recovery us-ascii encoding

lib/feed_reader/recovery/encoding.ex

@@ -8,14 +8,14 @@ defmodule FeedReader.Recovery.Encoding do
   to UTF-8, including declarations split across input chunks.
   """
 
-  def transcode(input, :utf8), do: input
+  def transcode(input, encoding) when encoding in [:utf8, :ascii], do: input
 
   def transcode(input, :latin1),
     do: Stream.map(input, &:unicode.characters_to_binary(&1, :latin1, :utf8))
 
   def normalize_declaration(input, :utf8), do: input
 
-  def normalize_declaration(input, :latin1) do
+  def normalize_declaration(input, encoding) when encoding in [:latin1, :ascii] do
     Stream.transform(input, fn -> {:prolog, <<>>} end, &rewrite_chunk/2, &flush/1)
   end
 

lib/feed_reader/recovery/recovery.ex

@@ -10,7 +10,7 @@ defmodule FeedReader.Recovery do
   """
 
   @doc "Returns the encoding declared by an XML prefix, defaulting to UTF-8."
-  @spec declared_encoding(binary()) :: :utf8 | :latin1
+  @spec declared_encoding(binary()) :: :utf8 | :latin1 | :ascii
   def declared_encoding(prefix) do
     case Regex.run(~r/<\?xml\b[^>]*\bencoding\s*=\s*["']([^"']+)["']/is, prefix,
            capture: :all_but_first
@@ -34,6 +34,7 @@ defmodule FeedReader.Recovery do
   defp normalize_encoding(encoding) do
     case String.downcase(encoding) do
       value when value in ["iso-8859-1", "latin1", "latin-1"] -> :latin1
+      "us-ascii" -> :ascii
       _utf8 -> :utf8
     end
   end

test/feed_reader/recovery/encoding_test.exs

@@ -2,6 +2,7 @@ defmodule FeedReader.Recovery.EncodingTest do
   use ExUnit.Case, async: true
 
   alias FeedReader.Recovery.Encoding
+  alias FeedReader.Recovery
 
   test "passes UTF-8 input through unchanged" do
     chunks = ["<title>", "café", "</title>"]
@@ -29,4 +30,15 @@ defmodule FeedReader.Recovery.EncodingTest do
     assert chunks |> Encoding.normalize_declaration(:latin1) |> Enum.join() ==
              "<?xml version=\"1.0\" encoding=\"UTF-8\"?><rss/>"
   end
+
+  test "rewrites a US-ASCII declaration to UTF-8" do
+    chunks = ["<?xml version=\"1.0\" encoding=\"us-ascii\"?>", "<rss/>"]
+
+    assert chunks |> Encoding.normalize_declaration(:ascii) |> Enum.join() ==
+             "<?xml version=\"1.0\" encoding=\"UTF-8\"?><rss/>"
+  end
+
+  test "recognizes US-ASCII declarations" do
+    assert Recovery.declared_encoding("<?xml encoding=\"us-ascii\"?>") == :ascii
+  end
 end