get .js & .css sources from html

getHtmlSrc.pl

#!/usr/bin/env perl
# downloads src .js and .css files from html; use if you are behind proxy and cant source external content
# usage
# ./getHtmlSrc.pl /path/to/some.html

# v0.2 - prints local links after src name
# v0.3 - archive (out.html.src.gz) then remove all .js & .css
# v0.4 - outputs out.html with swapped src files for local ones 

use strict;
use warnings;
use IO::All;

my (@lines, $file, $content, $name, $url, $hot, @files, $nurl);
my $locLibDir = 'llib';
my $fileOut = 'out.html';

if ($ARGV[0]) {
    $file = $ARGV[0];
    `rm $fileOut && touch $fileOut`;
} else {
    print "give me path to file\n";
    exit;
}

io("$file") > $content;
@lines = split(/\n/, $content);

foreach (@lines) {
    if (/(\<script\ src\=\"(http|\/\/)|(\<link\ rel.*\/\/.*))/) {
        $name = $url = $nurl =  $_;

        $name =~ s/.*\/(.*?\.(js|css)).*/$1/;
        push @files, $name;
        $url =~ s/(.*\/\/)(.*?)(\".*)/$2/;
        
        $nurl =~ s/(.*)(\/\/.*?)(\".*)/$1$name$3/;
        $nurl =~ s/(.*)(http\:)(.*)/$1$3/;

        print "$name -> ";
        $hot = "curl\ \-LO\ http\:\/\/$url 1>/dev/null 2>/dev/null";
        `$hot`;
        print "$nurl\n";
        "$nurl\n" >> io($fileOut);
    } else {
        "$_\n" >> io($fileOut);
    }
}

sub archive {
    my $toTar = join ' ', @files;
    `tar -zcvf "$fileOut".'src.gz' $toTar`;
    `rm $toTar`;
}

archive;