Script that uses Tesseract, Poppler and ImageMagick utilities to OCR a PDF consisting of mere images and make it searchable

ocrpdf.rb

#!/usr/bin/env ruby

main_lang = ARGV[0]
input_pdf = ARGV[1]
temp_dir = 'temp'

if main_lang && input_pdf
  `mkdir -p #{temp_dir}`
  print "Splitting PDF into separate pages... "
  `pdfseparate "#{input_pdf}" #{temp_dir}/page_%d.pdf`
  print "\nConverting to image, then running OCR: page "
  bs = 0
  Dir["#{temp_dir}/*.pdf"].each do |pdf_in|
    prefix, number = pdf_in.match(/(page_)(\d+)/).captures
    bs_str = ""; bs.times {bs_str += "\b"}
    print "#{bs_str}#{number} "
    bs = number.size + 1
    tiff = pdf_in.gsub(/\.pdf/, '.tiff')
    `convert -density 300 #{pdf_in} -depth 8 -strip -background white -alpha off #{tiff} 2>/dev/null`
    `rm #{pdf_in}`
    outname = "#{prefix}#{'%04i' % number.to_i}"
    `tesseract #{tiff} #{File.join(temp_dir, outname)} -l #{main_lang} pdf 2>/dev/null`
    `rm #{tiff}`
  end
  puts
  `pdfunite #{temp_dir}/page_????.pdf "#{input_pdf.gsub(/\.pdf/, '')} (searchable).pdf"`
  `rm -f #{temp_dir}/page_????.pdf`
  `rmdir #{temp_dir}`
else
  puts "Usage: ocrpdf.rb <main language (CAREFUL! FORMAT IS UNUSUAL; SEE BELOW!)> <input PDF>\n\n"
  puts `tesseract --list-langs`.split("\n").join(" ")
end