Fix an incorrect autocorrect for Performance/RedundantStringChars

Follow up to rubocop/rubocop-performance#249 (comment).

This PR fixes an incorrect autocorrect for `Performance/RedundantStringChars`
when using `str.chars[0]`.

changelog/fix_incorrect_autocorrect_for_performance_redundant_string_chars.md

@@ -0,0 +1 @@
+* [#273](https://github.com/rubocop/rubocop-performance/pull/273): Fix an incorrect autocorrect for `Performance/RedundantStringChars` when using `str.chars[0]`. ([@koic][])

lib/rubocop/cop/performance/redundant_string_chars.rb

@@ -82,21 +82,29 @@ def build_message(method, args)
 
         def build_good_method(method, args)
           case method
-          when :[], :slice
+          when :slice
             "[#{build_call_args(args)}].chars"
-          when :first
-            if args.any?
-              "[0...#{args.first.source}].chars"
-            else
-              '[0]'
-            end
+          when :[], :first
+            build_good_method_for_brackets_or_first_method(method, args)
           when :take
             "[0...#{args.first.source}].chars"
           else
             ".#{method}"
           end
         end
 
+        def build_good_method_for_brackets_or_first_method(method, args)
+          first_arg = args.first
+
+          if first_arg&.range_type?
+            "[#{build_call_args(args)}].chars"
+          elsif method == :first && args.any?
+            "[0...#{args.first.source}].chars"
+          else
+            first_arg ? "[#{first_arg.source}]" : '[0]'
+          end
+        end
+
         def build_bad_method(method, args)
           case method
           when :[]

spec/rubocop/cop/performance/redundant_string_chars_spec.rb

@@ -34,6 +34,28 @@
     RUBY
   end
 
+  it 'registers an offense and corrects when using `str.chars[0]`' do
+    expect_offense(<<~RUBY)
+      str.chars[0]
+          ^^^^^^^^ Use `[0]` instead of `chars[0]`.
+    RUBY
+
+    expect_correction(<<~RUBY)
+      str[0]
+    RUBY
+  end
+
+  it 'registers an offense and corrects when using `str.chars[42]`' do
+    expect_offense(<<~RUBY)
+      str.chars[42]
+          ^^^^^^^^^ Use `[42]` instead of `chars[42]`.
+    RUBY
+
+    expect_correction(<<~RUBY)
+      str[42]
+    RUBY
+  end
+
   it 'registers an offense and corrects when using `str.chars.first(2)`' do
     expect_offense(<<~RUBY)
       str.chars.first(2)