A Ruby module for searching for FEC images

paper_chase.rb

# encoding: utf-8
require 'net/http'
require 'nokogiri'
require 'uri'

module PaperChase

  class Report
    attr_accessor :type, :amended, :filed_on, :from_date, :end_date, :pages, :page_by_page, :pdf_url
    
    def initialize(params={})
      @type = params[:type]
      @amended = params[:amended]
      @filed_on = params[:filed_on]
      @from_date = params[:from_date]
      @end_date = params[:end_date]
      @pages = params[:pages]
      @page_by_page = params[:page_by_page]
      @pdf_url = params[:pdf_url]
    end
  end

  class Committee
    attr_accessor :id, :name, :city, :state, :party, :designation, :type, :candidate_state, :candidate_office

    def initialize(params={})
      @id = params[:id]
      @name = params[:name]
      @city = params[:city]
      @state = params[:state]
      @party = params[:party]
      @designation = params[:designation]
      @type = params[:type]
      @candidate_state = params[:candidate_state]
      @candidate_office = params[:candidate_office]
    end

    def uri
      URI.parse("http://images.nictusa.com/cgi-bin/fecimg/?#{@id}")
    end

    def page
      @page ||= Net::HTTP.get(uri)
    end

    def doc
      @doc ||= Nokogiri::HTML(page)
    end

    def reports
      return @reports if @reports

      table = doc.css("table")[2]
      return [] unless table

      @reports = doc.css("table")[2].css('td[align="center"] a').map do |href|
        row = href.parent.parent
        cells = row.css("td")
        PaperChase::Report.new({
          :type => cells[0].text,
          :amended => cells[1].text,
          :filed_on => cells[2].text,
          :from_date => cells[3].text,
          :end_date => cells[4].text,
          :pages => cells[5].text,
          :page_by_page => "http://images.nictusa.com/#{cells[6].css("a")[0].attr("href")}",
          :pdf_url => "http://images.nictusa.com/#{cells[7].css("a")[0].attr("href")}"
        })
      end
      @reports
    end
  end

  class Search

    def initialize(search_params={})
      @search_params = make_params(search_params)
      @uri = URI.parse("http://images.nictusa.com/cgi-bin/fecimg/")
      @response = search
    end

    def make_params(search_params)
      allowed = [:filerid, :name, :treas, :city, :img_num, :state, :party, :type]
      params = {}
      allowed.each do |param|
        params[param.to_s] = search_params[param] || ''
      end
      params
    end

    def search
      Net::HTTP.post_form(@uri, @search_params)
    end

    def results(&block)
      @response.body
      doc = Nokogiri::HTML(@response.body)
      rows = []
      fields = [:id, :name, :city, :state, :party, :designation, :type, :candidate_state, :candidate_office]
      doc.css("tr")[1..-1].each do |row|
        result = PaperChase::Committee.new(Hash[*fields.zip(row.css("td").map { |x| x.text.gsub("\302\240", ' ').strip }).flatten(1)])
        if block_given?
          yield result
        else
          rows << result
        end
      end
      block_given? ? nil : rows
    end

  end

end