A script to generate a Graphviz file for what happens to bills in congress

congress_bill_transitions.rb

require 'set'
require 'pp'
require 'json'
require 'date'
require 'bigdecimal'

module BillStateGraphs

  class BillAction
    attr_accessor :acted_at

    def initialize (hash)
      @hash = hash
      @acted_at = DateTime.parse(hash['acted_at'])
    end

    def status
      @hash['status']
    end
  end

  class Bill
    attr_accessor :actions

    def initialize (hash)
      @hash = hash
      @actions = @hash['actions'].map{ |act| BillAction.new(act) }
      @actions.sort_by!(&:acted_at)
    end

    def transitions
      transition_list = []
      status = nil
      while actions.length > 0
        act = actions.shift
        if not act.status.nil? and act.status != status
          transition_list.push [status, act.status]
          status = act.status
        end
      end
      return transition_list
    end
  end

  def self.node_id_for_status (status)
    # Graphviz doesn't like colons
    if status.nil?
      "nil"
    else
      status.gsub(/[^_a-zA-Z0-9]/, '_')
    end
  end

  def self.run (congress_dir, congresses, matrix_path, graph_path)
    statuses = Set.new
    matrix = {}
    total = BigDecimal.new(0)
    congresses.map(&:to_i).each do |cong_num|
      bill_file_paths = Dir.glob(File.join(congress_dir,
                                           cong_num.to_s,
                                           'bills',
                                           '*',
                                           '*',
                                           'data.json'))

      bill_file_paths.each do |path|
        File.open(path, 'r') do |bill_file|
          hash = JSON.load(bill_file)
          bill = Bill.new(hash)
          bill.transitions.each do |a, b|
            statuses.add(a)
            statuses.add(b)
            row = matrix.fetch(a, {})
            cnt = row.fetch(b, 0) + 1
            row[b] = cnt
            matrix[a] = row
            total += 1
          end
        end
      end
    end

    statuses = statuses.to_a.map(&:to_s).sort!
    maxlen = statuses.map(&:length).max
    format = "%#{maxlen}s " * (statuses.length + 1) + "\n"
    File.open(matrix_path, 'w') do |matrix_file|
      matrix_file.printf(format, "", *statuses)
      statuses.each do |a|
        row = matrix.fetch(a, {})
        values = statuses.map{ |b| row.fetch(b, '') }
        matrix_file.printf(format, a, *values)
      end
    end

    File.open(graph_path, 'w') do |graph_file|
      graph_file.puts("digraph {")
      matrix.each do |a, row|
        row.each do |b, cnt|
          pct = (BigDecimal.new(cnt) * 100 / total).round(2).to_f.to_s
          graph_file.puts("#{node_id_for_status(a)} -> #{node_id_for_status(b)} [label=\"#{cnt}\\n(#{pct}%)\"] ;")
        end
      end
      graph_file.puts("}")
    end
  end
end