Naive solution for finding a map page on a PDF by word frequency

map_extract.rb

require 'rmagick'
require 'rtesseract'

## Utility for extracting the map page from a survey_map pdf
class Utils::MapExtract
  attr_reader :pdf_file, :images

  def initialize(pdf_file)
    @pdf_file = pdf_file
    @images = []
  end

  # Returns an io instance of the page with fewest words
  def process!
    Magick::Image.read(pdf_file).each_with_index do |pdf, idx|
      filename = "#{basename}_#{idx}.png"
      tmp = "/tmp/#{filename}"
      pdf.write(tmp)
      word_count = word_count_on_page(tmp)
      images << { file: tmp, size: word_count }
    end
    io = File.read(map[:file])
    cleanup
    io
  end

  def map
    @map ||= images.sort_by { |i| i[:size] }.first
  end

  def cleanup
    images.each { |img| File.delete img[:file] }
  end

  def basename
    File.basename(pdf_file, '.pdf')
  end

  def word_count_on_page(img)
    RTesseract.new(img).to_s.split(' ').count
  end
end