#!/usr/bin/perl
#
# Converts Roman numerals to arabic numerals
#
# Supports iv, ix, xl, xc, cd, cm convention
# Supports iiii, xxxx, convention
# Does NOT support ic, im, xm, convention
#
# written 27/6/00

sub arabise
{
    $roman = $_[0];

# delete any carriage returns
    if ($roman =~ /\n/)
    {
	$roman =~ s/[\n]//g;
    }
# lower case letters
    if ($roman =~ /[A-Z]/)
    {
	$roman =~ tr/[A-Z]/[a-z]/;
    }
# report dodgy roman numerals    
    if ($roman =~ /i[ldcm]|v[vxldcm]|x[dm]|l[lcdm]|d[dm]/ ||
	$roman =~ /[iv]+ix|[iv]+iv|ix[xivcdlm]|xl[xcdml]|xc[cxmd]|cm[mdc]/ ||
	$roman =~ /iiiii|xxxxx|ccccc/)
    {
	print STDERR "\n\tWarning \"$roman\" not a valid Roman numeral.\n";
	print STDERR "\tThis will probably cause an erroneous output.\n\n";
    }
    
    $units = 0;
    $tens = 0;
    $hundreds = 0;
    $thousands = 0;
    $i_counter = 0;
    $x_counter = 0;
    $c_counter = 0;
    $m_counter = 0;
    
    if ($roman =~ /[ixv]/)
    {
	$units = $roman;
	$units =~ s/xl//;
	$units =~ s/xc//;
	$units =~ s/[mdcl]//;
	$units =~ s/^[x]+//;
	
# units
	if ($units =~ /iv$/)
	{
	    $units = 4;
	}
	elsif ($units =~ /ix$/)
	{
	    $units = 9;
	}
	elsif ($units =~ /v$/)
	{
	    $units = 5;
	}
	elsif ($units =~ /[i]$/)
	{
	    $temp_units = $units;
	    while ($temp_units =~ s/i//)
	    {
		$i_counter++;
	    }
	    if ($units =~ /v/)
	    {
		$units = $i_counter + 5;
	    }
	    else
	    {
		$units = $i_counter;
	    }
	    $i_counter = 0;
	}
	else
	{
	    $units = 0;
	}
    }
    
# tens
    if ($roman =~ /[xl]/)
    {
	$tens = $roman;
	$tens =~ s/[md]+//;
	$tens =~ s/iv$//;
	$tens =~ s/ix$//;
	$tens =~ s/v[i]+$//;
	$tens =~ s/[i]+$//;
	$tens =~ s/^[c]+//;
	
	$temp_tens = $tens;
	while ($temp_tens =~ s/x//)
	{
	    $x_counter++;
	}
	if ($tens =~ /l/)
	{
	    if ($tens =~ /xl/)
	    {
		$tens = 4;
	    }
	    elsif ($tens =~ /lx/)
	    {
		$tens = 5 + $x_counter;
	    }
	    else
	    {
		$tens = 5;
	    }
	}
	elsif ($tens =~ /xc/)
	{
	    $tens = 9;
	}
	else
	{
	    $tens = $x_counter;
	}
    }
    
# hundreds
    if ($roman =~ /[cd]/)
    {
	$hundreds = $roman;
	
	if ($hundreds =~ /^xc/)
	{
	    $hundreds = 0;
	}
	elsif ($hundreds =~ /c/)
	{
	    $hundreds =~ s/xc//;
	    $hundreds =~ s/[xvil]+//;
	    $temp_hundreds = $hundreds;
	    while ($temp_hundreds =~ s/c//)
	    {
		$c_counter++;
	    }
	    if ($hundreds =~ /d/)
	    {
		if ($hundreds =~ /cd/)
		{
		    $hundreds = 4;
		}
		else
		{
		    $hundreds = 5 + $c_counter;
		}
	    }
	    elsif ($hundreds =~ /cm/)
	    {
		$hundreds = 9;
	    }
	    else
	    {
		$hundreds = $c_counter;
	    }
	}
	else
	{
	    $hundreds = 5;
	}
    }
# thousands
    if ($roman =~ /m/)
    {
	$thousands = $roman;
	if ($thousands =~ /^cm/)
	{
	    $thousands = 0;
	}
	elsif ($thousands =~ /m/)
	{
	    $thousands =~ s/cm//;
	    $temp_thousands = $thousands;
	    while ($temp_thousands =~ s/m//)
	    {
		$m_counter++;
	    }
	    $thousands = $m_counter;
	}
    }
    $arabic_number = $thousands . $hundreds . $tens . $units;
    return $arabic_number;
}
1; # returns true (a perl requirement)