X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=40980def8198d134496ed429a3efd494352c4c95;hb=dbcfac11b45b367185a3b18a2db3e3fb07e8f20d;hp=185156375fa36ed69d7d277a76c9a81d96b8f5e4;hpb=33681747857eec90ff56ea0342d2ea179c4f856e;p=akkoma diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index 185156375..40980def8 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -1,13 +1,11 @@ # Pleroma: A lightweight social networking server -# Copyright © 2017-2019 Pleroma Authors +# Copyright © 2017-2020 Pleroma Authors # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.Web.RichMedia.Parser do - @hackney_options [ + @options [ pool: :media, - recv_timeout: 2_000, - max_body: 2_000_000, - with_body: true + max_body: 2_000_000 ] defp parsers do @@ -77,19 +75,33 @@ defmodule Pleroma.Web.RichMedia.Parser do end defp parse_url(url) do + opts = + if Application.get_env(:tesla, :adapter) == Tesla.Adapter.Hackney do + Keyword.merge(@options, + recv_timeout: 2_000, + with_body: true + ) + else + @options + end + try do - {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options) + {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: opts) html + |> parse_html() |> maybe_parse() + |> Map.put(:url, url) |> clean_parsed_data() |> check_parsed_data() rescue e -> - {:error, "Parsing error: #{inspect(e)}"} + {:error, "Parsing error: #{inspect(e)} #{inspect(__STACKTRACE__)}"} end end + defp parse_html(html), do: Floki.parse_document!(html) + defp maybe_parse(html) do Enum.reduce_while(parsers(), %{}, fn parser, acc -> case parser.parse(html, acc) do @@ -99,7 +111,8 @@ defmodule Pleroma.Web.RichMedia.Parser do end) end - defp check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do + defp check_parsed_data(%{title: title} = data) + when is_binary(title) and byte_size(title) > 0 do {:ok, data} end