Perl6: large gzipped files read line by line

Update Now tested, which revealed an error, now fixed.

Solution #2

use Compress::Zlib;

my $file   = "data.txt.gz" ;
my $handle = try open $file or die "Error reading $file: $!" ;
my $zwrap  = zwrap($handle, :gzip) ;

for $zwrap.lines {
    .print
}

CATCH { default { die "Error reading $file: $_" } }

$handle.close ;

I've tested this with a small gzipped text file.

I don't know much about gzip etc. but figured this out based on:

• Knowing P6;

• Reading Compress::Zlib's README and choosing the zwrap routine;

• Looking at the module's source code, in particular the signature of the zwrap routine our sub zwrap ($thing, :$zlib, :$deflate, :$gzip);

• And trial and error, mainly to guess that I needed to pass the :gzip adverb.

---

Please comment on whether my code works for you. I'm guessing the main thing is whether it's fast enough for the large files you have.

A failed attempt at solution #5

With solution #2 working I would have expected to be able to write just:

use Compress::Zlib ;
.print for "data.txt.gz".&zwrap(:gzip).lines ;

But that fails with:

No such method 'eof' for invocant of type 'IO::Path'

This is presumably because this module was written before the reorganization of the IO classes.

That led me to @MattOates' IO::Handle like object with .lines ? issue. I note no response and I saw no related repo at https://github.com/MattOates?tab=repositories.

I am focusing on the Inline::Perl5 solution that you tried.

For the call to $gz.gzreadline($_): it seems like gzreadline tries to return the line read from the zip file by modifying its input argument $_ (treated as an output argument, but it is not a true Perl 5 reference variable^[1]), but the modified value is not returned to the Perl 6 script.

Here is a possoble workaround: Create a wrapper module in the curent directory, e.g. ./MyZlibWrapper.pm:

package MyZlibWrapper;
use strict;
use warnings;
use Compress::Zlib ();
use Exporter qw(import);

our @EXPORT = qw(gzopen);
our $VERSION = 0.01;

sub gzopen {
    my ( $fn, $mode ) = @_;
    my $gz = Compress::Zlib::gzopen( $fn, $mode );
    my $self = {gz => $gz}; 
    return bless $self, __PACKAGE__;
}

sub gzreadline {
    my ( $self ) = @_;
    my $line = "";
    my $res = $self->{gz}->gzreadline($line);
    return [$res, $line];
}

sub gzclose {
    my ( $self ) = @_;
    $self->{gz}->gzclose();
}    

1;

Then use Inline::Perl5 on this wrapper module instead of Compress::Zlib. For example ./p.p6:

use v6;
use lib:from<Perl5> '.';
use MyZlibWrapper:from<Perl5>;
my $file = 'data.txt.gz';
my $mode = 'rb';
my $gz = gzopen($file, $mode);
loop {
    my ($res, $line) = $gz.gzreadline();
    last if $res == 0;
    print $line;
}
$gz.gzclose();

_{[1] In Perl 5 you can modify an input argument that is not a reference, and the change will be reflected in the caller. This is done by modifying entries in the special @_ array variable. For example: sub quote { $_[0] = "'$_[0]'" } $str = "Hello"; quote($str) will quote $str even if $str is not passed by reference.}