Style/FrozenStringLiteralComment - invalid byte sequence in UTF-8

[snutij]

Hello 👋 I have a tricky issue, with Style/FrozenStringLiteralComment. Regex give me nightmare, so can't help more than providing as much as possible information. Do not hesitate if you need something else!

Given a file foo.rb:

"\xE3\xD3\x8B\xE3\x83\xBC\x83\xE3\x83\xE3\x82\xB3\xA3\x82\x99" \
        "\xE3\x83\xB3\xE3\x83\x8F\xE3\x82\x9A\xC3\xBD;foo@bar.com"

Which is a valid syntax:

ruby -c foo.rb
Syntax OK

---

Expected behavior

Should not trigger an error. And probably propose a fix to add the comment.

Actual behavior

bin/rubocop -d --only Style/FrozenStringLiteralComment foo.rb

An error occurred while Style/FrozenStringLiteralComment cop was inspecting foo.rb.
invalid byte sequence in UTF-8
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/style/frozen_string_literal_comment.rb:145:in `match?'
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/style/frozen_string_literal_comment.rb:145:in `last_special_comment'
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/style/frozen_string_literal_comment.rb:195:in `insert_comment'
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/style/frozen_string_literal_comment.rb:159:in `block in missing_offense'
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/base.rb:377:in `correct'
/Users/snutij/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/rubocop-1.54.1/lib/rubocop/cop/base.rb:181:in `add_offense'

Steps to reproduce the problem

In Style::FrozenStringLiteralComment the method last_special_comment receive an argument processed_source,

(ruby) processed_source.raw_source
"\"\\xE3\\xD3\\x8B\\xE3\\x83\\xBC\\x83\\xE3\\x83\\xE3\\x82\\xB3\\xA3\\x82\\x99\" \\\n        \"\\xE3\\x83\\xB3\\xE3\\x83\\x8F\\xE3\\x82\\x9A\\xC3\\xBD;foo@bar.com\"\n"

(ruby) processed_source.ast
s(:dstr,
  s(:str, "\xE3Ӌー\x83\xE3\x83コ\xA3\x82\x99"),
  s(:str, "ンパý;foo@bar.com"))

Its raised an error on Regex match check:

(ruby) next_token
#<RuboCop::AST::Token:0x00000001078413b0
 @pos=#<Parser::Source::Range /Users/justinrabiller/Documents/www/ruby-lsp/test/requests/diagnostics_expectations_test.rb 0...62>,
 @text="\xE3Ӌー\x83\xE3\x83コ\xA3\x82\x99",
 @type=:tSTRING>

(ruby) next_token&.text
"\xE3Ӌー\x83\xE3\x83コ\xA3\x82\x99"

(ruby) Encoding::ENCODING_PATTERN
/#.*coding\s?[:=]\s?(?:UTF|utf)-8/

(ruby) Encoding::ENCODING_PATTERN.match?(next_token&.text)
eval error: invalid byte sequence in UTF-8

rubocop/lib/rubocop/cop/style/frozen_string_literal_comment.rb

Lines 137 to 148 in 1a13d7a

	def last_special_comment(processed_source)
	token_number = 0
	if processed_source.tokens[token_number].text.start_with?(SHEBANG)
	token = processed_source.tokens[token_number]
	token_number += 1
	end
	next_token = processed_source.tokens[token_number]
	token = next_token if Encoding::ENCODING_PATTERN.match?(next_token&.text)
	token
	end

RuboCop version

bundle exec rubocop -V

1.54.1 (using Parser 3.2.2.3, rubocop-ast 1.29.0, running on ruby 3.2.2) [arm64-darwin22]
  - rubocop-minitest 0.31.0
  - rubocop-rake 0.6.0
  - rubocop-sorbet 0.7.0