Fixing a bug where UTF-8 encodings resulted in incorrect char offset

 calculations.
Co-authored-by: Vinicius Stock <vinistock@users.noreply.github.com>

Co-authored-by: Vinicius Stock <vinistock@users.noreply.github.com>

Author: ChallaHalla

lib/ruby_lsp/document.rb

@@ -176,7 +176,8 @@ class Scanner
       def initialize(source, encoding)
         @current_line = 0 #: Integer
         @pos = 0 #: Integer
-        @source = source.codepoints #: Array[Integer]
+        @source = encoding == Encoding::UTF_8 ? source.bytes : source.codepoints #: Array[Integer]
+        @text = source
         @encoding = encoding
       end
 
@@ -200,7 +201,14 @@ def find_char_position(position)
 
         # The final position is the beginning of the line plus the requested column. If the encoding is UTF-16, we also
         # need to adjust for surrogate pairs
-        requested_position = @pos + position[:character]
+        requested_position = if @encoding == Encoding::UTF_8
+          character_offset = @text.byteslice(@pos, position[:character]) #: as !nil
+            .length
+          @pos + character_offset
+
+        else
+          @pos + position[:character]
+        end
 
         if @encoding == Encoding::UTF_16LE
           requested_position -= utf_16_character_position_correction(@pos, requested_position)

test/ruby_document_test.rb

@@ -59,6 +59,20 @@ def foo
     RUBY
   end
 
+  def test_multibyte_character_offsets_are_bytes_in_utf8
+    document = RubyLsp::RubyDocument.new(source: +<<~RUBY, version: 1, uri: @uri, global_state: @global_state)
+      bá
+    RUBY
+
+    document.push_edits(
+      [{ range: { start: { line: 0, character: 3 }, end: { line: 0, character: 3 } }, text: "r" }], version: 2
+    )
+
+    assert_equal(<<~RUBY, document.source)
+      bár
+    RUBY
+  end
+
   def test_deletion_full_node
     document = RubyLsp::RubyDocument.new(source: +<<~RUBY, version: 1, uri: @uri, global_state: @global_state)
       def foo