Review 

• Initially we were reading bytes and printing bytes
• The regex replaced bytes with bytes
• ‘use utf8’ made the regex replacement a character string
• Which upgraded each string it modified
• We needed to decode the bytes on input
• and encode to bytes on output