这是我查看的博客
http://blog.csdn.net/chuanchuan608/article/details/17915959
直接引用里面的关键语句：
When you use client_socket.send(data),replace it by client_socket.send(data.encode()). When you get datausing data = client_socket.recv(512), replace it by data =client_socket.recv(512).decode()
翻译过来，也就是说在你使用python3的时候，如果想要使用套接字组件send发送str类型的东西，记得先encode()一下。同理，当你想用recv接受str东西的时候，记得decode()一下。
为什么会是这样呢？
In python 3, bytes strings and unicodestrings are now two different types. Since sockets are not aware of string encodings, they are using raw bytes strings, that have a slightly differentinterface from unicode strings.
So, now, whenever you have a unicode stringthat you need to use as a byte string, you need toencode() it. And whenyou have a byte string, you need to decode it to use it as a regular(python 2.x) string.
Unicode strings are quotes enclosedstrings. Bytes strings are b"" enclosed strings
原来在python3里面，bytes strings和unicode strings已经是不同的两个类型的（是不是说python2里面是同一类型？不清楚）。由于套接字并不关心字符串的编码方式，他们使用原字节字符串（raw bytes strings）进行传输，而这个类型又和unicode strings又有一点小小的不同。
因此，现在任何时候你要是有一个unicode strings需要把它转化为byte string，你需要对他进行encode操作。当你有bytes string的时候，你需要decode操作。
unicode strings =encode()=》 byte strings
byte strings =decode()=》 unicode strings