#!/usr/bin/perl
# Complete check of a stateless encoding.
# Usage: check-stateless TOP SRCDIR CHARSET

use warnings;
use strict;

if (@ARGV < 3) {
	print "Usage: check-stateless <top> <srcdir> <charset\n";
	exit;
}

my $top     = shift @ARGV;
my $srcdir  = shift @ARGV;
my $charset = shift @ARGV;

# charset, modified for use in filenames.
my $charsetf = $charset;
$charsetf =~ s/:/-/g;

# iconv in one direction.
command("$srcdir/table-from $charset > $srcdir/tmp-$charsetf.TXT");

# iconv in the other direction.
command("$srcdir/table-to $charset | sort > $srcdir/tmp-$charsetf.INVERSE.TXT");

# Check 1: charmap and iconv forward should be identical.
command("cmp $srcdir/$charsetf.TXT $srcdir/tmp-$charsetf.TXT 2> /dev/null");

# Check 2: the difference between the charmap and iconv backward.
command("sed -e '/	.* 0x/d' < $srcdir/$charsetf.TXT > $srcdir/tmp-noprecomposed-$charsetf.TXT");

if (-f "$srcdir/$charsetf.IRREVERSIBLE.TXT") {
	command("cat $srcdir/tmp-noprecomposed-$charsetf.TXT $srcdir/$charsetf.IRREVERSIBLE.TXT | sort | uniq -u > $srcdir/tmp-orig-$charsetf.INVERSE.TXT");
} else {
	command("cp $srcdir/tmp-noprecomposed-$charsetf.TXT $srcdir/tmp-orig-$charsetf.INVERSE.TXT");
}
command("cmp $srcdir/tmp-orig-$charsetf.INVERSE.TXT $srcdir/tmp-$charsetf.INVERSE.TXT 2> /dev/null");

command("rm -f $srcdir/tmp-$charsetf.TXT $srcdir/tmp-$charsetf.INVERSE.TXT $srcdir/tmp-noprecomposed-$charsetf.TXT $srcdir/tmp-orig-$charsetf.INVERSE.TXT");

sub command {
	my $cmd = shift;
	print "> $cmd\n";
	my @output = `$cmd 2>&1`;
	foreach my $line (@output) {
		print "| $line";
	}
	my $status = $? / 256;
	die "$cmd:\nexit status $status\n" if $status;
	return @output;
}

# For a new encoding:
# You can create the "$charsetf".TXT like this:
#   ./table-from "$charset" > "$charsetf".TXT
# You can create the "$charsetf".IRREVERSIBLE.TXT like this:
#   ./table-to "$charset" | sort > "$charsetf".INVERSE.TXT
#   diff "$charsetf".TXT "$charsetf".INVERSE.TXT | grep '^[<>]' | sed -e 's,^. ,,' > "$charsetf".IRREVERSIBLE.TXT