Washington Mutual OFX Downloads Scraper

I put this together to automaticaally download my bank statements from WaMu, soon to be JP Morgan Chase.


class WamuRobot < OnlineAccountRobot
  BASE_URL = "https://online.wamu.com/Servicing/Servicing.aspx"

  def initialize(username, password, opts={})
    super(opts)
    agent.keep_alive = false

    page = request { agent.get "https://online.wamu.com/IdentityManagement/Logon.aspx" }
    form = page.form('frmLogin')

    form['txtUserID'] = username
    form['password'] = Encryption.decrypt(password)
    form['chkRememberMe'] = 'off'

    page = request { agent.submit form }

    if page.uri.to_s.match("https://online.wamu.com/banking/offers")
      page = request { agent.get "https://online.wamu.com/Servicing/Servicing.aspx?targetPage=AccountSummary" }
    end

    if page.uri.to_s == "https://online.wamu.com/Servicing/Servicing.aspx?targetPage=AccountSummary"
      process_summary(page)
    elsif page.body.match(/your entry is not valid/)
      raise InvalidLoginError.new("Login invalid")
    else
      raise unknown_page(page)
    end
  end

  def download(acct_number, from=nil, to=Date.today)
    from ||= to - 30.days
    index = acct_index(acct_number)

    page = request { agent.get "#{BASE_URL}?targetPage=TransactionDownload&accountIndex=#{index}" }
    form = page.form('__aspnetForm')

    form["_ctl3:txtFromDate"] = from.strftime('%m/%d/%Y')
    form["_ctl3:txtToDate"] = to.strftime('%m/%d/%Y')
    form["_ctl3:softwareFormatDropDownList"] = "OFX"

    page = request { form.click_button }

    if page.body.match(/Verify transactions to download/)
      output = request { page.form('__aspnetForm').click_button }
    else
      raise unknown_page(page)
    end

    if output.is_a?(WWW::Mechanize::File) && output.body.match(/DATA:OFXSGML/)
      return output.body
    else
      if output.body.match(/No Transactions available/)
        raise DataNotAvailableError.new("No transactions during that time period")
      else
        raise unknown_page(output)
      end
    end
  end

  def process_summary(doc)
    table = doc.search("//table[@id='depositSummaryRptr']")
    (table/"tr").search("a[href*='accountIndex']").each do |a|
      link = a.attributes['href'].to_s.match(/accountIndex=(\d+)/)
      text = a.parent.inner_text.strip.match(/(\d+)/)

      accounts[text[1]] = link[1]
    end
  end

  def acct_index(number)
    last_four = number.to_s.match(/.*(\d{4})$/)[1]
    accounts[last_four]
  end

  #protected
  def request(time=2)
    retryable(:on => StandardError, :times => 2) do
      sleep time
      yield
    end
  end

  def agent
    @agent ||= WWW::Mechanize.new {|a| a.log = logger }
  end


#  private
  def accounts
    @accounts ||= {}
  end

end