utf8cut.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

def utf8cut(ba, l):
    if len(ba) <= l:
        return ba

    pos = l
    while (pos > 0):
        b = ba[pos]
        if (b & 0x80 == 0) or (b & 0xc0 == 0xc0):
            break
        else:
            pos -= 1

    return ba[0:pos]

x = bytearray("1世2界1")

for i in range(len(x)):
    print i, "= [", str(utf8cut(x, i)), "]"