extract-elements-with-cfi.rb

require 'English'
require 'epub/parser'
require 'epub/parser/cfi'
require 'nokogiri'

def usage
  <<EOS

USAGE:
 ruby #{$PROGRAM_NAME} ELEMENT EPUB

EOS
end

def main(argv)
  elem_name = argv.shift
  epub_path = argv.shift
  if elem_name.nil? or epub_path.nil?
    abort usage
  end

  spine_step = {:step => 6}

  epub = EPUB::Parser.parse(epub_path)
  epub.package.spine.each_itemref.with_index do |itemref, i|
    itemref_step = {
      :step => (i + 1) * 2,
      :id => itemref.id
    }
    itemref.item.content_document.nokogiri.search(elem_name).each do |elem|
      path = find_path(elem)
      path_str = "/#{spine_step[:step]}/#{itemref_step[:step]}"
      path_str << "[#{itemref_step[:id]}]" if itemref_step[:id]
      path_str << "!"
      path.each do |elem_step|
        path_str << "/#{elem_step[:step]}"
        path_str << "[#{elem_step[:id]}]" if elem_step[:id]
      end
      puts path_str
      puts elem.content
      puts
    end
  end
end

def find_path(elem)
  steps = []
  until elem.parent.document?
    index = elem.parent.element_children.index(elem)
    steps.unshift({
      :step => (index + 1) * 2,
      :id => elem["id"]
    })
    elem = elem.parent
  end
  steps
end

main ARGV