#!/usr/bin/perl -w
use strict;
use warnings;
use utf8;
use open qw(:std :utf8);


# Filter_3_01_replace_unknown_characters.pl
# 2010-04-14
# Wolfgang Schmidle

# what does it do?
# input
# output

my $filtername = "Filter_3_01_replace_unknown_characters.pl";


# text input

my @text;
while(<>) { push @text, $_; }


# read default parameters

my $filter = "mpdl-workflow-settings.pl";
my $result = `perl $filter`;

my $aux = ($result =~ m! aux=([^ ]+) ! ? $1 : "" );

while ($result =~ m! unknown=([^ ]+)!g) { push @ARGV, "$aux/$1"; }


# read the "parameters" block

my %unknown = ();

my $inText = 0;
my $inParameters = 0;

foreach (@text) {
	last if m!<pb[ ^]!;	
	if (m!parameters:!) { $inParameters = 1; }
	if ($inParameters) {
		last if m!^ *$!;
		while (m!unknown *= *([^ \n]+)!) { push @ARGV, "$aux/$1"; }
	}
}


# read helper files

while (<>) {
	chomp;
	s!#.*$!!; # remove comments
	s! +$!!;
	next if m!^$!;

	if (m!<(\d\d\d)> +see +<(\d\d\d)>(.*)!) {
		unless (exists $unknown{$2}) { die "invalid reference:\n$_\n"; }
		$unknown{$1} = "$unknown{$2} $3"; 
		next;
	}
	if (m!<(\d\d\d)> *= *([^ ]+)!) { $unknown{$1} = $2; }
}


# do not replace codes whose meaning is unclear

foreach my $code (keys %unknown) {
	if ($unknown{$code} =~ /\?/) {
		$unknown{$code} = "$code"; # in other words, do not replace this code
	}
}


# evaluate and remove the "unknown" block

my $inUnknown = 0;

foreach (@text) {
	last if m!<pb[ ^]!;	
	if (m!unknown:!) { 
		$inUnknown = 1; 
		$_ = "";
		next;
	}
	if ($inUnknown) {
		if (m!^ *$!) {
			$_ = "";
			last;
		}
		unless (m!<(\d\d\d)> *= *([^ ]+)!) { 
			die "$filtername: unrecognised unknown character code:\n$_\n"; 
		}
		$unknown{$1} = $2;
		$_ = "";
	}
}


# go through the text

foreach (@text) {
	s!<(\d\d\d)>!$unknown{$1}!g;
}


# text output

print @text;


# TO DO: