X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=c8a767935333c50fae6037789866bcee49e1299f;hb=e1a1c8e7de5e10fa64d168dc5d35a80b96767395;hp=9deb038451e2e8a8cdd63af7e891653a426017cf;hpb=514c899275a32e6ef63305f9424c50344d41b12e;p=akkoma

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 9deb03845..c8a767935 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,5 +1,5 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parser do
@@ -86,33 +86,36 @@ defmodule Pleroma.Web.RichMedia.Parser do
       end
 
     try do
-      {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: opts)
+      rich_media_agent = Pleroma.Application.user_agent() <> "; Bot"
+
+      {:ok, %Tesla.Env{body: html}} =
+        Pleroma.HTTP.get(url, [{"user-agent", rich_media_agent}], 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
 
@@ -123,11 +126,7 @@ 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