X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=0d25233388a67ff977bbee8622bab2393ab944e9;hb=f4447d82b814e4710a0d7499bc0707773ac1e440;hp=6da83c6e4ab34265be75c1a623980aedd0d022c4;hpb=fd2f1258fb0249694d1ed7f5ba5a2f4baa858f20;p=akkoma diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index 6da83c6e4..0d2523338 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -1,25 +1,52 @@ +# 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, - Pleroma.Web.RichMedia.Parsers.TwitterCard, - Pleroma.Web.RichMedia.Parsers.OEmbed + @hackney_options [ + pool: :media, + recv_timeout: 2_000, + max_body: 2_000_000, + with_body: true ] - if Mix.env() == :test do + defp parsers do + Pleroma.Config.get([:rich_media, :parsers]) + end + + def parse(nil), do: {:error, "No URL provided"} + + if Pleroma.Config.get(:env) == :test do def parse(url), do: parse_url(url) else - def parse(url), - do: 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 - Enum.reduce_while(@parsers, %{}, fn parser, acc -> + Enum.reduce_while(parsers(), %{}, fn parser, acc -> case parser.parse(html, acc) do {:ok, data} -> {:halt, data} {:error, _msg} -> {:cont, acc} @@ -27,11 +54,23 @@ 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 get_parsed_data(data) do - {:ok, data} + defp check_parsed_data(data) do + {:error, "Found metadata was invalid or incomplete: #{inspect(data)}"} + end + + 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