[ruby/prism] Fix parser translator during string escaping with invalid utf-8

Instead, prefer `scan_byte` over `get_byte` since that already returns the byte as an integer, sidestepping conversion issues.

Fixes ruby/prism#3582

ruby/prism@7f3008b2b5

Author: Earlopain

lib/prism/polyfill/scan_byte.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+require "strscan"
+
+# Polyfill for StringScanner#scan_byte, which didn't exist until Ruby 3.4.
+if !(StringScanner.instance_methods.include?(:scan_byte))
+  StringScanner.include(
+    Module.new {
+      def scan_byte # :nodoc:
+        get_byte&.b&.ord
+      end
+    }
+  )
+end

lib/prism/prism.gemspec

@@ -88,6 +88,7 @@ Gem::Specification.new do |spec|
     "lib/prism/pattern.rb",
     "lib/prism/polyfill/append_as_bytes.rb",
     "lib/prism/polyfill/byteindex.rb",
+    "lib/prism/polyfill/scan_byte.rb",
     "lib/prism/polyfill/unpack1.rb",
     "lib/prism/polyfill/warn.rb",
     "lib/prism/reflection.rb",

lib/prism/translation/parser/lexer.rb

@@ -3,6 +3,7 @@
 
 require "strscan"
 require_relative "../../polyfill/append_as_bytes"
+require_relative "../../polyfill/scan_byte"
 
 module Prism
   module Translation
@@ -762,12 +763,12 @@ def escape_read(result, scanner, control, meta)
           elsif (value = scanner.scan(/M-\\?(?=[[:print:]])/))
             # \M-x where x is an ASCII printable character
             escape_read(result, scanner, control, true)
-          elsif (byte = scanner.get_byte)
+          elsif (byte = scanner.scan_byte)
             # Something else after an escape.
-            if control && byte == "?"
+            if control && byte == 0x3f # ASCII '?'
               result.append_as_bytes(escape_build(0x7f, false, meta))
             else
-              result.append_as_bytes(escape_build(byte.ord, control, meta))
+              result.append_as_bytes(escape_build(byte, control, meta))
             end
           end
         end

test/prism/fixtures/strings.txt

@@ -146,6 +146,10 @@ baz
 
 %Q{abc}
 
+%Q(\«)
+
+%q(\«)
+
 %^#$^#
 
 %@#@#