utf8check.rb

UTF8_REGEXP = / \A (?: [\x00-\x7F] | [\xC2-\xDF] [\x80-\xBF] | [\xE1-\xEF] [\x80-\xBF]{2} |
                       [\xF1-\xF7] [\x80-\xBF]{3} | [\xF9-\xFB] [\x80-\xBF]{4} |
                       [\xFD-\xFD] [\x80-\xBF]{5} ) \Z /x

# Return true if string is valid UTF-8
def utf8?(string)
  string.split(//u).all? { |c| c =~ UTF8_REGEXP}
end