BigEndianStructure中的字符串 [英] Strings in BigEndianStructure

问题描述

我想做这样的事情：

  from ctypes import * 
 
 class Packet（ BigEndianStructure）：
 _fields_ = [（ length，c_ushort），
（ session，c_uint），
（ command，c_ushort）] 
 
 class PacketString（BigEndianStructure）：
 _fields_ = [（ length，c_ushort），
（ value，c_char_p）] 
 
 class InitialPacket（Packet）：
 _fields_ = [（ time，PacketString）] 
  

但是由于c_char_p只能是以本机字节顺序。但是，也许还有其他方法可以使字符串的长度在字符串之前指定。我只是喜欢如何轻松地从套接字读取/写入结构。以及如何仅定义_fields_，然后像这样使用它：

  initialPacket = InitialPacket（）
 initialPacket.command = 128 
  

问题是：如何在BigEndianStructure中创建可变长度字段？因为Python不允许我使用c_char_p。脚本根本不会运行。这是错误的消息：

 跟踪（最近一次通话最近）：
文件 C：\PKOEmu\test .py，第8行，在< module>中
类PacketString（BigEndianStructure）：
文件 C：\Python27\lib\ctypes\_endian.py，第34行，位于__setattr__ 
fields.append（（ _other_endian（typ））+ rest）
文件 C：\Python27\lib\ctypes\_endian.py，第24行，在_other_endian 
中引发TypeError（其他字节序：％s％typ）
 TypeError：此类型不支持其他字节序：< class'ctypes.c_char_p'> 
  

解决方案

此类型：

 类PacketString（BigEndianStructure）：
 _fields_ = [（ length，c_ushort），
（ value，c_char_p）] 
  

…甚至忽略了字节序问题，也没有做您认为的事情。这是一个包含ushort长度的结构，然后是指向内存中其他地方的实际字符串数据的指针。

换句话说，它就像下面的C结构：

  struct PacketString {
无符号短长度； 
 char *值; 
}; 
  

您要查找的是一个长度前缀的字符串，该字符串直接插入结构。为此，C结构为：

  struct PacketString {
无符号短长度； 
个字符值[1]； 
}; 
  

这称为结构破解。这实际上不是合法的C语言，但是它恰好适用于每个已知的C89编译器以及大多数C99和C ++编译器。有关详细信息，请参见 C常见问题解答条目。

那么，您可以在 ctypes 中做同样的事情吗？好吧，是的，但是它没有那么有用：

  class PacketString（BigEndianStructure）：
 _fields_ = [（ length ，c_ushort），
（ value，c_char * 0）] 
  

变得复杂请参阅文档中的变量大小的数据类型有关详细信息。特别是，您不能在 p.value 上调用调整大小；您需要计算自己调整 p 大小的大小，然后将 p._fields_ [1] 的类型更改为

嗯，这就是为什么文档说：

将可变大小的数据类型与ctypes一起使用的另一种方法是使用Python的动态特性，并根据具体情况在所需的大小已知后重新定义数据类型。 / p>

换句话说：

  class LocalPacketString（BigEndianStructure）：
 _fields_ = [（ length，c_ushort），
（ value，c_char * length）] 
 ps = LocalPacketString（length，buff）
  

但是，您可能会注意到，这与单独保存类型相比并没有节省很多工作。

---

总而言之，struct hack甚至都不是有效的C，它与<$ c $的映射也不太好c> ctypes 。 ctypes.Structure 并不是表示长度可变的字符串的好方法。

I want to do something like that:

from ctypes import *

class Packet(BigEndianStructure):
    _fields_ = [("length", c_ushort),
                ("session", c_uint),
                ("command", c_ushort)]

class PacketString(BigEndianStructure):
    _fields_ = [("length", c_ushort),
                ("value", c_char_p)]

class InitialPacket(Packet):
    _fields_ = [("time", PacketString)]

However I getting error because c_char_p can only be in native byte order. But maybe there is some other way how I can make strings whose length is specified right before them. I just like how structs are easy to read from/write to socket. And how you define just _fields_ and then can use it like that:

initialPacket = InitialPacket()
initialPacket.command = 128

The question is: How do I create variable-length field in BigEndianStructure? Because Python wouldn't allow me to use c_char_p. Script simply wont run at all. This is error:

Traceback (most recent call last):
  File "C:\PKOEmu\test.py", line 8, in <module>
    class PacketString(BigEndianStructure):
  File "C:\Python27\lib\ctypes\_endian.py", line 34, in __setattr__
    fields.append((name, _other_endian(typ)) + rest)
  File "C:\Python27\lib\ctypes\_endian.py", line 24, in _other_endian
    raise TypeError("This type does not support other endian: %s" % typ)
TypeError: This type does not support other endian: <class 'ctypes.c_char_p'>

解决方案

This type:

class PacketString(BigEndianStructure):
    _fields_ = [("length", c_ushort),
                ("value", c_char_p)]

… doesn't do what you think it does, even ignoring the endianness issue. It's a struct that contains a ushort length, and then a pointer to the actual string data somewhere else in memory.

In other words, it's just like this C structure:

struct PacketString {
    unsigned short length;
    char *value;
};

What you're looking for is a length-prefixed string, where the string is directly inline inside the struct. For that, the C structure is:

struct PacketString {
    unsigned short length;
    char value[1];
};

This is called the "struct hack". This is not actually legal C, but it happens to work with every known C89 compiler, and most C99 and C++ compilers. See the C FAQ entry fro details.

So, can you do the same thing in ctypes? Well, yes, but it's not as useful:

class PacketString(BigEndianStructure):
    _fields_ = [("length", c_ushort),
                ("value", c_char * 0)]

This can get complicated; see Variable-sized data types in the docs for details. In particular, you can't call resize on p.value; you need to calculate how much to resize p itself, and then change the type of p._fields_[1] to the have the right type, and then…

Well, this is why the docs say:

Another way to use variable-sized data types with ctypes is to use the dynamic nature of Python, and (re-)define the data type after the required size is already known, on a case by case basis.

In other words:

class LocalPacketString(BigEndianStructure):
    _fields_ = [("length", c_ushort),
                ("value", c_char * length)]
ps = LocalPacketString(length, buff)

However, you may notice that this isn't really saving you much work over just keeping the types separate.

---

In summary, the struct hack isn't even valid C, and it doesn't map very well to ctypes. A ctypes.Structure is not a good way to represent a variable-length length-prefixed string.

这篇关于BigEndianStructure中的字符串的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！