canonicalize feed urls

gistfile1.txt

  def test_feedburner_aliased
    series = Series.create title: 'foo', url: 'http://feeds.feedburner.com/InTheBleachersPodcast'
    assert_not_nil Series.find_by_url('http://feeds.feedburner.com/inthebleacherspodcast')
    assert_not_nil Series.find_by_url('http://feeds.feedburner.com/inthebleacherspodcast/')
    assert_not_nil Series.find_by_url('http://feeds.feedburner.com/inthebleacherspodcast.xml')
    assert_not_nil Series.find_by_url('http://feeds.feedburner.com/inthebleacherspodcast?format=rss')
    assert_not_nil Series.find_by_url('http://feeds2.feedburner.com/inthebleacherspodcast?format=rss')
    assert_not_nil (series = Series.find_by_url('http://feeds.feedburner.com/InTheBleachersPodcast'))
    assert_equal 'http://feeds.feedburner.com/inthebleacherspodcast', series.url
  end 

  def self.canonicalize_uri(uri)

    uri = Addressable::URI.heuristic_parse(uri)

    uri.scheme = 'http' unless uri.scheme=='http' or uri.scheme=='https'
    if uri.host =~ /feedburner.com$/
      uri.host = 'feeds.feedburner.com' # ignore feeds2 alias
      uri.path = uri.path.downcase.gsub(/\/$/, '').gsub(/\..+/,'')
      uri.query=nil
    end

    uri.to_s.strip

  end