Update handling of spaces between nodes

Fixes #14

lib/html_sanitize_ex/parser.ex

@@ -10,7 +10,8 @@ defmodule HtmlSanitizeEx.Parser do  @doc """
   @type html_tree :: tuple | list
 
   @my_root_node "html_sanitize_ex"
-  @linebreak [239, 188, 191]
+  @replacement_linebreak [239, 188, 191]
+  @replacement_space [239, 189, 191]
 
   @spec parse(binary) :: html_tree
 
@@ -22,7 +23,9 @@ defmodule HtmlSanitizeEx.Parser do  @doc """
   end
 
   defp before_parse(html) do
-    String.replace(html, ~r/(>)(\r?\n)/, "\\1 #{@linebreak} \\2")
+    html
+    |> String.replace(~r/(>)(\r?\n)/, "\\1 #{@replacement_linebreak} \\2")
+    |> String.replace(~r/(>)(\ +)(<)/, "\\1 #{@replacement_space}\\2\\3")
   end
 
   def to_html(tokens) do
@@ -36,7 +39,9 @@ defmodule HtmlSanitizeEx.Parser do  @doc """
   end
 
   defp after_to_html(html) do
-    String.replace(html, ~r/(\ ?#{@linebreak} )(\r?\n)/, "\\2")
+    html
+    |> String.replace(~r/(\ ?#{@replacement_linebreak} )(\r?\n)/, "\\2")
+    |> String.replace(~r/(\&gt\;|>)(\ +)(#{@replacement_space})(\ +)(\&lt\;|<)/, "\\1\\4\\5")
   end
 
   defp ensure_list(list) do

lib/html_sanitize_ex/scrubber.ex

@@ -9,15 +9,9 @@ defmodule HtmlSanitizeEx.Scrubber do
 
   def scrub(html, scrubber_module) do
     html
-    |> before_scrub
     |> scrubber_module.before_scrub
     |> HtmlSanitizeEx.Parser.parse
     |> HtmlSanitizeEx.Traverser.traverse(scrubber_module)
     |> HtmlSanitizeEx.Parser.to_html
   end
-
-  defp before_scrub(html) do
-    html
-    |> String.replace(~r/(>)(\ +)(<)/, "\\1&#32;\\3")
-  end
 end

test/html5_test.exs

@@ -66,4 +66,10 @@ defmodule HtmlSanitizeExScrubberHTML5Test do
     expected = ~s(<a href="mailto:[email protected]">Email Us</a>)
     assert expected == full_html_sanitize(input)
   end
+
+  test "does encode script in textarea, but preserves white-space" do
+    input = ~s(<textarea> <script></script></textarea>)
+    expected = ~s(<textarea> &lt;script&gt;&lt;/script&gt;</textarea>)
+    assert expected == full_html_sanitize(input)
+  end
 end