From 012691279d213f249c167fd5af5850ac5e91a54e Mon Sep 17 00:00:00 2001
From: "Yukihiro \"Matz\" Matsumoto" <matz@ruby.or.jp>
Date: Tue, 26 May 2026 23:10:45 +0900
Subject: [PATCH] mruby-string-ext: skip scrub UTF-8 assertions on non-UTF-8
 builds
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When MRB_UTF8_STRING is undefined, String#scrub is a no-op that
returns the receiver as-is (the #else branch added in ccb62ceb57).
The unit tests assumed UTF-8 semantics unconditionally, so a build
of just mruby-string-ext without UTF-8 strings failed all five scrub
tests.

Guard the UTF-8-dependent assertions with `skip unless "あ".length
== 1`, and add a paired test that asserts the no-op behaviour on the
non-UTF-8 build (skipped on UTF-8 builds).

Verified against the build config from the report:
  - UTF-8 (host-debug): 5 tests pass, 1 skip
  - non-UTF-8 (noutf8):  5 skip, 1 test pass

Closes #6860.

Co-authored-by: Claude <noreply@anthropic.com>
---
 mrbgems/mruby-string-ext/test/string.rb | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/mrbgems/mruby-string-ext/test/string.rb b/mrbgems/mruby-string-ext/test/string.rb
index c76a7ef15..c0fd381db 100644
--- a/mrbgems/mruby-string-ext/test/string.rb
+++ b/mrbgems/mruby-string-ext/test/string.rb
@@ -793,6 +793,9 @@ assert('String#-@') do
 end
 
 assert('String#scrub default replacement (U+FFFD)') do
+  # scrub has UTF-8 semantics; on builds without MRB_UTF8_STRING it
+  # degrades to a no-op (verified separately below).
+  skip unless "あ".length == 1
   assert_equal "\u{FFFD}",       "\xE3\x81".scrub
   assert_equal "abc\u{FFFD}def", "abc\x80def".scrub
   assert_equal "\u{FFFD}",       "\x80\x81\x82".scrub   # run collapsed
@@ -802,6 +805,7 @@ assert('String#scrub default replacement (U+FFFD)') do
 end
 
 assert('String#scrub rejects malformed sequences') do
+  skip unless "あ".length == 1
   # overlong, UTF-16 surrogate, codepoint above U+10FFFF
   assert_equal "\u{FFFD}", "\xC0\xAF".scrub             # overlong "/"
   assert_equal "\u{FFFD}", "\xED\xA0\x80".scrub         # surrogate U+D800
@@ -809,16 +813,19 @@ assert('String#scrub rejects malformed sequences') do
 end
 
 assert('String#scrub with replacement string') do
+  skip unless "あ".length == 1
   assert_equal "abc?def", "abc\x80def".scrub("?")
   assert_equal "abcdef",  "abc\x80def".scrub("")
   assert_equal "abc<bad>def", "abc\x80def".scrub("<bad>")
 end
 
 assert('String#scrub raises on invalid replacement') do
+  skip unless "あ".length == 1
   assert_raise(ArgumentError) { "abc\x80".scrub("\xFF") }
 end
 
 assert('String#scrub with block') do
+  skip unless "あ".length == 1
   assert_equal "abc<80>def",
                "abc\x80def".scrub { |b| "<" + b.bytes.first.to_s(16) + ">" }
   # Block not called when string is already valid
@@ -833,3 +840,11 @@ assert('String#scrub with block') do
   # explicit and doesn't drift accidentally.
   assert_equal "abc42def", "abc\x80def".scrub { 42 }
 end
+
+assert('String#scrub is a no-op without MRB_UTF8_STRING') do
+  skip if "あ".length == 1
+  # Method is still defined and returns a (string-equal) copy.
+  assert_equal "abc\x80def", "abc\x80def".scrub
+  assert_equal "abc\x80def", "abc\x80def".scrub("?")
+  assert_equal "abc\x80def", "abc\x80def".scrub { |_| "?" }
+end