Get the IANA Language Tag registry, parse it, and write it to a tab-delimited file

iana-lang-tags.pl

#!perl -w

# Parse out "http://www.iana.org/assignments/language-subtag-registry/language-subtag-registry
# And write it to a tab-delimited file.
# Format described at http://tools.ietf.org/html/bcp47#section-3.1.2

# TODO: use some Excel writer to split into sheets by Type

use LWP::Simple;
$_ = get("http://www.iana.org/assignments/language-subtag-registry/language-subtag-registry") or die;
s{\n }{}g; # continuation lines in Comments

open STDOUT,">iana-lang-tags.txt" or die "can't open STDOUT: $!\n";
binmode STDOUT, ':encoding(UTF-8)' or die "can't set binmode UTF-8: $!\n";
  # http://perldoc.perl.org/perlunifaq.html#Is-there-a-way-to-automatically-decode-or-encode%3f
my @cols = qw(Type Scope Prefix Tag Subtag Suppress-Script Description Macrolanguage Added Deprecated Preferred-Value Comments);
print join("\t",@cols), "\n";

foreach (split/\n%%\n/) { # %% separated block
  next if /File-Date:/; # first block is date stamp, not data
  my %hash;
  foreach (split/\n/) {
    my ($key,$val) = split(/: /,$_,2);
    if ($hash{$key}) {$hash{$key} .= ", $val"} # 'Description', 'Comments', and 'Prefix' are multivalued
    else {$hash{$key} = $val};
  };
  foreach (@cols) {
    print $hash{$_} || "", "\t";
  };
  print "\n";
}