parser.rb

class Parser
  attr_accessor :results
  @@default_services = {}

  def parse(document)
    document = Nokogiri::XML(document).remove_namespaces!
    found    = false
    @@default_services.each do |service, content|
      if document.xpath(content[:identifier]).length > 0
        self.send("parse_#{service.to_s}", document)
        found = true
        break
      end
    end
      raise "Parser not found" unless found
  end

  def self.define_model(model)
    @@result_model = model
  end

  def self.define_interface(name, component)
    raise "Interface name already used" unless @@default_services[name].nil?
    @@default_services[name] = component

    define_method("parse_#{name.to_s}".to_sym) do |document|
      document = Nokogiri::XML(document).remove_namespaces! if document.class != Nokogiri::XML::Document
      @results ||= []

      document.xpath(component[:finder]).each do |a|
        result = @@result_model.new

        component[:fields].each do |key, path|
          next if path.nil?

          if path.class == Proc
            value = path.call(a)
          else
            value = path.match(/^@/) ?
              a.attribute(path.gsub(/@/, "")).text :
              a.xpath(path).text
          end
          result.send("#{key.to_s}=".to_sym, value)
        end

        @results << result
      end

    @results.length
    end
  end
end

# Example Default Services
default_services = {
  :factiva => {
    :finder  => "//result",
    :identifier => "/PPSResponse/ppsresultResponse",
    :fields => {
      :source  => "sourceName",
      :title   => "headline/paragraph",
      :content => lambda { |result|
        content = ""
        %w{leadParagraph/paragraph tailParagraphs/paragraph}.each do |search|
          content += result.xpath(search).inject("") { |c, p| c = c + p.text }
        end
        content
      },
      :date      => "publicationDate/date",
      :url       => nil,
      :custom_id => "accessionNo",
      :language  => nil
    }
  }
}

Parser.define_model(Article)

default_services.each do |name, component|
  Parser.define_interface(name, component)
end