#!/usr/bin/perl 

use Getopt::Long qw(:config no_auto_abbrev bundling);
use Pod::Usage;
my($l,$of,$dpi,$out,$nompsep,$r,$rr,$rrr,$b);
my $debug;

GetOptions(
  "help|h|?" => \$help,
  "man|m" => \$man,
  "rrr" => \$rrr,
  "rr" => \$rr,
  "r" => \$r,
  "b" => \$b,
  "n" => \$nompsep,
  "output|o=s" => \$out,
  "l=s" => \$l,
  "d=s" => \$dpi,
  "debug" => \$debug,
  "of|f=s" => \$of,
  ) or die "Specify the --help (or -?) option for usage information.\n";

pod2usage(2) if $help;
pod2usage(-exitstatus => 0, -verbose => 2) if $man;

$l   ||= "por";
$of  ||= "smarttext";
$dpi ||= "300";
$r = "-rotate 90"  if $r;
$r = "-rotate 180" if $rr;
$r = "-rotate 270" if $rrr;

print STDERR "Lingua: $l\n" if $debug;
print STDERR "Output format: $of\n" if $debug;
print STDERR "dpi: $dpi\n" if $debug;

use strict;
my $f=shift or pod2usage(2);

my $aux = $f;
my $out = $f."_ocr.txt";
my $pat = $f;

if ($f =~ /(.*)\.(pdf)$/i) {  
  $aux = "_$$..bmp" ;
  $out = $1."_ocr.txt";
  print STDERR "*...$f --> bmp\n";
  syst("pdftoppm -mono -r $dpi '$f' $aux");
  $pat = "_$$.";
}
elsif ($f =~ /(.*)\.(ps)$/i) {  
  $aux = "_$$..bmp" ;
  $out = $1."_ocr.txt";
  print STDERR "*...$f --> bmp\n";
  syst("convert $r -compress none -density $dpi -monochrome '$f' BMP3:$aux");
  $pat = "_$$.";
}
elsif ($f =~ /(.*)\.(tiff?|png|gif|jpg|pbm|ppm)$/i) {
  $aux = "_$$..bmp" ;
  $out = $1."_ocr.txt";
  print STDERR "...$f --> bmp\n";
  syst("convert $r -compress none -monochrome '$f' BMP3:$aux");
  $pat = "_$$.";
}
elsif ($f =~ /(.*)\.([^.]+)$/) {
  $aux = "_$$..bmp" ;
  $out = $1."_ocr.txt";
  print STDERR "...$f --> bmp\n";
  syst("convert $r -compress none -monochrome '$f' BMP3:$aux");
  $pat = "_$$.";
}

my $n = 1;
print STDERR "...$f --> ocr\n";
for( <$pat*.bmp>,<$pat*.pbm> ){
  print STDERR "$_\n" if $debug;
  my $k = sprintf("%03d",$n);
  syst("cuneiform -l $l -f $of -o '$f.txt.$k' '$_' ");
  $n ++;
  unlink ($_)       unless $debug;
}
undef $/;
open(O,">$out") or die ("cant create $out\n");
for (<$f.txt.*>){
  open(F,$_) or die ("cant read $_\n");
  print O <F>;
  print O "\n\cL\n" unless $nompsep;
  close F;
  unlink $_         unless $debug;
}
close O;

sub syst {
  my $cmd = shift;
  print STDERR "$cmd\n" if $debug ; 
  if(system ($cmd) == 0){ 
         return 1}
  else { warn "** ERROR ************ system $cmd failed: $!$?\n";
         return 0 }
}

__END__

=head1 NAME

wocr - Wrapper for Cuneiform OCR

=head1 SYNOPSIS

 wocr [-options] imagefile
    -l eng          # language = English (def:por)
    -f text         # output format (def:smarttext)
    -d 400          # dpi used in translation PDF ou PS to BMP (def: 300)
    -m              # manpage of wocr
    -h              # usage
    -n              # no page separator "CTR F"
    --debug         # dont remove temp. files
    --r             # rotate pages 90
    --rr            # rotate pages 180
    --rrr           # rotate pages 270
    -b
 

=head1 DESCRIPTION

Convert files to BMP V3, run Cuneiform OCR on all bmp files, concatenate outputs.

Known extensions and formats:

 pdf ps png gif jpg tif tiff pbm ppm

=head2 OPTION

 -l eng          # language = English (def:por)
 -f text         # output format (def:smarttext)
 -d 400          # dpi used in translation PDF ou PS to BMP (def: 300)
 -m              # manpage
 -h              # usage
 -n              # no page separator "CTR F"
 --debug         # dont remove temp. files
 --r             # rotate pages 90
 --rr            # rotate pages 180
 --rrr           # rotate pages 270
 -b
 
=head2 Dependencies

Image Magick  (to convert to BMP)

=head1 AUTHOR

J.Joao Almeida, jj@di.uminho.pt

=head1 SEE ALSO

perl, cuneiform, Image Magick

=cut