From 1ff55bb09dca302d42951059a73e6d237fd8c338 Mon Sep 17 00:00:00 2001
From: Takashi Kokubun <takashikkbn@gmail.com>
Date: Tue, 4 Jun 2024 14:58:37 -0700
Subject: merge revision(s) 05553cf22d43dd78b8f30cc4591230b5c000c538: [Backport
 #20517]

	[Bug #20517] Make a multibyte character one token at meta escape
---
 parse.y                   |  6 +++++-
 test/ripper/test_lexer.rb | 54 ++++++++++++++++++++++++++++++++++++++++++++---
 version.h                 |  2 +-
 3 files changed, 57 insertions(+), 5 deletions(-)

diff --git a/parse.y b/parse.y
index 8755904b9a..dd96d6136e 100644
--- a/parse.y
+++ b/parse.y
@@ -8047,7 +8047,11 @@ read_escape(struct parser_params *p, int flags)
             }
             return read_escape(p, flags|ESCAPE_META) | 0x80;
         }
-        else if (c == -1 || !ISASCII(c)) goto eof;
+        else if (c == -1) goto eof;
+        else if (!ISASCII(c)) {
+            tokskip_mbchar(p);
+            goto eof;
+        }
         else {
             int c2 = escaped_control_code(c);
             if (c2) {
diff --git a/test/ripper/test_lexer.rb b/test/ripper/test_lexer.rb
index 7bff858705..65d4ce3b5b 100644
--- a/test/ripper/test_lexer.rb
+++ b/test/ripper/test_lexer.rb
@@ -302,9 +302,8 @@ world"
       [[6, 2], :on_tstring_content, "3\n", state(:EXPR_BEG)],
       [[7, 0], :on_heredoc_end, "H1\n", state(:EXPR_BEG)],
     ]
-    assert_equal(code, Ripper.tokenize(code).join(""))
-    assert_equal(expected, result = Ripper.lex(code),
-                 proc {expected.zip(result) {|e, r| break diff(e, r) unless e == r}})
+
+    assert_lexer(expected, code)
 
     code = <<~'HEREDOC'
     <<-H1
@@ -330,6 +329,55 @@ world"
       [[6, 0], :on_tstring_content, "  3\n", state(:EXPR_BEG)],
       [[7, 0], :on_heredoc_end, "H1\n", state(:EXPR_BEG)],
     ]
+
+    assert_lexer(expected, code)
+  end
+
+  def test_invalid_escape_ctrl_mbchar
+    code = %["\\C-\u{3042}"]
+    expected = [
+      [[1, 0], :on_tstring_beg, '"', state(:EXPR_BEG)],
+      [[1, 1], :on_tstring_content, "\\C-\u{3042}", state(:EXPR_BEG)],
+      [[1, 7], :on_tstring_end, '"', state(:EXPR_END)],
+    ]
+
+    assert_lexer(expected, code)
+  end
+
+  def test_invalid_escape_meta_mbchar
+    code = %["\\M-\u{3042}"]
+    expected = [
+      [[1, 0], :on_tstring_beg, '"', state(:EXPR_BEG)],
+      [[1, 1], :on_tstring_content, "\\M-\u{3042}", state(:EXPR_BEG)],
+      [[1, 7], :on_tstring_end, '"', state(:EXPR_END)],
+    ]
+
+    assert_lexer(expected, code)
+  end
+
+  def test_invalid_escape_meta_ctrl_mbchar
+    code = %["\\M-\\C-\u{3042}"]
+    expected = [
+      [[1, 0], :on_tstring_beg, '"', state(:EXPR_BEG)],
+      [[1, 1], :on_tstring_content, "\\M-\\C-\u{3042}", state(:EXPR_BEG)],
+      [[1, 10], :on_tstring_end, '"', state(:EXPR_END)],
+    ]
+
+    assert_lexer(expected, code)
+  end
+
+  def test_invalid_escape_ctrl_meta_mbchar
+    code = %["\\C-\\M-\u{3042}"]
+    expected = [
+      [[1, 0], :on_tstring_beg, '"', state(:EXPR_BEG)],
+      [[1, 1], :on_tstring_content, "\\C-\\M-\u{3042}", state(:EXPR_BEG)],
+      [[1, 10], :on_tstring_end, '"', state(:EXPR_END)],
+    ]
+
+    assert_lexer(expected, code)
+  end
+
+  def assert_lexer(expected, code)
     assert_equal(code, Ripper.tokenize(code).join(""))
     assert_equal(expected, result = Ripper.lex(code),
                  proc {expected.zip(result) {|e, r| break diff(e, r) unless e == r}})
diff --git a/version.h b/version.h
index 5da189ff33..af217e3158 100644
--- a/version.h
+++ b/version.h
@@ -11,7 +11,7 @@
 # define RUBY_VERSION_MINOR RUBY_API_VERSION_MINOR
 #define RUBY_VERSION_TEENY 2
 #define RUBY_RELEASE_DATE RUBY_RELEASE_YEAR_STR"-"RUBY_RELEASE_MONTH_STR"-"RUBY_RELEASE_DAY_STR
-#define RUBY_PATCHLEVEL 85
+#define RUBY_PATCHLEVEL 86
 
 #include "ruby/version.h"
 #include "ruby/internal/abi.h"
-- 
cgit v1.2.3