Python:套接字发送结构(以C结构为例)

时间:2013-01-31 07:33:13

标签: python c sockets struct python-3.x

下面是来自C的Struct,我正在尝试转换为Python, 并使用Socket发送结构

C:

        struct CommandReq
        {
            char    sMark[6];  //start flag\r\n*KW
            short   nPackLen;  //packet length
            short   nFlag;     //command ID 0x0002
            int     nGisIp;    //GIS port
            short   nPort;     //GIS Port
            char    sData[50]; //command string
            char    sEnd[2];   //end flag "\r\n"
        };
        //source code
        CommandReq stResq;
        memset(&stResq, 0, sizeof(stResq));
        sprintf(stResq.sMark, "\r\n%s", "*KW");
        stResq.nFlag = 0x0002;
        stResq.nPackLen = sizeof(stResq);
        stResq.nGisIp = 0;
        stResq.nPort = 0;
        strcpy(stResq.sData, "*KW,CC09C00001,015,080756,#");
        strncpy(stResq.sEnd, "\r\n", 2);

我使用namedtuple创建了Python结构,并使用socket发送此结构。 但不幸的是失败了。

的Python:

from collections import namedtuple
MyStruct = namedtuple("MyStruct", "sMark nPackLen nFlag nGisIp nPort sData sEnd")

编辑,回复以下答案

在弄清楚之后,python 3.x必须添加字符串.encode(“ascii”)

      format_ = "6shhih50s2s"
      MyStruct = namedtuple("MyStruct", "sMark nPackLen nFlag nGisIp nPort sData sEnd")
      tuple_to_send = MyStruct(sMark="\r\n{}".format("*KW").encode("ascii"), 
                 nPackLen=struct.calcsize(format_),
                 nFlag=0x0002,
                 nGisIp=0,
                 nPort=0,
                 sData= "*KW,NR09G05133,015,080756,#".encode("ascii"),
                 sEnd="\r\n".encode("ascii"))
      string_to_send = struct.pack(format_, *tuple_to_send._asdict().values()
      socket.sendto(string_to_send, self.client_address)

附加问题

下面是数据包正在发送结构

format_ = "6shhih50s2s" //total bytes should be 68 bytes?
    0d0a2a4b5700 //6 char \r\n*KW
    4600 // packet length 70 bytes
    0200 //command ID 0x0002,short 2 bytes
    00000000 //GIS port is integer 4 bytes
    0000 //GIS Port short 2 bytes
    0000 //unknown??what is this?
    2a4b572c4e5230394730353133332c3031352c3038303735362c230000000000000000000000000000000000000000000000 //sData[50] char 50 bytes
    0d0a //sEnd 2 char,2 bytes

2个未知字节怎么来?

1 个答案:

答案 0 :(得分:1)

namedtuple与c-struct无法真正比​​较。如果您习惯使用结构,可以查看struct-module以将结构化信息转换为字符串。

通常,Pythonistas更喜欢使用pickle-module进行序列化。

from collections import namedtuple
import pickle # or cPickle, it's faster
MyStruct = namedtuple("MyStruct", "sMark nPackLen nFlag nGisIp nPort sData sEnd")

tuple_to_send = MyStruct(sMark="abcdef", nPackLen=...)
string_to_send = pickle.dumps(tuple_to_send)

有两种口味的泡菜,picklecPickle。后者更快,但仅在CPython(大多数程序员使用)中可用,而pickle也可用于Jython,IronPython,......

如果你想坚持使用struct(例如,因为另一方需要这种格式),你的格式字符串将是

format_ = "6shhih50s2s"

所以你可以这样做:

import struct
from collections import namedtuple

format_ = "6shhih50s2s"

MyStruct = namedtuple("MyStruct", "sMark nPackLen nFlag nGisIp nPort sData sEnd")
tuple_to_send = MyStruct(sMark="\r\n{}".format("*KW"), 
                     nPackLen=struct.calcsize(format_),
                     nFlag=0x0002,
                     nGisIp=0,
                     nPort=0,
                     sData= "*KW,NR09G05133,015,080756,#",
                     sEnd="\r\n")


string_to_send = struct.pack(format_, *tuple_to_send._asdict().values())

BTW:namedtuples是不可变的,即你不能做

tuple_to_send.sMark = "bcdefg"

改变一些财产。如果您需要此类行为,则必须创建一个类。

修改

对于Python 3,字符串处理已更改。必须执行从unicode到字节的转换,例如

import struct
from collections import namedtuple

format_ = "6shhih50s2s"

MyStruct = namedtuple("MyStruct", "sMark nPackLen nFlag nGisIp nPort sData sEnd")
tuple_to_send = MyStruct(sMark="\r\n{}".format("*KW").encode("ascii"), 
                     nPackLen=struct.calcsize(format_),
                     nFlag=0x0002,
                     nGisIp=0,
                     nPort=0,
                     sData= "*KW,NR09G05133,015,080756,#".encode("ascii"),
                     sEnd=b"\r\n")


string_to_send = struct.pack(format_, *tuple_to_send._asdict().values())