#!/usr/bin/env ruby 

# By Michael Moore
# Placed in the Public Domain for any use by anybody. 
# A link to my site is appreicated, but by no means required http://bookscanned.com

# Make this an empty string if you want verbose output, or redirect 2>1 if you want everything in the TakOCR output 
quiet="2>/dev/null"

ENV['MAGICK_HOME']="/usr/local/ImageMagick-6.4.8/"
ENV['DYLD_LIBRARY_PATH']="#{ENV['MAGICK_HOME']}/lib"
ENV['PATH']="#{ENV['MAGICK_HOME']}/bin:/usr/local/bin:#{ENV['PATH']}"
ENV['TESSDATA_PREFIX']='/usr/local/share/'

tmpdir = '/tmp/takocr' + rand(1000000).to_s
Dir.mkdir(tmpdir,0700)

# Keep png files, don't imagemagick them
imgs = Array.new()

ARGV.each do |img| 
    if(File.extname(img).downcase == '.png')
	#OCRopus handles pngs fine
	imgs << img
    elsif(File.extname(img).downcase == '.pdf' || File.extname(img).downcase == '.tiff' || File.extname(img).downcase == '.tif')
	# Convert every page of multipage documents into a png file
	`convert -density 300 -units PixelsPerInch #{img} #{tmpdir}/#{File.basename(img)}%d.png #{quiet}`
    else
	# Convert other images to png 
	`convert -density 300 -units PixelsPerInch #{img} #{tmpdir}/#{File.basename(img)}.png #{quiet}`
    end
end

cmd = "ocroscript recognize #{imgs.join(' ')} #{Dir.entries(tmpdir).size > 2 ? tmpdir + "/*.png" : ''}"
puts `#{cmd} #{quiet}`