X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=ef5ead2daf1fdcad1309591344223c07513f5136;hb=61180ab6f4b85ab78de2eaf1bc1b974c9e7908af;hp=c06b0a0f2668338415d592fde838bff42e49e056;hpb=a0f101ee806af06bcd4271cd8d57d11ff85ea11a;p=akkoma diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index c06b0a0f2..ef5ead2da 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 @@ -77,34 +75,44 @@ 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 + |> parse_html() |> maybe_parse() - |> Map.put(:url, url) + |> 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(html) + 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 - {:ok, data} -> {:halt, data} - {:error, _msg} -> {:cont, acc} + data when data != %{} -> {:halt, data} + _ -> {:cont, acc} end 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 title != "" do {:ok, data} end @@ -115,11 +123,7 @@ defmodule Pleroma.Web.RichMedia.Parser do defp clean_parsed_data(data) do data |> Enum.reject(fn {key, val} -> - with {:ok, _} <- Jason.encode(%{key => val}) do - false - else - _ -> true - end + not match?({:ok, _}, Jason.encode(%{key => val})) end) |> Map.new() end