Do not rely on cached stripped content. We need the control to preserve the breaks.

2 files changed, 21 insertions, 6 deletions

diff --git a/lib/pleroma/web/metadata/utils.ex b/lib/pleroma/web/metadata/utils.ex
index a519fbf01..649319db5 100644
--- a/lib/pleroma/web/metadata/utils.ex
+++ b/lib/pleroma/web/metadata/utils.ex
@@ -3,19 +3,17 @@
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.Metadata.Utils do
-  alias Pleroma.Activity
   alias Pleroma.Emoji
   alias Pleroma.Formatter
   alias Pleroma.HTML
 
-  def scrub_html_and_truncate(%{data: %{"content" => content}} = object) do
+  def scrub_html_and_truncate(%{data: %{"content" => content}} = _object) do
     content
     # html content comes from DB already encoded, decode first and scrub after
-    |> HtmlEntities.decode()
-    |> String.replace(~r/<br\s?\/?>/, " ")
-    |> Activity.HTML.get_cached_stripped_html_for_activity(object, "metadata")
     |> Emoji.Formatter.demojify()
+    |> HTML.filter_tags(Pleroma.HTML.Scrubber.BreaksOnly)
     |> HtmlEntities.decode()
+    |> String.replace(~r/<br\s?\/?>/, "&#10;&#13;")
     |> Formatter.truncate()
   end
 
diff --git a/test/pleroma/web/metadata/utils_test.exs b/test/pleroma/web/metadata/utils_test.exs
index b5d3e3566..c5d241886 100644
--- a/test/pleroma/web/metadata/utils_test.exs
+++ b/test/pleroma/web/metadata/utils_test.exs
@@ -23,6 +23,22 @@ defmodule Pleroma.Web.Metadata.UtilsTest do
       assert Utils.scrub_html_and_truncate(note) == "Pleroma's really cool!"
     end
 
+    test "it replaces <br> with compatible HTML entity (objects)" do
+      user = insert(:user)
+
+      note =
+        insert(:note, %{
+          data: %{
+            "actor" => user.ap_id,
+            "id" => "https://pleroma.gov/objects/whatever",
+            "content" => "First line<br>Second line"
+          }
+        })
+
+      assert Utils.scrub_html_and_truncate(note) ==
+               "First line&#10;&#13;Second line"
+    end
+
     test "it returns text without encode HTML (binaries)" do
       assert Utils.scrub_html_and_truncate("Pleroma's really cool!") == "Pleroma's really cool!"
     end
@@ -44,7 +60,8 @@ defmodule Pleroma.Web.Metadata.UtilsTest do
     end
 
     test "it strips HTML tags and other entities (binaries)" do
-      assert Utils.scrub_html_and_truncate("<title>my title</title> <p>and a paragraph&#33;</p>") == "my title and a paragraph!"
+      assert Utils.scrub_html_and_truncate("<title>my title</title> <p>and a paragraph&#33;</p>") ==
+               "my title and a paragraph!"
     end
   end
 end