- def fetch_data_for_activity(%Activity{} = activity) do
- with true <- Pleroma.Config.get([:rich_media, :enabled]),
- %Object{} = object <- Object.normalize(activity.data["object"]),
- {:ok, page_url} <- HTML.extract_first_external_url(object, object.data["content"]),
+ @options [
+ pool: :media,
+ max_body: 2_000_000,
+ recv_timeout: 2_000
+ ]
+
+ @spec validate_page_url(URI.t() | binary()) :: :ok | :error
+ defp validate_page_url(page_url) when is_binary(page_url) do
+ validate_tld = Config.get([Pleroma.Formatter, :validate_tld])
+
+ page_url
+ |> Linkify.Parser.url?(validate_tld: validate_tld)
+ |> parse_uri(page_url)
+ end
+
+ defp validate_page_url(%URI{host: host, scheme: "https", authority: authority})
+ when is_binary(authority) do
+ cond do
+ host in Config.get([:rich_media, :ignore_hosts], []) ->
+ :error
+
+ get_tld(host) in Config.get([:rich_media, :ignore_tld], []) ->
+ :error
+
+ true ->
+ :ok
+ end
+ end
+
+ defp validate_page_url(_), do: :error
+
+ defp parse_uri(true, url) do
+ url
+ |> URI.parse()
+ |> validate_page_url
+ end
+
+ defp parse_uri(_, _), do: :error
+
+ defp get_tld(host) do
+ host
+ |> String.split(".")
+ |> Enum.reverse()
+ |> hd
+ end
+
+ def fetch_data_for_object(object) do
+ with true <- Config.get([:rich_media, :enabled]),
+ false <- object.data["sensitive"] || false,
+ {:ok, page_url} <-
+ HTML.extract_first_external_url_from_object(object),
+ :ok <- validate_page_url(page_url),