#!/usr/bin/env perl
use 5.014;
use warnings;
use utf8;
use open IO => ':encoding(utf-8)', ':std';
use Data::Dump 'pp';

our $VERSION = '1.01';

# translation table for deprecated code points
my %replace = (
	0xF015 => '⎇',  # alt
	0xF016 => '⇧',  # shift
	0xF017 => '⎈',  # control
);

# expect input data source at command line
@ARGV or die "Specify input source file or - for STDIN\n";

# convert each character line to perl code
# (assume no backslashes or curlies, so we can just q{} w/o escaping)
say "# automatically generated by $0";
say 'use utf8;';
say '+{';
while ($_ = readline) {
	my ($chrhex, $mnems, $sample, $name) = m{\A([0-9A-F ]{5}) (.{11}) (.)\h(.*)}i
		or warn("syntax error on line $.: $_"), next;
	$chrhex =~ s/ $//;
	my $chrnum = hex $chrhex;
	my $chr = chr $chrnum;
	$chr eq $sample
		or warn("character mismatch on line $.: $_"), next;
	$chr = $replace{$chrnum} or next if defined $replace{$chrnum};
	my $chrstr = pp($replace{$chrnum} // $chr);
	for my $mnem (split / /, $mnems) {
#		next if length $mnem != 2;
		say "q{$mnem} => $chrstr, # $name";
	}
}
say '}';

__END__

=head1 NAME

mkdigraphs-plan9 - Output digraph data from Plan9 keyboard data

=head1 SYNOPSIS

Extract digraphs from text specifications as a perl hash:

    mkdigraphs-plan9 keyboard >digraphs-plan9.inc.pl

Input can be the literal RFC (or similar) document:

    curl https://9fans.github.io/usr/local/plan9/lib/keyboard | mkdigraphs-plan9 -

Test by printing the character for DO (should be a dollar sign):

    perl -e'$di = do "digraphs-plan9.inc.pl"; print chr $di->{DO}'

=head1 DESCRIPTION

Parses the official RFC-1345 document, searching the
'character mnemonic table' for all digraph definitions.
If successful, Perl code is output resulting in a hash
with Unicode code points keyed by digraph.
Obsolete values (references to private use area)
are converted to modern alternatives.
Any errors and warnings are given at STDERR.

=head1 AUTHOR

Mischa POSLAWSKY <perl@shiar.org>

=head1 LICENSE

Licensed under the GNU Affero General Public License version 3.