#!/usr/bin/perl

use strict;
use warnings;
use File::stat;
use File::Basename;
use File::Copy;
use File::Path "make_path";
use DateTime;
use POSIX ();
use POSIX qw(strftime);

my $lang = "bel";

if ( $#ARGV < 0 ) {
  print "give a DJVU file as 1st argument \n" ;
  exit 0;
}

my $inputdjvu = $ARGV[0];
my $basedjvu = basename($inputdjvu, '.djvu');
my $tmp = "/srv/ftp/tmp/$basedjvu";
my $err;
if (-d $tmp) {
	print "$tmp already exists.\n"
} else {
	make_path($tmp, {error  => $err}) or die "Can not create output dir $tmp\n";
}
my $outputprefix = "$tmp/$basedjvu";
my $outputdjvu = $outputprefix . '.djvu';
my $logfile = $outputprefix . '.log';
`touch $logfile`;
copy($inputdjvu, $outputdjvu) or die "Can not copy $inputdjvu to $outputdjvu";

my $nbpages = `djvused "$inputdjvu" -e 'n'`;
chomp($nbpages);

my $fileprefix;
my $outtiff;
my $djvused;
my $pdfs = "";
my $elapsedtime = 0;
my $estimatedduration = 0;
my $i;
for ($i=1 ; $i <= $nbpages ; $i++) {
  my $start = Time::HiRes::gettimeofday();
  print "Page $i/$nbpages: ";
  my $zeropage = sprintf("%04d", $i);
  $fileprefix = $outputprefix . '.' . $zeropage;
  my $splitpagepnm = $fileprefix . '.pnm';

  # page extraction as an image
  `ddjvu -format=pnm -page="$i" "$inputdjvu" $splitpagepnm >> $logfile 2>&1` ;

  my $threshed = $fileprefix . '.thr.pnm';
  `localthresh -b 35 -m 3 -n yes $splitpagepnm $threshed >> $logfile 2>&1`;
  #  unlink($splitpagepnm) or warn "Cannot unlink $splitpagepnm: $!";

  my $outpnm = $fileprefix . '.out.pnm';
  `unpaper -v --layout none $threshed $outpnm >> $logfile 2>&1`;
  # unlink($threshed) or warn "Cannot unlink $threshed: $!";

  my $onepagedjvu = $fileprefix . '.djvu';
  `cjb2 -clean $outpnm $onepagedjvu >> $logfile 2>&1`;

  $outtiff = $fileprefix . '.out.tiff';
  `pnmtotiff $outpnm > $outtiff 2>> $logfile`;
  #  unlink($outpnm) or warn "Cannot unlink $outpnm: $!";

  my $hocr = $fileprefix . '.hocr.html';
  my $pdffile = $fileprefix . '.pdf';
  `tesseract $outtiff $fileprefix -l $lang hocr pdf txt >> $logfile 2>&1`;
  if (-z $pdffile) {
    print "$pdffile is empty!\n";
  } else {
    $pdfs = $pdfs . ' ' . $pdffile;
  }

  my $txtfile = "$fileprefix.txt";
  my $words = `cat $txtfile | wc -w`;
  chomp($words);
  my $correctwords = `hunspell -d be_BY -G $txtfile | wc -l`;
  chomp($correctwords);
  print(sprintf("%.0f", $words ? 100*$correctwords/$words : 0) . "% good, ");

  rename($fileprefix . '.hocr', $hocr);
  unlink($outtiff) or warn "Cannot unlink $outtiff: $!";

  my $tempdjvused = $fileprefix . '.djvused.txt';
  `hocr2djvused $hocr > $tempdjvused 2>> $logfile`;
  #unlink($hocr) or warn "Cannot unlink $hocr: $!";

  `djvused "$onepagedjvu" -f "$tempdjvused" -s >> $logfile 2>&1`;
  
  my $bwdjvu = $outputprefix . '.bw.djvu';
  if ($i == 1) {
    rename($onepagedjvu, $bwdjvu);
  } else {
    `djvm -i $bwdjvu $onepagedjvu >> $logfile 2>&1`;
  }

  $djvused = $fileprefix . ".out.djvused.txt" ;

  open(TXT, "$tempdjvused") ;
  open(TXTDJVU, ">$djvused") ;
  my $line;
  while ($line=<TXT>) {
    $line =~ s/select 1/select $i/g;
    print TXTDJVU $line
  }
  close (TXT);
  close (TXTDJVU);
  #  unlink($tempdjvused) or warn "Cannot unlink $tempdjvused: $!";

  #  print "Writing the text in the DJVU file\n";
  `djvused $outputdjvu -f $djvused -s >> $logfile 2>&1`;
  #  print "djvused $outputdjvu -f $djvused -s returned $?\n";
  #  unlink($djvused) or warn "Cannot unlink $djvused: $!";
  my $end = Time::HiRes::gettimeofday();
  my $iterationtime = $end - $start;
  $elapsedtime += $iterationtime;
  $estimatedduration = $elapsedtime / $i * $nbpages;
  my $timeleft = $estimatedduration - $elapsedtime;
  my $timeleftsec = $timeleft % 60;
  my $timeleftmin = (($timeleft - $timeleftsec) / 60) % 60;
  my $timelefthour = ($timeleft - $timeleftsec - $timeleftmin * 60) / 3600;
  printf("%.0f:%.0f:%.0f left.\n", $timelefthour, $timeleftmin, $timeleftsec);
}

`pdfunite $pdfs $outputprefix.pdf >> $logfile 2>&1`;
print "pdfunite returned $?\n";
#`rm -f -v $pdfs`;