feed_parser/lib/parser/rss2.ex

defmodule FeedParser.Parser.RSS2 do
  @moduledoc """
  A `FeedParser.Parser` that handles [RSS 2.0 feeds](https://cyber.harvard.edu/rss/rss.html).
  """

  alias FeedParser.XML
  require XML

  @behaviour FeedParser.Parser

  @impl FeedParser.Parser
  def accepts(data, content_type) do
    cond do
      content_type in ["application/rss+xml", "text/rss+xml"] ->
        case XML.parse(data) do
          {:error, _} -> false
          {:ok, doc} -> {true, doc}
        end

      content_type in ["text/xml", "application/xml"] ->
        case XML.parse(data) do
          {:error, _} ->
            false

          {:ok, doc} ->
            if XML.xmlElement(doc, :name) == :rss do
              {true, doc}
            else
              false
            end
        end

      true ->
        false
    end
  end

  @impl FeedParser.Parser
  def parse_feed(rss) do
    [channel] = :xmerl_xpath.string('/rss/channel', rss)
    title = text('/channel/title/text()', channel)
    link = text('/channel/link/text()', channel)
    image = text('/channel/image/url/text()', channel)

    last_updated =
      text('/channel/lastBuildDate/text()', channel)
      |> Timex.parse("{RFC1123}")
      |> case do
        {:ok, date} -> date
        _ -> nil
      end

    items =
      :xmerl_xpath.string('/channel/item', channel)
      |> Enum.map(fn item ->
        guid = text('/item/guid/text()', item)
        title = text('/item/title/text()', item)
        link = text('/item/link/text()', item)
        description = text('/item/description/text()', item)

        # http://purl.org/rss/1.0/modules/content/
        content = text('/item/content:encoded/text()', item)

        pubDate =
          text('/item/pubDate/text()', item)
          |> Timex.parse("{RFC1123}")
          |> case do
            {:ok, date} -> date
            _ -> nil
          end

        # from Dublin Core extension: https://www.rssboard.org/rss-profile#namespace-elements-dublin
        # todo: should this only be attempted if the xmlns:dc is defined?
        creator = text('/item/dc:creator/text()', item)

        %FeedParser.Item{
          guid: guid,
          title: title,
          url: link,
          links: [{link, nil}],
          content: content || description,
          date: pubDate,
          creator: creator
        }
      end)

    {:ok,
     %FeedParser.Feed{
       site_url: link,
       title: title,
       image_url: image,
       last_updated: last_updated,
       items: items
     }}
  end

  defp text(xpath, element) do
    case :xmerl_xpath.string(xpath, element) do
      [el] ->
        XML.xmlText(el, :value) |> List.to_string() |> String.trim()

      _ ->
        nil
    end
  end
end