解决NSData中包含非法UTF-8编码

1字节 0xxxxxxx
2字节 110xxxxx 10xxxxxx
3字节 1110xxxx 10xxxxxx 10xxxxxx
4字节 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx

@implementation NSData (UTF8)

- (NSData *)UTF8Data
{
    //保存结果
    NSMutableData *resData = [[NSMutableData alloc] initWithCapacity:self.length];
    
    //无效编码替代符号(常见 ? □ ?)
    NSData *replacement = [@"?" dataUsingEncoding:NSUTF8StringEncoding];
    
    uint64_t index = 0;
    const uint8_t *bytes = self.bytes;
    
    while (index < self.length)
    {
        uint8_t len = 0;
        uint8_t header = bytes[index];
        
        //单字节
        if ((header&0x80) == 0)
        {
            len = 1;
        }
        //2字节(并且不能为C0,C1)
        else if ((header&0xE0) == 0xC0)
        {
            if (header != 0xC0 && header != 0xC1)
            {
                len = 2;
            }
        }
        //3字节
        else if((header&0xF0) == 0xE0)
        {
            len = 3;
        }
        //4字节(并且不能为F5,F6,F7)
        else if ((header&0xF8) == 0xF0)
        {
            if (header != 0xF5 && header != 0xF6 && header != 0xF7)
            {
                len = 4;
            }
        }
        
        //无法识别
        if (len == 0)
        {
            [resData appendData:replacement];
            index++;
            continue;
        }
        
        //检测有效的数据长度(后面还有多少个10xxxxxx这样的字节)
        uint8_t validLen = 1;
        while (validLen < len && index+validLen < self.length)
        {
            if ((bytes[index+validLen] & 0xC0) != 0x80)
                break;
            validLen++;
        }
        
        //有效字节等于编码要求的字节数表示合法,否则不合法
        if (validLen == len)
        {
            [resData appendBytes:bytes+index length:len];
        }else
        {
            [resData appendData:replacement];
        }
        
        //移动下标
        index += validLen;
    }
    
    return resData;
}

@end