#!/usr/bin/perl
#-*-cperl-*-

unless (@ARGV) {
  my $self=$0;
  $self=~s(^.*/)();
  print "$self - fix broken Latin-2 encoding in OpenOffice.org\n";
  print "        documents imported from M\$ Office\n\n";
  print "May be distributed under GNU General Public Licence (Version 2)\n";
  print "Copyright (c) 2002 by Petr Pajas\n\n";
  print "Usage: $self <file> [<file> ...]\n\n";
}

my %map = (
         'È' => 'Č',
         '©' => 'Š',
         'è' => 'č',
         '«' => 'Ť',
         'Ì' => 'Ě',
         '®' => 'Ž',
         'ì' => 'ě',
         'Ï' => 'Ď',
         'ï' => 'ď',
         'Ò' => 'Ň',
         'ò' => 'ň',
         '¹' => 'š',
         'Ø' => 'Ř',
         'Ù' => 'Ů',
         'ø' => 'ř',
         'ù' => 'ů',
         '»' => 'ť',
         '¾' => 'ž',
       );

my $re=join('|',keys(%map));

foreach my $f (@ARGV) {
  print "Fixing $f:\n";
  my $tmpzip="/tmp/oo__convert-$$.zip";
  my $tmpdir="/tmp/oo__convert-$$";

  (system("unzip -d $tmpdir $f")==0) || die "$f: unzip failed!";
  chmod 0700, $tmpdir;
  foreach my $file (glob("$tmpdir/*.xml"),glob("$tmpdir/*/*.xml")) {
    print "  fixing XML file $file\n";
    open my $F,$file;

    my $content=join("",<$F>);
    close $F;
    $content=~s/$re/$map{$&}/eg;
    $content=~s/font(-name|-family)?=(["'])([^"']+) CE\d?\1/font$1=$2$3$2/ig;

    open my $F,">$file";
    print $F $content;
    close $F;
  }
  rename $f,"$f~";

  system "cd $tmpdir; zip -r $tmpzip *";
  system "mv $tmpzip $f";
  system "rm -rf $tmpdir";
}