pdf2png: runs 'mutool extract' and renumbers

pdf2png.rb

#!/usr/bin/ruby
# * coding: utf-8 *
class PDFExtract
  def self.main
    ARGV.each {|a| new.extract a }
  end

  def extract(pdf)
    images = parse_info(`mutool info #{pdf}`)

    images.each do |img|
      if nonempty?(img.renumbered_name)
        puts "skipping #{img}"
      else
        unless nonempty?(img.extracted_name)
          cmd = %{mutool extract -r "#{pdf}" "#{img.id}"}
          system cmd
          $? == 0 or raise "#{cmd} failed: #$?"
        end

        puts "renaming #{img}"
        File.rename(img.extracted_name, img.renumbered_name)
      end
    end
  end

  private

  def nonempty?(name)
    File.size?(name).to_i > 0
  end

  def parse_info(str)
    /^Images \((?<pages>\d+)/m =~ str or raise "parse error"
    images = $'.scan(/(?:DevGray|DevRGB|ICC|RGB|CMYK) \((\d+)/).each_with_index.map {|s, i| Image.new(i + 1, Integer(s[0])) }
    images.size == Integer(pages) or raise "unexpected parse result: #{images.size} RGBs vs #{pages} images"
    images
  end

  Image = Struct.new(:page, :id) do
    def to_s
      "#{extracted_name} -> #{renumbered_name}"
    end
  
    def extracted_name
      "img-%04d.png" % id
    end

    def renumbered_name
      "%04d.png" % page
    end
  end
end

PDFExtract.main if $0 == __FILE__