to-UTF-16.pl

   1 #! /usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5
   6 my($in_name, $out_name) = @ARGV;
   7
   8 open my $in_h, "<", $in_name or die "$in_name: $!";
   9 binmode $in_h or die "$in_name: $!";
  10 my $head;
  11 defined(sysread $in_h, $head, 4) or die "$in_name: $!";
  12
  13 my($skip, $charset);
  14 if ($head =~ /^\x00\x00\xFF\xFE/) {
  15         print "$in_name: Found big-endian UTF-32 BOM\n";
  16         ($skip, $charset) = (0, "UTF-32");
  17 } elsif ($head =~ /^\xFE\xFF\x00\x00/) {
  18         print "$in_name: Found little-endian UTF-32 BOM\n";
  19         ($skip, $charset) = (0, "UTF-32");
  20 } elsif ($head =~ /^\xFE\xFF/) {
  21         print "$in_name: Found big-endian UTF-16 BOM\n";
  22         ($skip, $charset) = (0, "UTF-16");
  23 } elsif ($head =~ /^\xFF\xFE/) {
  24         print "$in_name: Found little-endian UTF-16 BOM\n";
  25         ($skip, $charset) = (0, "UTF-16");
  26 } elsif ($head =~ /^\xEF\xBB\xBF/) {
  27         print "$in_name: Found UTF-8 BOM\n";
  28         ($skip, $charset) = (3, "UTF-8");
  29 } else {
  30         print "$in_name: No BOM found -- assuming UTF-8\n";
  31         ($skip, $charset) = (0, "UTF-8");
  32 }
  33
  34 sysseek($in_h, $skip, 0) or die "$in_name: $!";
  35
  36 open STDOUT, ">", $out_name or die "$out_name: $!";
  37 open STDIN, "<&", $in_h or die "$in_name: $!";
  38 exec "iconv", "-f", $charset, "-t", "UTF-16"
  39         or die "exec iconv: $!";