UTF-8 aware string chop

chop_utf8.rb

# UTF-8 aware string chop. Returns an array with two elements, the first
# contains the iven string excluding the last character, and the second,
# also the last, contains the last character.
def chop_utf8(s)
  return unless s

  a = s.unpack('C*')
  c, w = 0, 0
  lead, last = '', ''

  while c < a.length
    case a[c]
    when 0x00..0x7E; w = 1
    when 0xC2..0xDF; w = 2
    when 0xE0..0xEF; w = 3
    when 0xF0..0xF4; w = 4
    else w = 1 # other ASCII
    end

    if (c + w) >= a.length
      last = a[c..c+(w-1)].pack('c*')
    else
      lead << a[c..c+(w-1)].pack('c*')
    end

    c += w
  end

  [lead, last]
end