recover a wrong closed tag

lib/feed_reader/parser.ex

@@ -337,15 +337,99 @@ defmodule FeedReader.Parser do
   defp flush_entities({_mode, pending}), do: [pending]
 
   defp normalize_attribute_quotes(input) do
-    Stream.map(input, fn chunk ->
-      Regex.replace(
-        ~r/(=\")([^\"]*)(\")([^\"]*)(\")(\")(?=\s*(?:\/>|>))/,
-        chunk,
-        "\\1\\2"\\4"\\6"
-      )
-    end)
+    Stream.transform(
+      input,
+      fn -> <<>> end,
+      fn chunk, pending ->
+        {output, pending} = scan_attribute_markup(pending <> chunk, [])
+        {if(output == [], do: [], else: [IO.iodata_to_binary(Enum.reverse(output))]), pending}
+      end,
+      fn
+        <<>> -> []
+        pending -> [pending]
+      end
+    )
   end
 
+  defp scan_attribute_markup(<<>>, output), do: {output, <<>>}
+
+  defp scan_attribute_markup(data, output) do
+    case :binary.match(data, "<") do
+      :nomatch ->
+        {[data | output], <<>>}
+
+      {position, 1} when position > 0 ->
+        {prefix, rest} = :erlang.split_binary(data, position)
+        scan_attribute_markup(rest, [prefix | output])
+
+      {0, 1} ->
+        case markup_end(data) do
+          :incomplete -> {output, data}
+          {markup, rest} -> scan_attribute_markup(rest, [repair_markup_quotes(markup) | output])
+        end
+    end
+  end
+
+  defp markup_end(<<"<![CDATA[", _rest::binary>> = data), do: special_markup_end(data, "]]>")
+  defp markup_end(<<"<!--", _rest::binary>> = data), do: special_markup_end(data, "-->")
+
+  defp markup_end(data) do
+    case :binary.match(data, ">") do
+      :nomatch -> :incomplete
+      {position, 1} -> :erlang.split_binary(data, position + 1)
+    end
+  end
+
+  defp special_markup_end(data, closing) do
+    case :binary.match(data, closing) do
+      :nomatch -> :incomplete
+      {position, length} -> :erlang.split_binary(data, position + length)
+    end
+  end
+
+  defp repair_markup_quotes(<<"<![CDATA[", _rest::binary>> = markup), do: markup
+  defp repair_markup_quotes(<<"<!--", _rest::binary>> = markup), do: markup
+
+  defp repair_markup_quotes(<<"<", rest::binary>>),
+    do: "<" <> repair_tag_quotes(rest, nil, false, [])
+
+  defp repair_tag_quotes(<<>>, _quote, _expect, output),
+    do: IO.iodata_to_binary(Enum.reverse(output))
+
+  defp repair_tag_quotes(<<char, rest::binary>>, nil, expect, output)
+       when char in [32, 9, 10, 13],
+       do: repair_tag_quotes(rest, nil, expect, [<<char>> | output])
+
+  defp repair_tag_quotes(<<char, rest::binary>>, nil, true, output)
+       when char in [34, 39],
+       do: repair_tag_quotes(rest, {char, false}, false, [<<char>> | output])
+
+  defp repair_tag_quotes(<<char, rest::binary>>, nil, _expect, output),
+    do: repair_tag_quotes(rest, nil, char == ?=, [<<char>> | output])
+
+  defp repair_tag_quotes(<<quote, rest::binary>>, {quote, has_content}, _expect, output) do
+    cond do
+      delimiter?(rest) ->
+        repair_tag_quotes(rest, nil, false, [<<quote>> | output])
+
+      empty_attribute_quote?(rest, quote, has_content) ->
+        <<_quote, after_quote::binary>> = rest
+        repair_tag_quotes(after_quote, nil, false, [<<quote>> | output])
+
+      true ->
+        repair_tag_quotes(rest, {quote, true}, false, ["&quot;" | output])
+    end
+  end
+
+  defp repair_tag_quotes(<<char, rest::binary>>, {quote, _has_content}, _expect, output),
+    do: repair_tag_quotes(rest, {quote, true}, false, [<<char>> | output])
+
+  defp delimiter?(<<char, _rest::binary>>) when char in [32, 9, 10, 13, 47, 62, 63], do: true
+  defp delimiter?(_rest), do: false
+
+  defp empty_attribute_quote?(<<quote, rest::binary>>, quote, false), do: delimiter?(rest)
+  defp empty_attribute_quote?(_rest, _quote, _has_content), do: false
+
   defp event_stream(input, format, opts, chunk_size) do
     Stream.resource(
       fn -> start_parser(input, format, opts, chunk_size) end,

test/feed_reader/collector_test.exs

@@ -91,6 +91,14 @@ defmodule FeedReader.CollectorTest do
              FeedReader.parse_binary(xml, max_items: 1)
   end
 
+  test "does not repair quotes across adjacent XML elements" do
+    xml =
+      ~s(<rss version="2.0"><channel><title>Example</title><item><source url="">Source</source><category domain="region">News</category></item></channel></rss>)
+
+    assert {:ok, %ParseResult{items: [%Item{title: nil}]}} =
+             FeedReader.parse_binary(xml, max_items: 1, chunk_size: 1)
+  end
+
   test "rejects invalid chunks and binary arguments" do
     assert {:error, :invalid_chunk} =
              FeedReader.parse_stream([@xml, 123], max_items: 2)