convert-html-to-json.rb

# usage: ruby convert-html-to-json.rb PATH_TO_FILE

require 'rubygems'
require 'hpricot'
require 'activesupport'

def node_to_s(node, indent_level=0)
  return nil unless node.elem?

  ret = ""
  indentation = " " * indent_level
  ret << "\n%s%s(" % [indentation, node.name]
  ret << node.attributes.to_json.gsub(',', ', ') unless node.attributes.empty?
  if !node.children.empty? && node.children.detect{ |child| child.elem? }
    ret << "," unless node.attributes.empty?
    ret << node.children.map { |child| node_to_s(child, indent_level + 4) }.compact.join(',')
    ret << "\n%s)" % indentation
  elsif !node.children.empty? || !node.inner_text.blank?
    ret << ", " unless node.attributes.empty?
    ret << "'%s'" % [node.inner_text]
    ret << ")"
  else
    ret << ")"
  end
  return ret
end

file = File.read(ARGV[0])
doc = Hpricot.parse(file)
print node_to_s(doc.root)