feed_parser/lib/parser/atom.ex

defmodule FeedParser.Parser.Atom do
  @moduledoc """
  A `FeedParser.Parser` that handles [Atom feeds](https://validator.w3.org/feed/docs/atom.html).
  """

  alias FeedParser.XML
  require XML

  @behaviour FeedParser.Parser

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

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

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

      _ ->
        false
    end
  end

  @impl FeedParser.Parser
  def parse_feed(feed) do
    title = text('/feed/title/text()', feed)
    link = attr('/feed/link/@href', feed)
    icon = text('/feed/icon/text()', feed)
    feed_author = texts('/feed/author/name/text()', feed)

    updated =
      text('/feed/updated/text()', feed)
      |> Timex.parse("{ISO:Extended}")
      |> case do
        {:ok, date} -> date
        _ -> nil
      end

    items =
      :xmerl_xpath.string('/feed/entry', feed)
      |> Enum.map(fn entry ->
        id = text('/entry/id/text()', entry)
        title = text('/entry/title/text()', entry)

        links =
          :xmerl_xpath.string('/entry/link', entry)
          |> Enum.map(fn link ->
            value = attr('/link/@href', link)
            rel = attr('/link/@rel', link)
            {value, rel}
          end)

        url =
          (Enum.find(links, fn {_value, rel} -> rel == "alternate" end) || List.first(links))
          |> case do
            url when is_binary(url) -> url
            {url, _rel} -> url
          end

        updated =
          text('/entry/updated/text()', entry)
          |> Timex.parse("{ISO:Extended}")
          |> case do
            {:ok, date} -> date
            _ -> nil
          end

        author =
          (texts('/entry/author/name/text()', entry) || feed_author)
          |> case do
            nil -> nil
            authors -> Enum.join(authors, ", ")
          end

        content = text('/entry/content/text()', entry) || text('/entry/summary/text()', entry)

        %FeedParser.Item{
          guid: id,
          title: title,
          url: url,
          links: links,
          content: content,
          date: updated,
          creator: author
        }
      end)

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

  defp text(xpath, element) do
    case texts(xpath, element) do
      [text] -> text
      _ -> nil
    end
  end

  defp texts(xpath, element) do
    case :xmerl_xpath.string(xpath, element) do
      [] ->
        nil

      els ->
        Enum.map(els, fn el ->
          XML.xmlText(el, :value) |> List.to_string() |> String.trim()
        end)
    end
  end

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

      _ ->
        nil
    end
  end
end