pdf2zip.rb

#!/usr/bin/env ruby -wKU
# PDF to Zip file (including jpg files) script.
# sudo port install rb-cocoa
# sudo gem install zipruby
require 'osx/cocoa'
require 'rubygems'
require 'fileutils'
require 'zipruby'


class Pdf2Zip
  def find_pdf_rep source
    reps = source.representations
    (0...reps.count).each do |index|
      pdf = reps.objectAtIndex(index)
      return pdf if pdf.isKindOfClass(OSX::NSPDFImageRep)
    end
    return nil
  end

  def calc_size pdf_rep, fit=800
    width   = pdf_rep.pixelsWide
    height  = pdf_rep.pixelsHigh
    if width > height
      height  = (fit.to_f / width.to_f * height.to_f).to_i
      width   = fit
    else
      width   = (fit.to_f / height.to_f * width.to_f).to_i
      height  = fit
    end
    OSX::NSMakeSize(width, height)
  end

  def pdf2zip pdf_path, output_folder, suggestion_size=800, compression_factor=0.7
    output_file   = File.join( output_folder, File.basename(pdf_path, File.extname(pdf_path)) )
    return false if File.file?("#{output_file}.zip")
    temp_folder   = "#{output_file}.temp"

    puts "Creating zip: #{pdf_path} to #{output_file}"
    source      = OSX::NSImage.alloc.initWithContentsOfFile(pdf_path)
    source.setScalesWhenResized(true)
    source.setDataRetained(true)
    pdf_rep     = find_pdf_rep(source)
    prop_jpeg   = OSX::NSDictionary.dictionaryWithObjectsAndKeys(
                  OSX::NSNumber.numberWithFloat(compression_factor),
                  OSX::NSImageCompressionFactor,
                  nil
    )

    FileUtils.mkdir_p(temp_folder) unless File.directory?(temp_folder)

    (0...pdf_rep.pageCount).each do |index|
      path = File.join(temp_folder, "#{sprintf('%04d', index)}.jpg")
      pdf_rep.setCurrentPage(index)
      source.recache

      size = calc_size(pdf_rep, suggestion_size)
      destination_rect = OSX::NSMakeRect(0, 0, size.width, size.height)
      image = OSX::NSImage.alloc.initWithSize(size)
      image.lockFocus
      # OSX::NSApplication.sharedApplication
      OSX::NSGraphicsContext.currentContext.setImageInterpolation(OSX::NSImageInterpolationHigh)

      OSX::NSColor.whiteColor.set
      OSX::NSRectFillUsingOperation(destination_rect, OSX::NSCompositeClear)
      pdf_rep.drawInRect(destination_rect)
      bitmap = OSX::NSBitmapImageRep.alloc.initWithFocusedViewRect(destination_rect)
      data = bitmap.representationUsingType_properties(OSX::NSJPEGFileType, prop_jpeg)
      bitmap.release
      data.writeToFile_atomically(path, true)

      image.unlockFocus
      image.release
    end

    source.release
    Zip::Archive.open("#{output_file}.zip", Zip::CREATE) do |arc|
      Dir.glob(File.join(temp_folder, '**', '*.jpg')) do |f|
        arc.add_file(f)
      end
    end
    FileUtils.rm_rf(temp_folder)
  end
end

if __FILE__ == $0
  require 'optparse'
  size = 800
  compression_factor = 0.7

  opt = OptionParser.new
  opt.on('-s', '--size SIZE') { |v| size = v.to_i }
  opt.on('-c', '--compression-factor COMPRESSION_FACTOR') { |v| compression_factor = v.to_f }
  opt.parse!

  input   = File.expand_path(ARGV[0])
  output  = File.expand_path(ARGV[1])

  if File.directory?(input)
    Dir.glob(File.join(input, '**', '*.pdf')) do |file|
      output_folder = File.join(output, File.dirname(file[input.bytesize .. -1]))
      Pdf2Zip.new.pdf2zip(file, output_folder, size, compression_factor)
    end
  else
    Pdf2Zip.new.pdf2zip(input, output, size, compression_factor)
  end
end