ocr

#!/usr/bin/perl
#
# OCR and image (extract text) from a file/url
# - wrapper for tesseract and imagemagick
#
# -samy kamkar 2020/12/06

use strict;
use y;;;
die "usage: $0 <path/url to image>\n" unless @ARGV == 1;

my $TESSERACT = "tesseract"; # ocr engine
my $WGET = "wget";
my $CONVERT = "convert"; # imagemagick

my @convopts = ('-draw', 'color 0,0 floodfill', qw/-fuzz 50% -fill white -opaque white -bordercolor black -border 2 -fill white -alpha off -negate -units pixelsperinch -density 72/);

my $DIR = "/tmp";

my $file = shift;
my $rand = time();
if (!-e $file)
{
  my $nf = "$DIR/.ocr.get.$rand";
  runv($WGET, $file, "-O", $nf);
  $file = $nf;
}

my $img = "$DIR/.ocr.new.$rand.png";
runv($CONVERT, $file, @convopts, $img);
#system("open", $img);

# tesseract needs to operate in the directory for some reason?
# otherwise get no such file (in a specific version)

chdir($DIR);
# run tesseract and output to file, it adds .txt to end
my $out = "$DIR/.ocr.out.$rand";
runv($TESSERACT, $img, $out);
$out .= ".txt";

# read our file
print cat($out);
#open(OUT, "<$out") || die $!;
#print while <OUT>;
#close(OUT);