X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=4341141df8504d67ba2d5b5129dc8ff611e24e42;hb=833161b5d21f85e2276cd0cee3e148ecbe6e1f05;hp=3746feaf688b275b8a70cc5bb15ac91a40191b58;hpb=0b54c3d6432dea77542596e34057f8d3fc69ca4c;p=akkoma diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index 3746feaf6..4341141df 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -1,17 +1,47 @@ +# Pleroma: A lightweight social networking server +# Copyright © 2017-2019 Pleroma Authors +# SPDX-License-Identifier: AGPL-3.0-only + defmodule Pleroma.Web.RichMedia.Parser do - @parsers [Pleroma.Web.RichMedia.Parsers.OGP] + @parsers [ + Pleroma.Web.RichMedia.Parsers.OGP, + Pleroma.Web.RichMedia.Parsers.TwitterCard, + Pleroma.Web.RichMedia.Parsers.OEmbed + ] + + @hackney_options [ + pool: :media, + timeout: 2_000, + recv_timeout: 2_000, + max_body: 2_000_000 + ] + + def parse(nil), do: {:error, "No URL provided"} if Mix.env() == :test do def parse(url), do: parse_url(url) else - def parse(url), - do: {:commit, Cachex.fetch!(:rich_media_cache, url, fn _ -> parse_url(url) end)} + def parse(url) do + 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: @hackney_options) - 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 @@ -23,11 +53,23 @@ defmodule Pleroma.Web.RichMedia.Parser 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 clean_parsed_data(data) do + data + |> Enum.reject(fn {key, val} -> + with {:ok, _} <- Jason.encode(%{key => val}) do + false + else + _ -> true + end + end) + |> Map.new() end end