Update encode() to prevent double-encode existing entities

lib/html_entities.ex

@@ -24,7 +24,7 @@ defmodule HtmlEntities do
   @external_resource "lib/html_entities_list.txt"
 
   @doc "Decode HTML entities in a string."
-  @spec decode(String.t) :: String.t
+  @spec decode(String.t()) :: String.t()
   def decode(string) do
     decode(string, "")
   end
@@ -67,17 +67,37 @@ defmodule HtmlEntities do
   defp decode_entity(_), do: :error
 
   @doc "Encode HTML entities in a string."
-  @spec encode(String.t) :: String.t
+  @spec encode(String.t()) :: String.t()
   def encode(string) do
-    for <<x <- string>>, into: "" do
-      case x do
-        ?' -> "&apos;"
-        ?" -> "&quot;"
-        ?& -> "&amp;"
-        ?< -> "&lt;"
-        ?> -> "&gt;"
-        _ -> <<x>>
-      end
+    encode(string, "")
+  end
+
+  defp encode(<<?&, rest::binary>> = entity, acc) do
+    case entity do
+      <<head::bytes-size(4), remaining::binary>> when head in ["&lt;", "&gt;"] ->
+        encode(remaining, acc <> head)
+
+      <<?&, ?a, ?m, ?p, ?;, remaining::binary>> ->
+        encode(remaining, acc <> "&amp;")
+
+      <<head::bytes-size(6), remaining::binary>> when head in ["&quot;", "&apos;"] ->
+        encode(remaining, acc <> head)
+
+      _ ->
+        encode(rest, acc <> encode_entity(?&))
     end
   end
+
+  defp encode(<<head, rest::binary>>, acc) do
+    encode(rest, acc <> encode_entity(head))
+  end
+
+  defp encode(<<>>, acc), do: acc
+
+  defp encode_entity(?'), do: "&apos;"
+  defp encode_entity(?"), do: "&quot;"
+  defp encode_entity(?&), do: "&amp;"
+  defp encode_entity(?<), do: "&lt;"
+  defp encode_entity(?>), do: "&gt;"
+  defp encode_entity(other), do: <<other>>
 end

test/html_entities_test.exs

@@ -25,4 +25,20 @@ defmodule HtmlEntitiesTest do
   test "Encoding does replace unsafe characters" do
     assert encode("'\"&<>") == "&apos;&quot;&amp;&lt;&gt;"
   end
+
+  test "Encoding when existing escaped characters exist" do
+    assert encode("this has both unsafe characters &<> and an existing &amp; entity") ==
+             "this has both unsafe characters &amp;&lt;&gt; and an existing &amp; entity"
+  end
+
+  test "Multiple Encodings of the same string will consistent with one single encoding pass" do
+    original_string = "this has both unsafe characters &<> and an existing &amp; entity"
+
+    first_pass = encode(original_string)
+    second_pass = encode(first_pass)
+    third_pass = encode(second_pass)
+
+    assert first_pass == second_pass
+    assert third_pass == second_pass
+  end
 end