Manpages - IO_Compress_FAQ.3perl

Although Compress::Zlib has a pair of functions called compress and uncompress, they are not related to the Unix programs of the same name. The Compress::Zlib module is not compatible with Unix compress.

If you have the uncompress program available, you can use this to read compressed files

open F, “uncompress -c $filename |”; while (<F>) { …

Alternatively, if you have the gunzip program available, you can use this to read compressed files

open F, “gunzip -c $filename |”; while (<F>) { …

and this to write compress files, if you have the compress program available

open F, “| compress -c $filename ”; print F “data”; … close F ;

The Archive::Tar module can optionally use Compress::Zlib (via the IO::Zlib module) to access tar files that have been compressed with gzip. Unfortunately tar files compressed with the Unix compress utility cannot be read by Compress::Zlib and so cannot be directly accessed by Archive::Tar.

If the uncompress or gunzip programs are available, you can use one of these workarounds to read .tar.Z files from Archive::Tar

Firstly with uncompress

use strict; use warnings; use Archive::Tar; open F, “uncompress -c $filename |”; my $tar = Archive::Tar->new(*F); …

and this with gunzip

use strict; use warnings; use Archive::Tar; open F, “gunzip -c $filename

Similarly, if the compress program is available, you can use this to write a .tar.Z file

use strict; use warnings; use Archive::Tar; use IO::File; my $fh = IO::File->new( “| compress -c >$filename” ); my $tar = Archive::Tar->new(); … $tar->write($fh); $fh->close ;

This is easier that you might expect if you realise that all the IO::Compress::* objects are derived from IO::File and that all the IO::Uncompress::* modules can read from an IO::File filehandle.

So, for example, say you have a file compressed with gzip that you want to recompress with bzip2. Here is all that is needed to carry out the recompression.

use IO::Uncompress::Gunzip :all; use IO::Compress::Bzip2 :all; my $gzipFile = “somefile.gz”; my $bzipFile = “somefile.bz2”; my $gunzip = IO::Uncompress::Gunzip->new( $gzipFile ) or die “Cannot gunzip $gzipFile: $GunzipError\n” ; bzip2 $gunzip => $bzipFile or die “Cannot bzip2 to $bzipFile: $Bzip2Error\n” ;

Note, there is a limitation of this technique. Some compression file formats store extra information along with the compressed data payload. For example, gzip can optionally store the original filename and Zip stores a lot of information about the original file. If the original compressed file contains any of this extra information, it will not be transferred to the new compressed file using the technique above.

support? The following compression formats are supported by IO::Compress::Zip and IO::Uncompress::Unzip

• Store (method 0) No compression at all.
• Deflate (method 8) This is the default compression used when creating a zip file with IO::Compress::Zip.
• Bzip2 (method 12) Only supported if the IO-Compress-Bzip2 module is installed.
• Lzma (method 14) Only supported if the IO-Compress-Lzma module is installed.

Yes, both the IO-Compress-Zip and IO-Uncompress-Unzip modules support the zip feature called Zip64. That allows them to read/write files/buffers larger than 4Gig.

If you are creating a Zip file using the one-shot interface, and any of the input files is greater than 4Gig, a zip64 complaint zip file will be created.

zip “really-large-file” => “my.zip”;

Similarly with the one-shot interface, if the input is a buffer larger than 4 Gig, a zip64 complaint zip file will be created.

zip \$really_large_buffer => “my.zip”;

The one-shot interface allows you to force the creation of a zip64 zip file by including the Zip64 option.

zip $filehandle => “my.zip”, Zip64 => 1;

If you want to create a zip64 zip file with the OO interface you must specify the Zip64 option.

my $zip = IO::Compress::Zip->new( “whatever”, Zip64 => 1 );

When uncompressing with IO-Uncompress-Unzip, it will automatically detect if the zip file is zip64.

If you intend to manipulate the Zip64 zip files created with IO-Compress-Zip using an external zip/unzip, make sure that it supports Zip64.

In particular, if you are using Info-Zip you need to have zip version 3.x or better to update a Zip64 archive and unzip version 6.x to read a zip64 archive.

