merge revision(s) a8ba1dd: [Backport #19455]

nagachika · nagachika · commit be09d77b966c · 2023-07-17T17:23:31.000+09:00
Use UTF-8 encoding for literal extended regexps with UTF-8 characters
	 in comments

	Fixes [Bug #19455]
	---
	 re.c                     | 9 ++++++++-
	 test/ruby/test_regexp.rb | 7 +++++++
	 2 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/re.c b/re.c
@@ -2926,7 +2926,11 @@ unescape_nonascii0(const char **pp, const char *end, rb_encoding *enc,
           case '#':
             if (extended_mode && !in_char_class) {
                 /* consume and ignore comment in extended regexp */
-                while ((p < end) && ((c = *p++) != '\n'));
+                while ((p < end) && ((c = *p++) != '\n')) {
+                    if ((c & 0x80) && !*encp && enc == rb_utf8_encoding()) {
+                        *encp = enc;
+                    }
+                }
                 break;
             }
             rb_str_buf_cat(buf, (char *)&c, 1);
@@ -2961,6 +2965,9 @@ unescape_nonascii0(const char **pp, const char *end, rb_encoding *enc,
                         switch (c = *p++) {
                           default:
                             if (!(c & 0x80)) break;
+                            if (!*encp && enc == rb_utf8_encoding()) {
+                                *encp = enc;
+                            }
                             --p;
                             /* fallthrough */
                           case '\\':
diff --git a/test/ruby/test_regexp.rb b/test/ruby/test_regexp.rb
@@ -200,6 +200,13 @@ def test_nonextended_section_of_extended_regexp_bug_19379
     RUBY
   end
 
+  def test_utf8_comment_in_usascii_extended_regexp_bug_19455
+    assert_separately([], <<-RUBY)
+      assert_equal(Encoding::UTF_8, /(?#\u1000)/x.encoding)
+      assert_equal(Encoding::UTF_8, /#\u1000/x.encoding)
+    RUBY
+  end
+
   def test_union
     assert_equal :ok, begin
       Regexp.union(
diff --git 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 90
+#define RUBY_PATCHLEVEL 91
 
 #include "ruby/version.h"
 #include "ruby/internal/abi.h"
