module ConsultaSunat::ParseContent

Public Class Methods

clean_page(page) click to toggle source
# File lib/consulta_sunat/parse_content.rb, line 20
def self.clean_page page
  page.encode!('UTF-8', 'iso-8859-1')
  page.gsub(/\r\n/, "\n")
end
parse_page(page) click to toggle source
# File lib/consulta_sunat/parse_content.rb, line 4
def self.parse_page page
  page = clean_page page
  file_path = File.join(File.dirname(__FILE__), "sunat_match_data.yaml")
  sunat_match_data = YAML.load_file(file_path)
  sunat_data = {}
  sunat_match_data.each do |key, expr|
    sunat_data[key] = if result = page.match(expr)
      result[1]
    else
      "-"
    end
  end

  return sunat_data
end