowsize},")\n")
}
return $self;
}
#############################
########## METHODS ##########
#############################
=head1 METHODS
=head2 C<uncompress(\$dataref)>
Takes a reference to a compressed data string, uncompresses the data
string and returns a reference to the uncompressed string.
=cut
sub uncompress :method
{
my $self = shift;
my ($dataref) = @_;
my $subname = (caller(0))[3];
debug(2,"DEBUG[",$subname,"]");
croak($subname,"(): no data provided!\n")
unless($dataref);
croak($subname,"(): data is not a reference!\n")
unless(ref $dataref);
my $lengthbits = $self->{lengthbits} || 4;
my $offsetbits = $self->{offsetbits} || 12;
my $windowsize = $self->{windowsize};
my $window = $self->{windowinit};
my $max_uncoded = $self->{maxuncoded} || 2;
my $max_encoded = (1 << $lengthbits) + $max_uncoded;
my $windowpos = $self->{windowstart} || 0;
if(length($window) != $windowsize)
{
croak($subname,"(): \n window initialization string size (",
length($window),
" does not match specified number of offset bits (",
$windowsize,")\n")
}
if($self->{verbose} > 1)
{
print {*STDERR} ("Uncompressing with ",$lengthbits," length bits, ",
$offsetbits," offset bits, windowpos = ",
$windowpos,"\n");
}
my $bitsize = length($$dataref) * 8;
my $bitvector = Bit::Vector->new($bitsize);
my $bitoffset;
my $bitpos = 0;
my $bits;
my $encflag;
my $dataoffset = 0;
my $lzss_offset;
my $lzss_length;
my $lzss_text;
my $uncompressed;
# Unfortunately, the bitstream needs to be processed from left to
# right, and Bit::Vector really likes to process from right to
# left, so we have to process by chunk instead of using
# Block_Store
while($bitpos < $bitsize)
{
$bitoffset = $bitsize-$bitpos;
$bits = unpack('C',substr($$dataref,$dataoffset,1));
$bitvector->Chunk_Store(8,$bitoffset-8,$bits);
$dataoffset++;
$bitpos += 8;
}
$bitpos = 0;
while($bitpos < $bitsize - 8)
{
# If there are less than 8 bits left, there's nothing to
# decode and we can stop. Otherwise, uncompress.
if($self->{verbose} or $EBook::Tools::debug > 0)
{
my $percent = int( ($bitpos / $bitsize) * 100);
print("Uncompressing text... [",$percent,"%]\r")
if($percent % 5 == 0);
}
$bitoffset = $bitsize - $bitpos;
$encflag = $bitvector->Chunk_Read(1,$bitoffset-1);
$bitpos++;
$bitoffset--;
if($encflag == UNCODED)
{
if($bitoffset - 8 < 0)
{
debug(1,"DEBUG: ran out of bits at bit position ",
$bitpos,"!");
last;
}
$bitoffset -= 8;
$bitpos += 8;
$bits = $bitvector->Chunk_Read(8,$bitoffset);
$uncompressed .= chr($bits);
substr($window,$windowpos,1,chr($bits));
$windowpos++;
$windowpos %= $windowsize;
}
else
{
if($bitoffset - ($offsetbits + $lengthbits) < 0)
{
if($bitoffset >= 16)
{
carp($subname,,"(): ran out of bits at bit position ",
$bitpos," [",$bitoffset," bits remaining]!");
}
last;
}
if($self->{screwybits} and $offsetbits > 8)
{
my $lowoffset;
my $highoffset;
my $extrabits = $offsetbits - 8;
$bitoffset -= 8;
$lowoffset = $bitvector->Chunk_Read(8,$bitoffset);
$bitoffset -= $extrabits;
$highoffset = $bitvector->Chunk_Read($extrabits,$bitoffset);
$lzss_offset = (256 * $highoffset) + $lowoffset;
}
else
{
$bitoffset -= $offsetbits;
$lzss_offset = $bitvector->Chunk_Read($offsetbits,$bitoffset);
}
$bitoffset -= $lengthbits;
$lzss_length = $bitvector->Chunk_Read($lengthbits,$bitoffset);
if($lzss_offset > $windowsize)
{
carp($subname,"(): LZSS offset points outside window!\n");
last;
}
if($lzss_length == 0
and $lzss_offset == 0)
{
if($bitoffset >= 16)
{
carp($subname,"(): invalid LZSS code at bit position ",
$bitpos," [",$bitoffset," bits remaining]!\n");
}
last;
}
$bitpos += ($offsetbits + $lengthbits);
$lzss_length += $max_uncoded + 1;
foreach my $lzss_pos (0 .. ($lzss_length - 1) )
{
my $windowoffset;
my $char = substr($window,
($lzss_offset + $lzss_pos) % $windowsize,
1);
substr($window,$windowpos,1,$char);
$windowpos++;
$windowpos %= $windowsize;
$uncompressed .= $char;
}
}
}
if($self->{verbose} or $EBook::Tools::debug > 0)
{
print("Finished uncompressing text.\n");
}
return \$uncompressed;
}
################################
########## PROCEDURES ##########
################################
########## END CODE ##########
=head1 BUGS AND LIMITATIONS
=over
=item * Compression not yet implemented.
=item * The LZSS algorithm isn't documented in the POD.
=item * Once compression is implemented and the algorithm is
documented, this module is a good candidate for being forked off into
its own project. Relying on it to keep its current name may be
unwise.
=back
=head1 AUTHOR
Zed Pobre <zed@debian.org>
The design of this module was based on the C LZSS library by Michael
Dipperstein, version 0.5.2, at http://michael.dipperstein.com/lzss/
=head1 LICENSE AND COPYRIGHT
Copyright 2008 Zed Pobre
Licensed to the public under the terms of the GNU GPL, version 2.
=cut
1;
__END__