From b31a934a804aed3f35442ceafe2080b0955e7317 Mon Sep 17 00:00:00 2001
From: dcc <dcc@logografos.com>
Date: Wed, 15 May 2024 00:57:23 -0700
Subject:  total rebase

---
 priv/scrubbers/search_indexing.ex | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
 create mode 100644 priv/scrubbers/search_indexing.ex

(limited to 'priv/scrubbers/search_indexing.ex')

diff --git a/priv/scrubbers/search_indexing.ex b/priv/scrubbers/search_indexing.ex
new file mode 100644
index 0000000..02756ab
--- /dev/null
+++ b/priv/scrubbers/search_indexing.ex
@@ -0,0 +1,24 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.HTML.Scrubber.SearchIndexing do
+  @moduledoc """
+  An HTML scrubbing policy that scrubs things for searching.
+  """
+
+  require FastSanitize.Sanitizer.Meta
+  alias FastSanitize.Sanitizer.Meta
+
+  # Explicitly remove mentions
+  def scrub({:a, attrs, children}) do
+    if(Enum.any?(attrs, fn {att, val} -> att == "class" and String.contains?(val, "mention") end),
+      do: nil,
+      # Strip the tag itself, leave only children (text, presumably)
+      else: children
+    )
+  end
+
+  Meta.strip_comments()
+  Meta.strip_everything_not_covered()
+end
-- 
cgit v1.2.3