Renderer content disposition encoding

lib/mail/parsers/rfc_2822.ex

@@ -947,11 +947,11 @@ defmodule Mail.Parsers.RFC2822 do
     |> Enum.map(fn
       # Find parameters that are split into multiple parts
       {key, value} when is_binary(value) ->
-        case String.split(key, "*", parts: 2) do
+        case String.split(key, "*", parts: 2, trim: true) do
           [key, part] ->
             {{key, part}, value}
 
-          _ ->
+          [key] ->
             {key, value}
         end
 

lib/mail/renderers/rfc_2822.ex

@@ -118,6 +118,15 @@ defmodule Mail.Renderers.RFC2822 do
     render_header_value(key, value)
   end
 
+  @rfc2231_headers ["Content-Disposition", "Content-Type"]
+  defp render_header_value(key, value) when key in @rfc2231_headers and is_binary(value) do
+    value
+  end
+
+  defp render_header_value(key, [value | subtypes]) when key in @rfc2231_headers do
+    Enum.join([value | render_subtypes(subtypes, :rfc_2231)], "; ")
+  end
+
   defp render_header_value(_key, [value | subtypes]),
     do:
       Enum.join([encode_header_value(value, :quoted_printable) | render_subtypes(subtypes)], "; ")
@@ -143,27 +152,46 @@ defmodule Mail.Renderers.RFC2822 do
 
   defp render_address(email), do: validate_address(email)
 
-  defp render_subtypes([]), do: []
+  defp render_subtypes(subtypes, encoding \\ :quoted_printable)
+
+  defp render_subtypes([], _encoding), do: []
 
-  defp render_subtypes([{key, value} | subtypes]) when is_atom(key),
-    do: render_subtypes([{Atom.to_string(key), value} | subtypes])
+  defp render_subtypes([{key, value} | subtypes], encoding) when is_atom(key),
+    do: render_subtypes([{Atom.to_string(key), value} | subtypes], encoding)
 
-  defp render_subtypes([{"boundary", value} | subtypes]) do
-    [~s(boundary="#{value}") | render_subtypes(subtypes)]
+  defp render_subtypes([{"boundary", value} | subtypes], encoding) do
+    [~s(boundary="#{value}") | render_subtypes(subtypes, encoding)]
   end
 
-  defp render_subtypes([{key, value} | subtypes]) do
+  # RFC 2231 parameter encoding for Content-Type and Content-Disposition
+  defp render_subtypes([{key, value} | subtypes], :rfc_2231) do
     key = String.replace(key, "_", "-")
-    value = encode_header_value(value, :quoted_printable)
 
-    value =
-      if value =~ ~r/[\s()<>@,;:\\<\/\[\]?=]/ do
-        "\"#{value}\""
-      else
-        value
-      end
+    if contains_non_ascii?(value) do
+      value = encode_header_value(value, :rfc_2231)
+      ["#{key}*=UTF-8''#{value}" | render_subtypes(subtypes, :rfc_2231)]
+    else
+      value = maybe_wrap_in_quotes(value)
+      ["#{key}=#{value}" | render_subtypes(subtypes, :rfc_2231)]
+    end
+  end
 
-    ["#{key}=#{value}" | render_subtypes(subtypes)]
+  defp render_subtypes([{key, value} | subtypes], :quoted_printable) do
+    key = String.replace(key, "_", "-")
+    value = value |> encode_header_value(:quoted_printable) |> maybe_wrap_in_quotes()
+    ["#{key}=#{value}" | render_subtypes(subtypes, :quoted_printable)]
+  end
+
+  defp contains_non_ascii?(value) do
+    String.match?(value, ~r/[\x80-\xFF]/)
+  end
+
+  defp maybe_wrap_in_quotes(value) do
+    if value =~ ~r/[\s()<>@,;:\\<\/\[\]?=]/ do
+      "\"#{value}\""
+    else
+      value
+    end
   end
 
