#!/usr/bin/perl

# Just makes a list of all the filenames included under each .tar.gz file.
# It's probably useful for grepping.
#
# Designed to work with CPAN::Mini.
#
# Beware -- the output is pretty large. If you run:
#       ./dump_filenames.pl | gzip > filename_dump.txt.gz
# you will still get a 11mb file.

    use strict;
    use warnings;

    use Archive::Tar        ();     # Perl core
    use CGI::Tiny           ();
    use File::Basename      ();     # Perl core
    use Pod::Parser         ();     # Perl core
    use Text::LineNumber    ();

    use Data::Dumper;               # Perl core


    # the location of your CPAN::Mini mirror
    my $mirror_location = '/home/newcum/minicpan/mirror/';


-d $mirror_location or die "\$mirror_location is set incorrectly ($mirror_location)\n";

$Archive::Tar::WARN = 0;        # quiet warning messages

my @archives;
chdir($mirror_location);
open PIN, "-|", "find", "authors/", "-type", "f", "-o", "-type", "l"
        or die $!;
while (<PIN>) {
    chomp;
    push(@archives, $_)     if (/\.tar\.gz$/i);
}

# sort by distribution name -- without this, the results end up being sorted by author name
        sub distribution_name {  local $_ = shift;  s#^.*/##;  $_  }
@archives = sort {lc(distribution_name($a)) cmp lc(distribution_name($b))} @archives;

foreach my $archive (@archives) {
    process_archive($archive);
}



sub process_archive {
    my $tarball = shift;

    # tarballs that we get hung up on for one reason or another
    return if ($tarball =~ m#/Lingua-StanfordCoreNLP-#);

    my $tar = Archive::Tar->new($tarball);

    foreach my $filename ($tar->list_files()) {
        #next unless (filename_filter($filename));
        (my $filename_sans_package = $filename) =~ s#^[^/]*/##s;

        print "$tarball/$filename_sans_package\n";
    }
}