Merge branch 'truncate-rich-media' into 'develop'

Truncate the length of Rich Media title and description fields

See merge request pleroma/pleroma!4343

changelog.d/truncate-rich-media.change

@@ -0,0 +1 @@
+Truncate the length of Rich Media title and description fields

lib/pleroma/web/rich_media/parser.ex

@@ -4,6 +4,7 @@
 
 defmodule Pleroma.Web.RichMedia.Parser do
   alias Pleroma.Web.RichMedia.Helpers
+  import Pleroma.Web.Metadata.Utils, only: [scrub_html_and_truncate: 2]
   require Logger
 
   @config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
@@ -63,8 +64,20 @@ defmodule Pleroma.Web.RichMedia.Parser do
       not match?({:ok, _}, Jason.encode(%{key => val}))
     end)
     |> Map.new()
+    |> truncate_title()
+    |> truncate_desc()
   end
 
+  defp truncate_title(%{"title" => title} = data) when is_binary(title),
+    do: %{data | "title" => scrub_html_and_truncate(title, 120)}
+
+  defp truncate_title(data), do: data
+
+  defp truncate_desc(%{"description" => desc} = data) when is_binary(desc),
+    do: %{data | "description" => scrub_html_and_truncate(desc, 200)}
+
+  defp truncate_desc(data), do: data
+
   @spec validate_page_url(URI.t() | binary()) :: :ok | :error
   defp validate_page_url(page_url) when is_binary(page_url) do
     validate_tld = @config_impl.get([Pleroma.Formatter, :validate_tld])

test/pleroma/web/rich_media/parser_test.exs

@@ -61,6 +61,13 @@ defmodule Pleroma.Web.RichMedia.ParserTest do
               }}
   end
 
+  test "truncates title and description fields" do
+    {:ok, parsed} = Parser.parse("https://instagram.com/longtext")
+
+    assert String.length(parsed["title"]) == 120
+    assert String.length(parsed["description"]) == 200
+  end
+
   test "parses OEmbed and filters HTML tags" do
     assert Parser.parse("https://example.com/oembed") ==
              {:ok,

test/support/http_request_mock.ex

@@ -1494,6 +1494,11 @@ defmodule HttpRequestMock do
     {:ok, %Tesla.Env{status: 200, body: File.read!("test/fixtures/rich_media/twitter_card.html")}}
   end
 
+  def get("https://instagram.com/longtext", _, _, _) do
+    {:ok,
+     %Tesla.Env{status: 200, body: File.read!("test/fixtures/rich_media/instagram_longtext.html")}}
+  end
+
   def get("https://example.com/non-ogp", _, _, _) do
     {:ok,
      %Tesla.Env{status: 200, body: File.read!("test/fixtures/rich_media/non_ogp_embed.html")}}
@@ -1720,7 +1725,8 @@ defmodule HttpRequestMock do
     "https://example.com/twitter-card",
     "https://google.com/",
     "https://pleroma.local/notice/9kCP7V",
-    "https://yahoo.com/"
+    "https://yahoo.com/",
+    "https://instagram.com/longtext"
   ]
 
   def head(url, _query, _body, _headers) when url in @rich_media_mocks do