X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=b33411cc7a1d95f67aa43942a24ab8fde36c23da;hb=dafb6f0b5efac3997080f6b164dfb35515f31b24;hp=6da83c6e4ab34265be75c1a623980aedd0d022c4;hpb=de1da7b3d1c855df87262b8d24d6cdb6a6a518bc;p=akkoma

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 6da83c6e4..b33411cc7 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,3 +1,7 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
 defmodule Pleroma.Web.RichMedia.Parser do
   @parsers [
     Pleroma.Web.RichMedia.Parsers.OGP,
@@ -5,17 +9,32 @@ defmodule Pleroma.Web.RichMedia.Parser do
     Pleroma.Web.RichMedia.Parsers.OEmbed
   ]
 
+  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: 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: [pool: :media])
 
-    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
@@ -27,11 +46,31 @@ 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 string_is_valid_unicode(data) do
+    data
+    |> :unicode.characters_to_binary()
+    |> clean_string()
+  end
+
+  defp clean_string({:error, _, _}), do: {:error, "Invalid data"}
+  defp clean_string(data), do: {:ok, data}
+
+  defp clean_parsed_data(data) do
+    data
+    |> Enum.reject(fn {_, val} ->
+      case string_is_valid_unicode(val) do
+        {:ok, _} -> false
+        _ -> true
+      end
+    end)
+    |> Map.new()
   end
 end