Yes. Zip64 allows this. See previous question.

The primary reference for zip files is the appnote document available at http://www.pkware.com/documents/casestudies/APPNOTE.TXT

An alternatively is the Info-Zip appnote. This is available from ftp://ftp.info-zip.org/pub/infozip/doc/

The primary reference for gzip files is RFC 1952 http://www.faqs.org/rfcs/rfc1952.html

The primary site for gzip is http://www.gzip.org.

If the gunzip program encounters a file containing multiple gzip files concatenated together it will automatically uncompress them all. The example below illustrates this behaviour

$ echo abc | gzip -c >x.gz $ echo def | gzip -c >>x.gz $ gunzip -c x.gz abc def

By default IO::Uncompress::Gunzip will not behave like the gunzip program. It will only uncompress the first gzip data stream in the file, as shown below

$ perl -MIO::Uncompress::Gunzip=:all -e gunzip “x.gz” => \*STDOUT abc

To force IO::Uncompress::Gunzip to uncompress all the gzip data streams, include the MultiStream option, as shown below

$ perl -MIO::Uncompress::Gunzip=:all -e gunzip “x.gz” => \*STDOUT, MultiStream => 1 abc def

A bgzip file consists of a series of valid gzip-compliant data streams concatenated together. To read a file created by bgzip with IO::Uncompress::Gunzip use the MultiStream option as shown in the previous section.

See the section titled The BGZF compression format in http://samtools.github.io/hts-specs/SAMv1.pdf for a definition of bgzip.

The primary site for the zlib compression library is http://www.zlib.org.

The primary site for bzip2 is http://www.bzip.org.

If the bunzip2 program encounters a file containing multiple bzip2 files concatenated together it will automatically uncompress them all. The example below illustrates this behaviour

$ echo abc | bzip2 -c >x.bz2 $ echo def | bzip2 -c >>x.bz2 $ bunzip2 -c x.bz2 abc def

By default IO::Uncompress::Bunzip2 will not behave like the bunzip2 program. It will only uncompress the first bunzip2 data stream in the file, as shown below

$ perl -MIO::Uncompress::Bunzip2=:all -e bunzip2 “x.bz2” => \*STDOUT abc

To force IO::Uncompress::Bunzip2 to uncompress all the bzip2 data streams, include the MultiStream option, as shown below

$ perl -MIO::Uncompress::Bunzip2=:all -e bunzip2 “x.bz2” => \*STDOUT, MultiStream => 1 abc def

Pbzip2 (http://compression.ca/pbzip2/) is a parallel implementation of bzip2. The output from pbzip2 consists of a series of concatenated bzip2 data streams.

By default IO::Uncompress::Bzip2 will only uncompress the first bzip2 data stream in a pbzip2 file. To uncompress the complete pbzip2 file you must include the MultiStream option, like this.

bunzip2 $input => \$output, MultiStream => 1 or die “bunzip2 failed: $Bunzip2Error\n”;

Below is a mod_perl Apache compression module, called Apache::GZip, taken from http://perl.apache.org/docs/tutorials/tips/mod_perl_tricks/mod_perl_tricks.html#On_the_Fly_Compression

package Apache::GZip; #File: Apache::GZip.pm use strict vars; use Apache::Constants :common; use Compress::Zlib; use IO::File; use constant GZIP_MAGIC > 0x1f8b; use constant OS_MAGIC => 0x03; sub handler { my $r = shift; my ($fh,$gz); my $file = $r->filename; return DECLINED unless $fh=IO::File->new($file); $r->header_out(Content-Encoding=>gzip); $r->send_http_header; return OK if $r->header_only; tie *STDOUT,Apache::GZip,$r; print($_) while <$fh>; untie *STDOUT; return OK; } sub TIEHANDLE { my($class,$r) = @_; # initialize a deflation stream my $d = deflateInit(-WindowBits=>-MAX_WBITS()) || return undef; # gzip header -- dont ask how I found out $r->print(pack("nccVcc",GZIP_MAGIC,Z_DEFLATED,0,time(),0,OS_MAGIC)); return bless { r => $r, crc => crc32(undef), d => $d, l => 0 },$class; } sub PRINT { my $self = shift; foreach (@_) { # deflate the data my $data = $self->{d}->deflate($_); $self->{r}->print($data); # keep track of its length and crc $self->{l} + length($_); $self->{crc} = crc32($_,$self->{crc}); } } sub DESTROY { my $self = shift; # flush the output buffers my $data = $self->{d}->flush; $self->{r}->print($data); # print the CRC and the total length (uncompressed) $self->{r}->print(pack(“LL”,@{$self}{qw/crc l/})); } 1;

