Always use optimal encoding function (#512)

lib/floki/entities.ex

@@ -62,15 +62,51 @@ defmodule Floki.Entities do
   * greater-than sign - > - is replaced by ">".
 
   All other simbols are going to remain the same.
+
+  Optimized IO data implementation from Plug.HTML
   """
-  @spec encode(String.t()) :: String.t()
-  def encode(string) when is_binary(string) do
-    String.replace(string, ["'", "\"", "&", "<", ">"], fn
-      "'" -> "&#39;"
-      "\"" -> "&quot;"
-      "&" -> "&amp;"
-      "<" -> "&lt;"
-      ">" -> "&gt;"
-    end)
+  @spec encode(iodata()) :: iodata()
+  def encode(string) when is_binary(string), do: encode(string, 0, string, [])
+  def encode(data), do: encode(IO.iodata_to_binary(data))
+
+  escapes = [
+    {?<, "&lt;"},
+    {?>, "&gt;"},
+    {?&, "&amp;"},
+    {?", "&quot;"},
+    {?', "&#39;"}
+  ]
+
+  for {match, insert} <- escapes do
+    defp encode(<<unquote(match), rest::bits>>, skip, original, acc) do
+      encode(rest, skip + 1, original, [acc | unquote(insert)])
+    end
+  end
+
+  defp encode(<<_char, rest::bits>>, skip, original, acc) do
+    encode(rest, skip, original, acc, 1)
+  end
+
+  defp encode(<<>>, _skip, _original, acc) do
+    acc
+  end
+
+  for {match, insert} <- escapes do
+    defp encode(<<unquote(match), rest::bits>>, skip, original, acc, len) do
+      part = binary_part(original, skip, len)
+      encode(rest, skip + len + 1, original, [acc, part | unquote(insert)])
+    end
+  end
+
+  defp encode(<<_char, rest::bits>>, skip, original, acc, len) do
+    encode(rest, skip, original, acc, len + 1)
+  end
+
+  defp encode(<<>>, 0, original, _acc, _len) do
+    original
+  end
+
+  defp encode(<<>>, skip, original, acc, len) do
+    [acc | binary_part(original, skip, len)]
   end
 end

lib/floki/raw_html.ex

@@ -131,11 +131,7 @@ defmodule Floki.RawHTML do
   end
 
   defp build_attrs({attr, value}, encoder) do
-    if encoder == @encoder do
-      [attr, "=\"", html_escape(value) | "\""]
-    else
-      [attr, "=\"", value | "\""]
-    end
+    [attr, "=\"", encoder.(value) | "\""]
   end
 
   defp build_attrs(attr, _encoder), do: attr
@@ -164,53 +160,6 @@ defmodule Floki.RawHTML do
     end
   end
 
-  # html_escape
-  # Optimized IO data implementation from Plug.HTML
-
-  defp html_escape(data) when is_binary(data), do: html_escape(data, 0, data, [])
-  defp html_escape(data), do: html_escape(IO.iodata_to_binary(data))
-
-  escapes = [
-    {?<, "&lt;"},
-    {?>, "&gt;"},
-    {?&, "&amp;"},
-    {?", "&quot;"},
-    {?', "&#39;"}
-  ]
-
-  for {match, insert} <- escapes do
-    defp html_escape(<<unquote(match), rest::bits>>, skip, original, acc) do
-      html_escape(rest, skip + 1, original, [acc | unquote(insert)])
-    end
-  end
-
-  defp html_escape(<<_char, rest::bits>>, skip, original, acc) do
-    html_escape(rest, skip, original, acc, 1)
-  end
-
-  defp html_escape(<<>>, _skip, _original, acc) do
-    acc
-  end
-
-  for {match, insert} <- escapes do
-    defp html_escape(<<unquote(match), rest::bits>>, skip, original, acc, len) do
-      part = binary_part(original, skip, len)
-      html_escape(rest, skip + len + 1, original, [acc, part | unquote(insert)])
-    end
-  end
-
-  defp html_escape(<<_char, rest::bits>>, skip, original, acc, len) do
-    html_escape(rest, skip, original, acc, len + 1)
-  end
-
-  defp html_escape(<<>>, 0, original, _acc, _len) do
-    original
-  end
-
-  defp html_escape(<<>>, skip, original, acc, len) do
-    [acc | binary_part(original, skip, len)]
-  end
-
   # helpers
 
   # TODO: Use Enum.map_intersperse/3 when we require Elixir v1.10+

test/floki/entities_test.exs

@@ -5,23 +5,23 @@ defmodule Floki.EntitiesTest do
 
   describe "encode/1" do
     test "encode single-quote" do
-      assert Entities.encode("'") == "'"
+      assert IO.iodata_to_binary(Entities.encode("'")) == "'"
     end
 
     test "encode double-quote" do
-      assert Entities.encode("\"") == """
+      assert IO.iodata_to_binary(Entities.encode("\"")) == """
     end
 
     test "ampersand" do
-      assert Entities.encode("&") == "&"
+      assert IO.iodata_to_binary(Entities.encode("&")) == "&"
     end
 
     test "encode less-than sign" do
-      assert Entities.encode("<") == "&lt;"
+      assert IO.iodata_to_binary(Entities.encode("<")) == "&lt;"
     end
 
     test "encode greater-than sign" do
-      assert Entities.encode(">") == "&gt;"
+      assert IO.iodata_to_binary(Entities.encode(">")) == "&gt;"
     end
 
     test "does not encode others" do