feedparser

defmodule Feedparser do
  defmodule Utils do
    @moduledoc """
    Finds Rss/Atom feed urls in html strings.
    """

    @vsn 0.1

    @doc """
    Takes a chunk of html as input and returns all valid feed urls in it
    """

    @spec get_feeds(char_list) :: [char_list]
    def get_feeds(html) do
      urls = urls(html)
      |> List.flatten
      |> Enum.filter(&(feed?(&1)))
      |> Enum.map(&(extract_href &1))
      |> Enum.filter(&(&1 != nil))
    end

    @spec urls(char_list) :: [char_list]
    defp urls(html) do
      Regex.scan(~r/<(?:a|link) .*?>/, html)
    end

    @spec feed?(char_list) :: boolean
    defp feed?(url) do
      atom?(url) or rss?(url)
    end

    @spec atom?(char_list) :: boolean
    defp atom?(url), do: url =~ ~r/.*type=['"]application\/atom\+xml['"].*/

    @spec rss?(char_list) :: boolean
    defp rss?(url),  do: url =~ ~r/.*type=['"]application\/rss\+xml['"].*/

    @spec extract_href(char_list) :: char_list
    defp extract_href(url) do
      matches = Regex.run(~r/.*href=['"](.*?)['"].*/, url)
      if matches != nil and length(matches) == 2 do
        List.last matches
      end
    end
  end
end