X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=62e8fa610bbf6937877eae6c26071aea41013c6e;hb=a58813e7a14fda8eff2f2a659094444d20765e85;hp=d9c1684d58449a80771c2a3f8282d781955167f3;hpb=2aab4e03c3a2867abd4555dc776eebc8b0dba176;p=akkoma diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index d9c1684d5..62e8fa610 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -1,14 +1,75 @@ +# 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, + recv_timeout: 2_000, + max_body: 2_000_000, + with_body: true + ] + + 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 + try do + Cachex.fetch!(:rich_media_cache, url, fn _ -> + {:commit, parse_url(url)} + end) + rescue + e -> + {:error, "Cachex error: #{inspect(e)}"} + end + end + end - def parse(url) do - {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url) + defp parse_url(url) do + try do + {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options) + + html |> maybe_parse() |> clean_parsed_data() |> check_parsed_data() + rescue + e -> + {:error, "Parsing error: #{inspect(e)}"} + end + end - Enum.reduce_while(@parsers, %Pleroma.Web.RichMedia.Data{}, fn parser, acc -> + defp maybe_parse(html) do + Enum.reduce_while(@parsers, %{}, fn parser, acc -> case parser.parse(html, acc) do {:ok, data} -> {:halt, data} {:error, _msg} -> {:cont, acc} end end) end + + 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 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