PDF to SVG components

converter.rb

# Set up a couple of paths
lib_directory = "#{Rails.root}/lib/svgs/"
assets_directory = "#{Rails.root}/app/assets/images/pages/"
pdf_path = "/Users/iHiD/Downloads/SightSound.pdf"

(1..18).each do |id|

  # Setup image specific paths
  orignal_filename = "#{lib_directory}#{id}.svg"
  text_filename = "#{assets_directory}#{id}_text.svg"
  images_filename = "#{assets_directory}#{id}_images.svg"
  images_png_filename = "#{assets_directory}#{id}_images.png"
  images_jpg_filename = "#{assets_directory}#{id}_images.jpg"
  
  # Convert PDF to SVG using any third party library. 
  # There are loads of free ones. 
  # Poppler is a good place to start. Here I'm using pdf2svg
  `pdf2svg #{pdf_path} #{orignal_filename} #{id}`
  
  # Extract thumbnail from SVG
  `/Applications/Inkscape.app/Contents/Resources/bin/inkscape -e #{assets_directory}/thumbs/#{id}.png -w 110 -b white -z #{orignal_filename}`

  # Split an SVG into an image and a cut-down SVG. Nokogiri is just an XML parser.
  svg = File.read(orignal_filename)
  text_svg = Nokogiri::XML.parse(svg)
  
  # Remove all images, graphics and paths from the text part...
  text_svg.css("image").remove()
  text_svg.css("g").each do |node|
    node.remove() if node.keys.include?("clip-path")
  end
  text_svg.css("path").each do |node|
    if node.keys.include?("style")
      if node.attributes["style"].value.include?("fill-opacity:1")
        node.remove() 
      end
    end
  end
  
  # ...a turn this into a compressed svgz
  File.open(text_filename, 'w') {|f|  f.write(text_svg.to_xml) }
  `gzip -c #{text_filename} > #{text_filename}z`

  # Strip the text out of the images part...
  images_svg = svg.gsub(/<symbol overflow="visible" id="[^"]*">\s*<path[^>]*>\s*<\/symbol>\s*/, '') do |match|
    match.include?("glyph") ? "" : match
  end
  images_svg.gsub!(/<use xlink:href="#glyph[^\/]*\/>\s*/, '')
  File.open(images_filename, 'w') {|f|  f.write(images_svg) }

  # ...and convert it to a clean image (convert is part of ImageMagick)
  `/Applications/Inkscape.app/Contents/Resources/bin/inkscape -e #{images_png_filename} -z #{images_filename}`
  `convert #{images_png_filename} -background white -flatten #{images_jpg_filename}`
  
  # Remove tmp files
  `rm #{images_filename}`
  `rm #{images_png_filename}`
  
  # Get an XML file of the words, again using any one of lots of free libraries.
  `pdftotext -bbox -f #{id} -l #{id} #{pdf_path} #{Rails.root}/db/seed_data/words_#{id}.xml`
end