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


# Filter_2_04_check_unknown_characters.pl
# 2010-04-14
# Wolfgang Schmidle

# Mit Hilfe dieses Skriptes wird der unknown-Block angelegt. 
# input
# output


# 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; }
}


# go through the text

my %number;
my %line;
my $pb = 0;
my $i = 0; 

foreach (@text) {
	$i++;
	# if (m!<pb [^>]*n="(\d+)"!) { $pb = $1; }
	if (m!<pb[ >]!) { $pb++; }

	while (s!<(\d\d\d)>!<¥$1>!) {
		unless ($number{$1}++) {				
			$line{$1} = "<$1>: first occurrence in line $i (page $pb):\n$text[$i-2]$_$text[$i]";
			if ($unknown{$1}) {
				$line{$1} .= "suggested replacement: $unknown{$1}\n";
			}
			$line{$1} =~ s!¥!!g;
		}
	}
}


my $username = "kthoden";

my $output = "";

# print first occurrences

foreach my $code (sort keys %line) {
		$output .= "\n$line{$code}\n";
}


# print a suggestion for the "unknown" block

$output .= "unknown:\n";
foreach my $code (sort keys %line) {
		my $times = ( $number{$code} == 1 ? "time" : "times" );
		$output .= "<$code> = $unknown{$code} ($number{$code} $times)\n";
}

$output .= "\n";

open (OUTPUT, ">/Users/$username/Desktop/script-output.txt") or die "Can't write the raw text!\n";
print OUTPUT $output;
close(OUTPUT);

die "That's all.\n";