aboutsummaryrefslogtreecommitdiff
path: root/lib/pleroma/web/rich_media/parser.ex
diff options
context:
space:
mode:
Diffstat (limited to 'lib/pleroma/web/rich_media/parser.ex')
-rw-r--r--lib/pleroma/web/rich_media/parser.ex28
1 files changed, 25 insertions, 3 deletions
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index e67ecc47d..32dec9887 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -30,7 +30,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
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)}"}
@@ -46,11 +46,33 @@ defmodule Pleroma.Web.RichMedia.Parser do
end)
end
- defp get_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
- defp get_parsed_data(data) do
+ defp check_parsed_data(data) do
{:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
end
+
+ defp string_is_valid_unicode(data) when is_binary(data) do
+ data
+ |> :unicode.characters_to_binary()
+ |> clean_string()
+ end
+
+ defp string_is_valid_unicode(data), do: {:ok, data}
+
+ 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