image-text

#!/usr/bin/env perl

use strict;
use utf8;
use File::Copy;
$ENV{'PATH'} = $ENV{'PATH'}.':/usr/local/bin';

# Can't just use this because of the PATH issue.
# Hazel doesn't have it in PATH, so we put it there.
eval
{
  require Image::OCR::Tesseract;
  Image::OCR::Tesseract->import();
};

unless($@)
{
 my $file = shift(@ARGV);
 # Handle the tempfile so we don't accidentally trigger Hazel.
  my $time = time();
  my $tempfile = "/var/tmp/tesseract-$time.jpg";
  copy($file,$tempfile) or die("Copy failed: $!");
  my $output = Image::OCR::Tesseract::get_ocr($tempfile);
  die('No Tesseract Output!') unless $output;
  unlink($tempfile);

  # Remove leading whitespace and newlines.
  $output =~ s/^(\n|\s)+//;

  # Remove trailing newlines and whitespace.
  $output =~ s/(\n|\s)+$//;

# Remove all newlines
  $output =~ s/\n/ /g;

# Replace / with -
  $output =~ s/\//-/g;

# Replace newlines with |
  $output =~ s/\n/|/g;

  # Remove non-ascii characters.
  $output =~ s/[^[:ascii:]]//g;

  # Remove quotes and commas.
  $output =~ s/(\'|\"|\,)//g;

  # Remove redirect operators.
  $output =~ s/(\<|\>)//g;

  # Remove pound signs
  $output =~ s/\#//g;

  utf8::encode($output);
  print $output;
}
exit;