Fix Floki.text/2 when document contains a doctype declaration (#702)

foxbenjaminfox · web-flow · commit 2db3aa31054d · 2026-06-13T14:54:14.000-03:00
This fixes #701. The fix is analogous to that in #696.
diff --git a/lib/floki/deep_text.ex b/lib/floki/deep_text.ex
@@ -44,6 +44,7 @@ defmodule Floki.DeepText do
   defp get_text({"script", _, _}, acc, _, _, false, _), do: acc
   defp get_text({"style", _, _}, acc, _, _, _, false), do: acc
   defp get_text({:pi, _, _}, acc, _, _, _, _), do: acc
+  defp get_text({:doctype, _, _, _}, acc, _, _, _, _), do: acc
 
   defp get_text({"input", attrs, _}, acc, _, true, _, _) do
     [acc, Floki.TextExtractor.extract_input_value(attrs)]
diff --git a/test/floki/deep_text_test.exs b/test/floki/deep_text_test.exs
@@ -112,4 +112,10 @@ defmodule Floki.DeepTextTest do
 
     assert Floki.DeepText.get(nodes) == "foo"
   end
+
+  test "HTML that contains a doctype (<!doctype html>)" do
+    nodes = [{:doctype, "html", "", ""}, "foo"]
+
+    assert Floki.DeepText.get(nodes) == "foo"
+  end
 end
