From bb4ced0eb5cdf8744adea2da9c268734a6170060 Mon Sep 17 00:00:00 2001
From: Mark Felder <feld@feld.me>
Date: Fri, 11 Jun 2021 15:58:55 -0500
Subject: [PATCH] scrub_html_and_truncate/1 -> filter_html_and_truncate/1

They shouldn't share the same name when /1 was used for a different type of incoming data anyway
---
 lib/pleroma/web/metadata/utils.ex        |  7 ++++---
 test/pleroma/web/metadata/utils_test.exs | 26 +++++++++++---------------
 2 files changed, 15 insertions(+), 18 deletions(-)

diff --git a/lib/pleroma/web/metadata/utils.ex b/lib/pleroma/web/metadata/utils.ex
index 649319db5..f5b79a82c 100644
--- a/lib/pleroma/web/metadata/utils.ex
+++ b/lib/pleroma/web/metadata/utils.ex
@@ -7,7 +7,7 @@ defmodule Pleroma.Web.Metadata.Utils do
   alias Pleroma.Formatter
   alias Pleroma.HTML
 
-  def scrub_html_and_truncate(%{data: %{"content" => content}} = _object) do
+  def filter_html_and_truncate(%{data: %{"content" => content}} = _object) do
     content
     # html content comes from DB already encoded, decode first and scrub after
     |> Emoji.Formatter.demojify()
@@ -20,9 +20,10 @@ defmodule Pleroma.Web.Metadata.Utils do
   def scrub_html_and_truncate(content, max_length \\ 200) when is_binary(content) do
     content
     |> Emoji.Formatter.demojify()
-    |> HTML.filter_tags(Pleroma.HTML.Scrubber.BreaksOnly)
     |> HtmlEntities.decode()
-    |> String.replace(~r/<br\s?\/?>/, "&#10;&#13;")
+    |> String.replace(~r/<br\s?\/?>/, " ")
+    |> HTML.strip_tags()
+    |> HtmlEntities.decode()
     |> Formatter.truncate(max_length)
   end
 
diff --git a/test/pleroma/web/metadata/utils_test.exs b/test/pleroma/web/metadata/utils_test.exs
index c5d241886..9d94bf9ca 100644
--- a/test/pleroma/web/metadata/utils_test.exs
+++ b/test/pleroma/web/metadata/utils_test.exs
@@ -7,8 +7,8 @@ defmodule Pleroma.Web.Metadata.UtilsTest do
   import Pleroma.Factory
   alias Pleroma.Web.Metadata.Utils
 
-  describe "scrub_html_and_truncate" do
-    test "it returns text without encode HTML (objects)" do
+  describe "filter_html_and_truncate/1" do
+    test "it returns text without HTML" do
       user = insert(:user)
 
       note =
@@ -20,10 +20,10 @@ defmodule Pleroma.Web.Metadata.UtilsTest do
           }
         })
 
-      assert Utils.scrub_html_and_truncate(note) == "Pleroma's really cool!"
+      assert Utils.filter_html_and_truncate(note) == "Pleroma's really cool!"
     end
 
-    test "it replaces <br> with compatible HTML entity (objects)" do
+    test "it replaces <br> with compatible HTML entity (meta tags, push notifications)" do
       user = insert(:user)
 
       note =
@@ -35,31 +35,27 @@ defmodule Pleroma.Web.Metadata.UtilsTest do
           }
         })
 
-      assert Utils.scrub_html_and_truncate(note) ==
+      assert Utils.filter_html_and_truncate(note) ==
                "First line&#10;&#13;Second line"
     end
+  end
 
-    test "it returns text without encode HTML (binaries)" do
+  describe "scrub_html_and_truncate/2" do
+    test "it returns text without encode HTML" do
       assert Utils.scrub_html_and_truncate("Pleroma's really cool!") == "Pleroma's really cool!"
     end
 
-    test "it truncates to specified chars (binaries)" do
+    test "it truncates to specified chars" do
       assert Utils.scrub_html_and_truncate("Pleroma's really cool!", 10) == "Pleroma..."
     end
 
-    # push notifications and link previews should be able to display newlines
-    test "it replaces <br> with compatible HTML entity (binaries)" do
-      assert Utils.scrub_html_and_truncate("First line<br>Second line") ==
-               "First line&#10;&#13;Second line"
-    end
-
-    test "it strips emojis (binaries)" do
+    test "it strips emojis" do
       assert Utils.scrub_html_and_truncate(
                "Open the door get on the floor everybody walk the dinosaur :dinosaur:"
              ) == "Open the door get on the floor everybody walk the dinosaur"
     end
 
-    test "it strips HTML tags and other entities (binaries)" do
+    test "it strips HTML tags and other entities" do
       assert Utils.scrub_html_and_truncate("<title>my title</title> <p>and a paragraph&#33;</p>") ==
                "my title and a paragraph!"
     end