   @doc """
@@ -198,6 +226,11 @@ defmodule Mail.Renderers.RFC2822 do
     end
   end
 
+  defp encode_header_value(header_value, :rfc_2231) do
+    # RFC 2231: parameter*=UTF-8''percent-encoded-value
+    URI.encode(header_value, &URI.char_unreserved?/1)
+  end
+
   defp wrap_encoded_words(value) do
     :binary.split(value, "=\r\n", [:global])
     |> Enum.map(fn chunk -> <<"=?UTF-8?Q?", chunk::binary, "?=">> end)

test/mail/message_test.exs

@@ -262,13 +262,17 @@ defmodule Mail.MessageTest do
       |> Mail.put_attachment({file_name, "data"}, headers: [content_id: "attachment-id"])
       |> Mail.render()
 
-    encoded_header_value =
-      "=?UTF-8?Q?" <> Mail.Encoders.QuotedPrintable.encode("READMEüä.md") <> "?="
+    refute String.contains?(message, "=?UTF-8?Q?")
 
-    assert String.contains?(message, encoded_header_value)
+    assert String.contains?(message, "filename*=UTF-8''README%C3%BC%C3%A4.md")
 
     assert %Mail.Message{
-             headers: %{"content-disposition" => ["attachment", {"filename", ^file_name}]}
+             headers: %{
+               "content-disposition" => [
+                 "attachment",
+                 {"filename", "UTF-8''README%C3%BC%C3%A4.md"}
+               ]
+             }
            } = Mail.Parsers.RFC2822.parse(message)
   end
 

test/mail/renderers/rfc_2822_test.exs

@@ -367,6 +367,51 @@ defmodule Mail.Renderers.RFC2822Test do
     end
   end
 
+  describe "RFC 2047 encoding restrictions" do
+    test "Subject header uses RFC 2047 encoding for non-ASCII characters" do
+      header = Mail.Renderers.RFC2822.render_header("subject", "café")
+      assert header == "Subject: =?UTF-8?Q?caf=C3=A9?="
+
+      header = Mail.Renderers.RFC2822.render_header("subject", "Test 日本語 Subject")
+      assert header =~ "Subject: =?UTF-8?Q?"
+      assert header =~ "?="
+    end
+
+    test "From header uses RFC 2047 encoding for non-ASCII names" do
+      header = Mail.Renderers.RFC2822.render_header("from", {"José García", "jose@example.com"})
+      assert header == ~s(From: =?UTF-8?Q?"Jos=C3=A9 Garc=C3=ADa"?= <jose@example.com>)
+
+      header = Mail.Renderers.RFC2822.render_header("from", {"山田太郎", "yamada@example.com"})
+      assert header =~ "From: =?UTF-8?Q?"
+      assert header =~ "?= <yamada@example.com>"
+    end
+
+    test "Content-Disposition non-ASCII filename parameter uses RFC 2231 encoding" do
+      # RFC 2047 explicitly forbids encoded-words in Content-Disposition parameters
+      header =
+        Mail.Renderers.RFC2822.render_header(
+          "Content-Disposition",
+          ["attachment", filename: "café.pdf"]
+        )
+
+      # Should NOT contain RFC 2047 encoded-word markers
+      # Should use RFC 2231 encoding for non-ASCII characters
+      assert header == "Content-Disposition: attachment; filename*=UTF-8''caf%C3%A9.pdf"
+    end
+
+    test "Content-Disposition ASCII filename parameter does NOT use RFC 2047 encoding" do
+      # RFC 2047 explicitly forbids encoded-words in Content-Disposition parameters
+      header =
+        Mail.Renderers.RFC2822.render_header(
+          "Content-Disposition",
+          ["attachment", filename: "name.pdf"]
+        )
+
+      # Should NOT contain RFC 2047 encoded-word markers
+      assert header == "Content-Disposition: attachment; filename=name.pdf"
+    end
+  end
+
   describe "multipart configuration" do
     test "multipart/alternative with text/plain and text/html" do
       message =