Class and script to make json/csv from ICF classificator

parse-icf.rb

#!/usr/bin/env ruby
# As source you need HTML file from http://apps.who.int/classifications/icfbrowser/Default.aspx
# with unfolded tree.
# It uses Ajax to fetch nested entries, so unfold each manually or you can use JS console to autoclick.
require 'optparse'
require 'nokogiri'
require 'json'
require 'csv'

class ICFEntry < Struct.new(:id, :code, :title, :children)
  class << self;
    def make(src)
        make_node(src.css('> div')).first
    end

    private
    @@last_id = 0

    def gen_id
      @@last_id += 1
    end

    def make_node(node)
      tables = node.children.select{|c| c.name == 'table'}
      tables.map do |tbl|
        id = gen_id # Make id first for order
        content = tbl.text.strip
        matches = content.match /(\S\d*)\s(.*)/
        code,title = matches.nil? ? [nil,content] : matches.captures
        i = node.children.index(tbl)
        ni = i + 1
        children = if node.children.length > ni
          div = node.children[ni]
          if div.name == 'div'
            make_node(div)
          end
        end
        ICFEntry.new(id, code, title, children)
      end
    end
  end

  def to_h
    out = {code: code, title: title}
    out = out.merge({children: children.map{|c| c.to_h}}) unless children.nil?
    out
  end

  def to_sh
    if children.nil?
        code_title
    else
        {title: code_title, children: children.map{|c| c.to_sh}}
    end
  end

  def to_arr(parent=nil, acc=[])
    out = [code, title, parent.nil? ? nil : parent.id]
    acc << out
    unless children.nil?
      children.each do |child|
        child.to_arr(parent=self, acc=acc)
      end
    end
    acc
  end

  def code_title
    "#{code} #{title}"
  end

  def to_csv()
    CSV.generate(write_headers: true, headers: ["code", "title", "parent"]) do |csv|
      to_arr.each do |entry|
        csv << entry
      end
    end
  end

  def to_json(short=false)
    if short
      to_sh.to_json
    else
      to_h.to_json
    end
  end
end

options = {short: false}
options_parser = OptionParser.new do |opts|
  opts.banner = "Usage: convert.rb [options] html-source-file"
  opts.on("-t [TO]", "--to [TO]", [:csv,:json], "Output format") do |v|
    options[:to] = v
  end
  opts.on("-s", "--short", "Use short json representation") do |v|
    options[:short] = true
  end
end
options_parser.parse!

filename = ARGV.pop
unless filename
  puts options_parser
  exit(1)
end

TREE_ID = "ctl00_ContentPlaceHolder1_tree"
src = Nokogiri::HTML(open(filename))
src = src.css("##{TREE_ID}").first
dict = ICFEntry.make src

if options[:to] == :csv
  puts dict.to_csv
elsif options[:to] == :json
  puts dict.to_json(options[:short])
end