[ruby/yarp] For errored % characters, use TOKEN_PERCENT and not TOKEN_STRING_BEGIN

https://github.com/ruby/yarp/commit/2ec671fe1e

1 files changed, 6 insertions, 4 deletions

diff --git a/yarp/yarp.c b/yarp/yarp.c
index 9aa4e264c6..ee9aa1b172 100644
--- a/yarp/yarp.c
+++ b/yarp/yarp.c
@@ -6325,12 +6325,14 @@ parser_lex(yp_parser_t *parser) {
 
                 // % %= %i %I %q %Q %w %W
                 case '%': {
-                    // In a BEG state, if you encounter a % then you must be starting
-                    // something. In this case if there is no subsequent character then
-                    // we have an invalid token.
+                    // In a BEG state, if you encounter a % then you must be
+                    // starting something. In this case if there is no
+                    // subsequent character then we have an invalid token. We're
+                    // going to say it's the percent operator because we don't
+                    // want to move into the string lex mode unnecessarily.
                     if (lex_state_beg_p(parser) && (parser->current.end >= parser->end)) {
                         yp_diagnostic_list_append(&parser->error_list, parser->current.start, parser->current.end, "unexpected end of input");
-                        LEX(YP_TOKEN_STRING_BEGIN);
+                        LEX(YP_TOKEN_PERCENT);
                     }
 
                     if (!lex_state_beg_p(parser) && match(parser, '=')) {