print dina4/dina3 stats for PDF files in recursive folders

PDFstats.rb

require 'pdf-reader'


def check_pdf(input_path)
  dina5count = 0
  dina4count = 0
  dina3count = 0
  dina2count = 0
  dina1count = 0
  dina0count = 0

  threshold = 1.25

  reader = PDF::Reader.new(input_path)
  reader.pages.each do |page|
    bbox   = page.attributes[:MediaBox]
    width  = bbox[2] - bbox[0]
    height = bbox[3] - bbox[1]

    if height < width
      oldwidth = width
      width = height
      height = oldwidth
    end

    #logger.debug "#{width} #{height}"

    if (width <= 420*threshold) && (height <= 595*threshold)
      dina5count += 1      
    elsif (width <= 595*threshold) && (height <= 842*threshold)
      dina4count += 1
    elsif (width <= 842*threshold) && (height <= 1190*threshold)
      dina3count += 1
    elsif (width <= 1190*threshold) && (height <= 1684*threshold)
      dina2count += 1
    elsif (width <= 1684*threshold) && (height <= 2380*threshold)
      dina1count += 1
    elsif (width <= 2380*threshold) && (height <= 3368*threshold)
      dina0count += 1
    else
      logger.error "error: unkown filesize for mailing: #{page.inspect}}"
    end
  end
  [dina5count, dina4count, dina3count + dina2count + dina1count + dina0count]
end

docs = 0
dina4 = 0
dina3 = 0 
Dir["./**/*.pdf"].each do |f|
  dina5count, dina4count, dina3count = check_pdf(f)
  docs += 1
  puts "#{f}, #{dina5count+dina4count}, #{dina3count} "
  dina4 += dina5count + dina4count
  dina3 += dina3count
end
puts "all, #{docs}: #{dina4}, #{dina3}"