#!/usr/bin/perl
#
# EE00-E650 = 1968
# E650 E6FF EE00
if ($#ARGV != 2)
{
  print STDERR "usage: shiftutf8 rangefrom rangeto new_from < filein > fileout$/";
  exit (1);
}

$range_from = hex (shift);
$range_to = hex(shift);
$new_from = hex(shift);

#print $range_from ."\n";
#exit (0);

while (<>)
{
  # read chars 
  s/([\xc0-\xdf])([\x80-\xbf])/&utf8dec($1,$2)/ge;
  s/([\xe0-\xef])([\x80-\xbf])([\x80-\xbf])/&utf8dec($1,$2,$3)/ge;
  # now sgml
  # add number
  while (s/&#(\d+)\;/&sgmladd($1)/ge) {}
  while (s/&##(\d+)\;/&utf8enc($1)/ge) {}
  print;
}

exit (0);

sub sgmladd {
  if ($_[0] >= $range_from && $_[0] <= $range_to)
  {
    return sprintf ("&##%u;", $_[0] + $new_from-$range_from);
  }
  return sprintf ("&##%u;", $_[0]);
}

sub utf8enc {
  $str = $_[0];
  if ($_[0] >= 0x800) {
    $str  = chr (0xe0 | ($_[0] >> 12));
    $str .= chr (0x80 | (($_[0] >> 6) & 0x3f));
    $str .= chr (0x80 | ($_[0] & 0x3f));
  } elsif ($_[0] >= 0x80) {
    $str = chr (0xc0 | ($_[0] >> 6));
    $str .= chr (0x80 | ($_[0] & 0x3f));
  } else {
    $str .= chr ($str);
  }
  $str;
}

#
# Works for UCS2
#
sub utf8dec {
  $ret = 0x1f & unpack("C", $_[0]);
  for ($i=1; $i<=$#_; $i++) {
    $ret = ($ret << 6) + int (0x3f & unpack ("C", $_[$i]));
  }
  sprintf ("&#%u;", $ret);
}