如何将带有 null 字符的字符串从 C++ DLL 返回到 ZA7F5F35426B927411FC9231B56

Question

I'm working on a Python 3 application that calls a DLL developed in C++. The DLL reads a database record and returns the record buffer to the Python application. The issue that I am having is that the database record may contain x'00' (null) characters. When this occurs the record buffer that is returned to the Python application is truncated at the null character.

I am using memcpy to copy the record buffer to the Python return area. I thought memcpy copied the number of bytes specified, regardless of the content.

I am not a C++ programmer, so I quite possibly could be misunderstanding how memcpy works.

The following code snippets are examples of test code that will demonstrate the problem I am having.

This is the DLL:

#include "pch.h"
#include <stdio.h>
#include <string.h>
#include <stdlib.h>     
#include <iostream>

using namespace std;

#define LIBDLL extern "C" __declspec(dllexport)

LIBDLL int readData_VPRI2(char* buffer, unsigned long buffer_length )

{
    char unsigned buf[20];
    int unsigned buf_len;
    FILE* stream;
    errno_t err;
    err = fopen_s(&stream, "D:\\testfile", "rb");
    if (err != 0) return err;
    fread(buf, buffer_length, 1, stream);
    memcpy(buffer, buf, buffer_length);
    return 0;

}

This is the Python calling routine:

import ctypes
from ctypes import *

dll = ctypes.CDLL('C:\\Users\\rhkea\\source\\repos\\TestDLL\\x64\\Debug\\TestDLL.dll')

f = open("D:\\testfile", "wb")
test_string = bytes('ABCD\x00efg', encoding='utf-8')
f.write(test_string)
f.close()

char_arr = ctypes.c_char * 500                                      # set up the return area
buffer = char_arr()                                                 # assign the buffer
readData_VPRI2 = dll.readData_VPRI2                                 # get the DLL
readData_VPRI2.restype = ctypes.c_int                               # set the return type
readData_VPRI2.argtypes = (POINTER(c_char), c_long)                 # define the arguments

rc = readData_VPRI2(buffer, len(test_string))                       # call the DLL

print ("rc =", rc)
if rc==0:
    print ("buffer =", buffer.value)
    print ("buffer length = ", len(buffer.value))

The output of the Python execution is:

rc = 0
buffer = b'ABCD'
buffer length =  4

As shown, the buffer that is returned is truncated at the x'00'.

I'm guessing there may be something simple that I am overlooking or don't understand. Any guidance on how correct this problem would be greatly appreciated.

Thanks in advance.

Answer 1

As it turns out, the C++ code is working as designed. My question about memcpy is not the issue. It is working as documented.

What I discovered is that Python may actually be truncating the string. The cytpes.c_char.value returns the truncated string as described in my original question.

But if you use cyptes.c_char.raw the entire string (in my example, buffer ) is returned (eg, 500 bytes), padded with x'00'. So print(buffer.raw) returns:

b'AB\x00CDefg\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00... ...\x00'

Since I already know the length of the record I am expecting to be returned, I can simply slice it from the buffer.raw .

Thanks for your responses.