Python实现针对含中文字符串的截取功能示例

def subString(string,length):
  if length >= len(string):
        return string
  result = ''
  i = 0
  p = 0
  while True:
        ch = ord(string[i])
        #1111110x
        if ch >= 252:
            p = p + 6
        #111110xx
        elif ch >= 248:
            p = p + 5
        #11110xxx
        elif ch >= 240:
            p = p + 4
        #1110xxxx
        elif ch >= 224:
            p = p + 3
        #110xxxxx
        elif ch >= 192:
            p = p + 2
        else:
            p = p + 1
        if p >= length:
            break;
        else:
            i = p
  return string[0:i]

def cut_string_off(string,s_len):
    if len(string)==0 or s_len <=0:
        return string
    elif len(string)==1 or s_len >= len(string):
        return string
    elif s_len < len(string):
        len_num=0
        while len_num < s_len:
            tmp_c=ord(string[len_num])
            if tmp_c >0 and tmp_c <=0x7F:
                len_num+=1
                continue
            tmp_nextc=ord(string[len_num+1])
            if tmp_c >= 0x81 and tmp_c <=0xFE and tmp_nextc>=0x40 and tmp_nextc<=0xFE:
                len_num+=2
                continue
            else:
                len_num +=1;
                continue
            break
        tmp = string[0:len_num]
#    print utf2gbk(tmp)
    return tmp