Here’s the Apache configuration entry you’ll need to make use of it. Once set it will result in everything in the /compressed directory will be compressed automagically.

<Location /compressed> SetHandler perl-script PerlHandler Apache::GZip </Location>

Although at first sight there seems to be quite a lot going on in Apache::GZip, you could sum up what the code was doing as follows Ω- read the contents of the file in $r->filename, compress it and write the compressed data to standard output. That’s all.

This code has to jump through a few hoops to achieve this because

1. The gzip support in Compress::Zlib version 1.x can only work with a real filesystem filehandle. The filehandles used by Apache modules are not associated with the filesystem.
2. That means all the gzip support has to be done by hand - in this case by creating a tied filehandle to deal with creating the gzip header and trailer.

IO::Compress::Gzip doesn’t have that filehandle limitation (this was one of the reasons for writing it in the first place). So if IO::Compress::Gzip is used instead of Compress::Zlib the whole tied filehandle code can be removed. Here is the rewritten code.

package Apache::GZip; use strict vars; use Apache::Constants :common; use IO::Compress::Gzip; use IO::File; sub handler { my $r = shift; my ($fh,$gz); my $file = $r->filename; return DECLINED unless $fh=IO::File->new($file); $r->header_out(Content-Encoding=>gzip); $r->send_http_header; return OK if $r->header_only; my $gz = IO::Compress::Gzip->new( -, Minimal => 1 ) or return DECLINED ; print $gz $_ while <$fh>; return OK; }

or even more succinctly, like this, using a one-shot gzip

package Apache::GZip; use strict vars; use Apache::Constants :common; use IO::Compress::Gzip qw(gzip); sub handler { my $r = shift; $r->header_out(Content-Encoding=>gzip); $r->send_http_header; return OK if $r->header_only; gzip $r->filename => -, Minimal => 1 or return DECLINED ; return OK; } 1;

The use of one-shot gzip above just reads from $r->filename and writes the compressed data to standard output.

Note the use of the Minimal option in the code above. When using gzip for Content-Encoding you should always use this option. In the example above it will prevent the filename being included in the gzip header and make the size of the gzip data stream a slight bit smaller.

The Net::FTP module provides two low-level methods called stor and retr that both return filehandles. These filehandles can used with the IO::Compress/Uncompress modules to compress or uncompress files read from or written to an FTP Server on the fly, without having to create a temporary file.

Firstly, here is code that uses retr to uncompressed a file as it is read from the FTP Server.

use Net::FTP; use IO::Uncompress::Gunzip qw(:all); my $ftp = Net::FTP->new( … ) my $retr_fh = $ftp->retr($compressed_filename); gunzip $retr_fh => $outFilename, AutoClose => 1 or die “Cannot uncompress $compressed_file: $GunzipError\n”;

and this to compress a file as it is written to the FTP Server

use Net::FTP; use IO::Compress::Gzip qw(:all); my $stor_fh = $ftp->stor($filename); gzip “filename” => $stor_fh, AutoClose => 1 or die “Cannot compress $filename: $GzipError\n”;

file/buffer. A fairly common use-case is where compressed data is embedded in a larger file/buffer and you want to read both.

As an example consider the structure of a zip file. This is a well-defined file format that mixes both compressed and uncompressed sections of data in a single file.

For the purposes of this discussion you can think of a zip file as sequence of compressed data streams, each of which is prefixed by an uncompressed local header. The local header contains information about the compressed data stream, including the name of the compressed file and, in particular, the length of the compressed data stream.

