Page MenuHomePhorge
Files F85826693

No OneTemporary
Actions

Authored By
Unknown
Size
2 KB
Referenced Files
None
Subscribers
None

View Options

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index e67ecc47d..b33411cc7 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,56 +1,76 @@
# 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,
Pleroma.Web.RichMedia.Parsers.TwitterCard,
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
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
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
Enum.reduce_while(@parsers, %{}, fn parser, acc ->
case parser.parse(html, acc) do
{:ok, data} -> {:halt, data}
{:error, _msg} -> {:cont, acc}
end
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) 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

File Metadata

Mime Type
text/x-diff
Expires
Fri, Oct 30, 9:12 AM (18 h, 5 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1799947
Default Alt Text
(2 KB)

Event Timeline