feedex

defmodule Feedex do
  @moduledoc """
  A simple elixir feed parser.
  """
  require Logger

  alias Feedex.Helpers.{Fetch, Xml, Sanitizer}
  alias Feedex.Parsers.{Atom, RSS1, RSS2}

  @doc """
  Parse feed.

  ## Examples

      iex> Feedex.parse url
      {:ok, feed}
      # {:error, :empty_url} -> Logger.warn "empty url"
      # {:error, :fetch_error} -> Logger.warn "fetch error"
      # {:error, :invalid_xml} -> Logger.warn "invalid xml"
      # {:error, :unknown_feed_format} -> Logger.warn "unknown feed format"
      # {:error, :unknown} -> Logger.error "Unknown error catched"
  """
  @spec parse(String.t) :: String.t
  def parse(url) do
    try do
      with {:ok, content} <- Fetch.get(url),
           {:ok, xml}     <- Xml.preprocess(content),
           {:ok, parser}  <- select_parser(xml),
           {:ok, feed}    <- parser.parse(xml) do
        base_url = extract_base_url(feed.url)

        entries =
          feed.entries
          |> Enum.filter(fn(e) -> e.title != nil && e.content != nil end)
          |> Enum.sort(&(DateTime.compare(&1[:updated], &2[:updated]) != :gt))
          |> Enum.map(&sanitize_entry(&1, base_url))

        {:ok, %{feed | entries: entries}}
      end
    rescue
      e -> {:error, :unknown, e}
    end
  end

  # --

  defp select_parser(xml) do
    cond do
      Atom.valid? xml -> {:ok, Atom}
      RSS1.valid? xml -> {:ok, RSS1}
      RSS2.valid? xml -> {:ok, RSS2}
      true -> {:error, :unknown_feed_format}
    end
  end

  defp extract_base_url(url) do
    %{URI.parse(url) | path: "/"} |> URI.to_string
  end

  defp sanitize_entry(entry, base_url) do
    content =
      (entry.content || "")
      |> String.replace(~r/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/i, "")
      |> Sanitizer.basic_html
      |> String.replace(~r/<\s*img\ssrc\s*=\s*(['"])\//i, "<img src=\\1#{base_url}")

    %{entry | content: content}
  end

end