import base64


def pad_count(n, b):
    if n % b == 0:
        return 0
    else:
        return b - (n % b)


def pack_ints(l):
    """
    Pack a set of ints into a URL-safe string that can be used in a URL path
    component.
    """

    def do_pack(s, size):
        bits = []
        for i in s:
            b = '{:b}'.format(i)
            b = '0' * pad_count(len(b), size - 1) + b
            for j in range(0, len(b), size - 1):
                bits.append(str(int(j == 0)))  # lol
                bits.extend(b[j:j+(size-1)])
        return bits

    diffs = []
    last = 0
    for i in sorted(set(l)):
        diffs.append(i - last)
        last = i

    best, size = do_pack(diffs, 8), 8
    for i in range(3, 8):
        attempt = do_pack(diffs, i)
        if len(attempt) < len(best):
            best, size = attempt, i

    # Padding works by adding 0b1, then padding the rest of the byte with 0b0.
    # In "UTF-N" (UTF-5 but with any surrogate length) the first surrogate in
    # the padding must refer to the value 0, because 0b10... cannot have any
    # following 0b0... surrogates, and a list of differences of an ordered set
    # cannot contain 0, so it can be safely used as a "stop processing" marker.
    pad_size = pad_count(3 + len(best), 8)
    if pad_size == 0:
        padding = ''
    else:
        padding = list('1' + '0' * (pad_size - 1))
    bits = list('{:03b}'.format(size - 1)) + best + padding
    bytes = bytearray(int(''.join(bits[i:i+8]), 2)
                      for i in range(0, len(bits), 8))
    return base64.b64encode(bytes)


def unpack_ints(s):
    """
    Unpack a set of ints generated by pack_ints.
    """

    bytes = base64.b64decode(s)
    if len(bytes) == 0:
        return []
    bits = ''.join(item for sublist in ('{:08b}'.format(b) for b in bytes)
                   for item in sublist)
    size, bits = int(bits[:3], 2) + 1, bits[3:]
    diffs = []
    for i in range(0, len(bits), size):
        surr = bits[i:i+size]
        if surr[0] == '1':
            if len(surr) == 1 or int(surr[1:], 2) == 0:  # stop marker
                break
            diffs.append(int(surr[1:], 2))
        else:
            diffs[len(diffs)-1] <<= (size - 1)
            diffs[len(diffs)-1] += int(surr[1:], 2)

    l = []
    last = 0
    for i in diffs:
        last += i
        l.append(last)
    return l


if __name__ == '__main__':
    with open('/home/ilianaw/rrix.txt') as f:
        s = sorted(set(int(x) for x in f.readlines()))
    print(s)
    packed = pack_ints(s).decode('ascii')
    print(packed)
    unpacked = unpack_ints(packed)
    print(unpacked)