mruby-string-ext: skip four UTF-8 tests where they cannot run - #7139

Four tests in mrbgems/mruby-string-ext/test/string.rb hold their MRB_UTF8_STRING guard inside the assertion block rather than on the assert call:

assert('String#chop! on a binary string removes one byte') do
  if UTF8STRING
    s = "\u{1F600}".b
    s.chop!
    assert_equal "\xF0\x9F\x98".b, s
    ...
  end
end

On a build without MRB_UTF8_STRING the block still runs and reaches no assertion, which mrbtest counts as a warning:

Warn: String#inspect of a binary string escapes every byte => no assertion
Warn: String#chop! on a binary string removes one byte => no assertion
Warn: String#rindex on a binary string counts bytes => no assertion
Warn: a needle that spells no character is found nowhere => no assertion
...
Warning: 4

MRB_UTF8_STRING is turned on by mruby-encoding, which the default gembox does not carry, so a plain rake test is one of the builds that reports this.

The rest of the file already guards its UTF-8 tests on the assert call, so this PR moves these four to the same form:

assert('String#chop! on a binary string removes one byte') do
  s = "\u{1F600}".b
  s.chop!
  assert_equal "\xF0\x9F\x98".b, s
  ...
end if UTF8STRING

A build that cannot run the test no longer defines it. The assertions themselves are unchanged, and the diff is the removal of the inner if plus one level of indentation.

The guard is needed in all four cases, since each asserts behaviour that only a UTF-8 build has: "る".inspect comes back with every byte escaped without it, chop! cuts a byte instead of a character, and a needle such as "\x81" is searched for as a byte rather than refused as a sequence that spells no character.

Testing

rake test with the default gembox (MRB_UTF8_STRING off):

The four tests that ran without asserting anything are the only ones that disappear; nothing that used to assert stops asserting.

A full-core build (MRB_UTF8_STRING on) still runs all four: Total: 2269 / OK: 2259 / KO: 0 / Warning: 0.

Summary by CodeRabbit

• Tests
  • Simplified conditional handling for UTF-8-specific string tests.
  • Preserved existing coverage for string inspection, binary chop!, binary rindex, and malformed-needle behavior.