🐛 Use literal syntax for invalid UTF8 or non-UTF8 strings

lib/net/imap/command_data.rb

@@ -58,8 +58,8 @@ def send_data(data, tag = nil)
       end
     end
 
-    UNQUOTABLE_CHARS = /\0\r\n/n
-    ASTRING_SPECIALS = /[(){ \x00-\x1f\x7f%*"\\]/n
+    UNQUOTABLE_CHARS = /\0\r\n/
+    ASTRING_SPECIALS = /[(){ \x00-\x1f\x7f%*"\\]/
     private_constant :UNQUOTABLE_CHARS, :ASTRING_SPECIALS
 
     # Sends generic strings formatted as +astring+:
@@ -72,22 +72,25 @@ def send_string_data(str, tag = nil)
       if str.empty?
         # same as send_quoted_string(str), but incompatible encoding is allowed
         put_string('""')
-      elsif str.match?(UNQUOTABLE_CHARS)
-        send_literal(str, tag)
-      elsif !str.ascii_only?
-        if @utf8_strings
-          send_quoted_string(str)
-        else
-          send_literal(str, tag)
-        end
-      elsif str.match?(ASTRING_SPECIALS)
-        send_quoted_string(str)
-      else
+      elsif str.ascii_only? && !str.match?(ASTRING_SPECIALS)
         # valid +astring+ atom: non-empty, ASCII only, no ASTRING_SPECIALS
         put_string(str)
+      elsif text_encodable?(str) && !str.match?(UNQUOTABLE_CHARS)
+        send_quoted_string(str)
+      else
+        send_literal(str, tag)
       end
     end
 
+    # Encodable as +text+ (which is a superset of +quoted+):
+    # * ASCII only (for any ASCII compatible encoding)
+    # * or valid UTF-8 (when the connection supports it)
+    def text_encodable?(str)
+      str.ascii_only? || (@utf8_strings &&
+                          str.encoding == Encoding::UTF_8 &&
+                          str.valid_encoding?)
+    end
+
     def send_quoted_string(str) = QuotedString.new(data: str).send_data(self)
 
     def send_binary_literal(*, **) = send_literal(*, **, binary: true)

test/net/imap/test_imap.rb

@@ -994,18 +994,42 @@ def imap.test_args(*args) = send_command("TEST", *args)
       assert_equal("sync-literal-utf8 ({10+}\r\nαβγδε)".b,
                    server.commands.pop.args)
 
+      imap.test_args "utf8-with-wrong-encoding", "αβγδε".b
+      assert_equal("utf8-with-wrong-encoding {10+}\r\nαβγδε".b,
+                   server.commands.pop.args)
+
+      imap.test_args "invalid-utf8", "\x80".b.force_encoding("UTF-8")
+      assert_equal("invalid-utf8 {1+}\r\n\x80".b,
+                   server.commands.pop.args)
+
       # Before enabling UTF-8 strings, without non-synchronizing literals
       imap.config.max_non_synchronizing_literal = -1
       imap.test_args "sync-literal-utf8", ["αβγδε"]
       assert_equal("sync-literal-utf8 ({10}\r\nαβγδε)".b,
                    server.commands.pop.args)
 
+      imap.test_args "utf8-with-wrong-encoding", "αβγδε".b
+      assert_equal("utf8-with-wrong-encoding {10}\r\nαβγδε".b,
+                   server.commands.pop.args)
+
+      imap.test_args "invalid-utf8", "\x80".b.force_encoding("UTF-8")
+      assert_equal("invalid-utf8 {1}\r\n\x80".b,
+                   server.commands.pop.args)
+
       # After enabling UTF-8 strings
       imap.enable(:utf8)
       server.commands.pop.args => ["UTF8=ACCEPT"]
 
       imap.test_args "quoted-utf8", "αβγδε"
       assert_equal 'quoted-utf8 "αβγδε"'.b, server.commands.pop.args
+
+      imap.test_args "utf8-with-wrong-encoding", "αβγδε".b
+      assert_equal("utf8-with-wrong-encoding {10}\r\nαβγδε".b,
+                   server.commands.pop.args)
+
+      imap.test_args "invalid-utf8", "\x80".b.force_encoding("UTF-8")
+      assert_equal("invalid-utf8 {1}\r\n\x80".b,
+                   server.commands.pop.args)
     end
   end