To illustrate how to use InputLength here is a script that walks a zip file and prints out how many lines are in each compressed file (if you intend write code to walking through a zip file for real see Walking through a zip file in IO::Uncompress::Unzip ). Also, although this example uses the zlib-based compression, the technique can be used by the other IO::Uncompress::* modules.

use strict; use warnings; use IO::File; use IO::Uncompress::RawInflate qw(:all); use constant ZIP_LOCAL_HDR_SIG > 0x04034b50; use constant ZIP_LOCAL_HDR_LENGTH => 30; my $file = $ARGV[0] ; my $fh = IO::File->new( "<$file" ) or die "Cannot open $file: $!\n"; while (1) { my $sig; my $buffer; my $x ; ($x = $fh->read($buffer, ZIP_LOCAL_HDR_LENGTH)) = ZIP_LOCAL_HDR_LENGTH or die “Truncated file: $!\n”; my $signature = unpack (“V”, substr($buffer, 0, 4)); last unless $signature = ZIP_LOCAL_HDR_SIG; # Read Local Header my $gpFlag = unpack ("v", substr($buffer, 6, 2)); my $compressedMethod = unpack ("v", substr($buffer, 8, 2)); my $compressedLength = unpack ("V", substr($buffer, 18, 4)); my $uncompressedLength = unpack ("V", substr($buffer, 22, 4)); my $filename_length = unpack ("v", substr($buffer, 26, 2)); my $extra_length = unpack ("v", substr($buffer, 28, 2)); my $filename ; $fh->read($filename, $filename_length) = $filename_length or die “Truncated file\n”; $fh->read($buffer, $extra_length) = $extra_length or die "Truncated file\n"; if ($compressedMethod ! 8 && $compressedMethod != 0) { warn “Skipping file $filename - not deflated $compressedMethod\n”; $fh->read($buffer, $compressedLength) = $compressedLength or die "Truncated file\n"; next; } if ($compressedMethod = 0 && $gpFlag & 8 = 8) { die "Streamed Stored not supported for $filename\n"; } next if $compressedLength = 0; # Done reading the Local Header my $inf = IO::Uncompress::RawInflate->new( $fh, Transparent => 1, InputLength => $compressedLength ) or die “Cannot uncompress $file [$filename]: $RawInflateError\n” ; my $line_count = 0; while (<$inf>) { ++ $line_count; } print “$filename: $line_count\n”; }

The majority of the code above is concerned with reading the zip local header data. The code that I want to focus on is at the bottom.

while (1) { # read local zip header data # get $filename # get $compressedLength my $inf = IO::Uncompress::RawInflate->new( $fh, Transparent => 1, InputLength => $compressedLength ) or die “Cannot uncompress $file [$filename]: $RawInflateError\n” ; my $line_count = 0; while (<$inf>) { ++ $line_count; } print “$filename: $line_count\n”; }

The call to IO::Uncompress::RawInflate creates a new filehandle $inf that can be used to read from the parent filehandle $fh, uncompressing it as it goes. The use of the InputLength option will guarantee that at most $compressedLength bytes of compressed data will be read from the $fh filehandle (The only exception is for an error case like a truncated file or a corrupt data stream).

This means that once RawInflate is finished $fh will be left at the byte directly after the compressed data stream.

Now consider what the code looks like without InputLength

while (1) { # read local zip header data # get $filename # get $compressedLength # read all the compressed data into $data read($fh, $data, $compressedLength); my $inf = IO::Uncompress::RawInflate->new( \$data, Transparent => 1 ) or die “Cannot uncompress $file [$filename]: $RawInflateError\n” ; my $line_count = 0; while (<$inf>) { ++ $line_count; } print “$filename: $line_count\n”; }

The difference here is the addition of the temporary variable $data. This is used to store a copy of the compressed data while it is being uncompressed.

If you know that $compressedLength isn’t that big then using temporary storage won’t be a problem. But if $compressedLength is very large or you are writing an application that other people will use, and so have no idea how big $compressedLength will be, it could be an issue.

Using InputLength avoids the use of temporary storage and means the application can cope with large compressed data streams.

One final point Ω- obviously InputLength can only be used whenever you know the length of the compressed data beforehand, like here with a zip file.