#!/usr/bin/perl -s

use 5.012;
use Time::HiRes qw[tv_interval gettimeofday];
use Time::Piece;
use Capture::Tiny 'capture';

our $h;
our $lang   ||= 'por';
our $latin1 ||= undef;

if ($h) {
  print STDERR "$0 [-lang=por] [-latin1] file.png ...\n";
  exit;
}

sub progress($$$) { printf STDERR "\r[%s%%|ETA %s] %- 50s" => @_ }
sub format_time($) { my $t = Time::Piece->strptime(int($_[0]),"%s"); $t->hms }

my @filenames = @ARGV;

my $t0 = [gettimeofday];
my $eta = format_time(1_000_000);
my $nr_pages = @filenames;
my $i = 1;
printf STDERR "Using Tesseract to OCR %d files.\n", $nr_pages;

print STDERR "...";
while (my $file = shift @filenames) {
    my $percent = sprintf "%6.2f" => ($i/$nr_pages)*100;

    progress $percent, $eta => "Performing OCR on file ".substr($file,0,30)."...";

    my $out = $file;
    $out =~ s/.(?:png|jpg)$//;
    capture {
        `tesseract $file $out -l $lang`;
    };

    if ($latin1) {
        `iconv -c -f utf8 -t latin1 < $out.txt > _$$; mv _$$ $out.txt`;
    }

    if ($i % 2) {
        my $elapsed = tv_interval $t0;
        ## $i = $elapsed
        ## $nr_pages-$i  == $x
        $eta = format_time(($elapsed * ($nr_pages-$i))/$i);
    }
    $i++;
}
$eta = format_time(tv_interval $t0);
progress 100, '00:00:00' => "done in [$eta]";

printf STDERR "\n";