+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
defmodule Pleroma.Web.RichMedia.Parser do
@parsers [
Pleroma.Web.RichMedia.Parsers.OGP,
def parse(url), do: parse_url(url)
else
def parse(url) do
- with {:ok, data} <- Cachex.fetch(:rich_media_cache, url, fn _ -> parse_url(url) end) do
- data
- else
- _e ->
- {:error, "Parsing error"}
+ try do
+ Cachex.fetch!(:rich_media_cache, url, fn _ ->
+ {:commit, parse_url(url)}
+ end)
+ rescue
+ e ->
+ {:error, "Cachex error: #{inspect(e)}"}
end
end
end
defp parse_url(url) do
- {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url)
+ try do
+ {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: [pool: :media])
- html |> maybe_parse() |> get_parsed_data()
+ html |> maybe_parse() |> clean_parsed_data() |> check_parsed_data()
+ rescue
+ e ->
+ {:error, "Parsing error: #{inspect(e)}"}
+ end
end
defp maybe_parse(html) do
end)
end
- defp get_parsed_data(data) when data == %{} do
- {:error, "No metadata found"}
+ defp check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do
+ {:ok, data}
+ end
+
+ defp check_parsed_data(data) do
+ {:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
end
- defp get_parsed_data(data) do
- {:ok, data}
+ defp string_is_valid_unicode(data) do
+ data
+ |> :unicode.characters_to_binary()
+ |> clean_string()
+ end
+
+ defp clean_string({:error, _, _}), do: {:error, "Invalid data"}
+ defp clean_string(data), do: {:ok, data}
+
+ defp clean_parsed_data(data) do
+ data
+ |> Enum.reject(fn {_, val} ->
+ case string_is_valid_unicode(val) do
+ {:ok, _} -> false
+ _ -> true
+ end
+ end)
+ |> Map.new()
end
end