charconv

   1 #!/usr/bin/env perl
   2
   3 #=======================================================================
   4 # charconv
   5 # File ID: 6e32b188-5d37-11df-98d6-90e6ba3022ac
   6 # Konverterer fra cp865 til Unicode. Alternativ konverteringsfil
   7 # spesifiseres med «-i»-parameteret, ellers brukes default filnavn. -p
   8 # skriver ut en enkel HTML-header og -slutt.
   9 # Lisens: GNU GPL v2 eller seinere.
  10 #=======================================================================
  11
  12 use strict;
  13 use warnings;
  14 require 'getopts.pl';
  15
  16 my $map_file = "$ENV{HOME}/src/svn/div/trunk/ibm2ent/cp865.txt";
  17 my %Table;
  18
  19 $| = 1;
  20
  21 Getopts('hi:pP');
  22 defined($main::opt_i) && ($map_file = $main::opt_i);
  23
  24 for (my $a = 0; $a < 256; $a++) {
  25         $Table{$a} = widechar($a);
  26 }
  27
  28 open(MapFP, "<$map_file") || die("$map_file: Can’t open map file: $!");
  29
  30 while (<MapFP>) {
  31         (/^#/ || /^\s*$/) && next;
  32         if (/^0x(..)\s+0x(....)/) {
  33                 my ($Cp, $Uni) = (hex($1), hex($2));
  34                 if ($Cp != $Uni) {
  35                         $Table{$Cp} = widechar($Uni);
  36                 }
  37         } else {
  38                 print STDERR "Line $.: Invalid line: \"$_\"\n";
  39         }
  40 }
  41
  42 $main::opt_P && print("<html><head></head><body><pre>\n") ||
  43 $main::opt_p && print("<pre>\n");
  44
  45 while(<>) {
  46         s/(.)/$Table{ord($1)}/g;
  47         print;
  48 }
  49
  50 $main::opt_P && print("</pre></body></html>\n") ||
  51 $main::opt_p && print("</pre>\n");
  52
  53 # Henta fra h2u,v 1.5 (r386) 2002/11/20 00:09:40
  54 sub widechar {
  55         my $Val = shift;
  56         my $allow_invalid = 0;
  57         if ($Val < 0x80) {
  58                 return sprintf("%c", $Val);
  59         } elsif ($Val < 0x800) {
  60                 return sprintf("%c%c", 0xC0 | ($Val >> 6),
  61                                        0x80 | ($Val & 0x3F));
  62         } elsif ($Val < 0x10000) {
  63                 unless ($allow_invalid) {
  64                         if  (($Val >= 0xD800 && $Val <= 0xDFFF) || ($Val eq 0xFFFE) || ($Val eq 0xFFFF)) {
  65                                 $Val = 0xFFFD;
  66                         }
  67                 }
  68                 return sprintf("%c%c%c", 0xE0 |  ($Val >> 12),
  69                                          0x80 | (($Val >>  6) & 0x3F),
  70                                          0x80 |  ($Val        & 0x3F));
  71         } elsif ($Val < 0x200000) {
  72                 return sprintf("%c%c%c%c", 0xF0 |  ($Val >> 18),
  73                                            0x80 | (($Val >> 12) & 0x3F),
  74                                            0x80 | (($Val >>  6) & 0x3F),
  75                                            0x80 |  ($Val        & 0x3F));
  76         } elsif ($Val < 0x4000000) {
  77                 return sprintf("%c%c%c%c%c", 0xF8 |  ($Val >> 24),
  78                                              0x80 | (($Val >> 18) & 0x3F),
  79                                              0x80 | (($Val >> 12) & 0x3F),
  80                                              0x80 | (($Val >>  6) & 0x3F),
  81                                              0x80 | ( $Val        & 0x3F));
  82         } elsif ($Val < 0x80000000) {
  83                 return sprintf("%c%c%c%c%c%c", 0xFC |  ($Val >> 30),
  84                                                0x80 | (($Val >> 24) & 0x3F),
  85                                                0x80 | (($Val >> 18) & 0x3F),
  86                                                0x80 | (($Val >> 12) & 0x3F),
  87                                                0x80 | (($Val >>  6) & 0x3F),
  88                                                0x80 | ( $Val        & 0x3F));
  89         } else {
  90                 return widechar(0xFFFD);
  91         }
  92 } # widechar()
  93
  94 __END__
  95
  96 #### End of file charconv ####