tags:

views:

359

answers:

4

Hi

I am doing a simple client implementation of TFTP. Here i need to send read request in following format

/* send request 

      2 bytes    string     1 byte     string   1 byte
------------------------------------------------------
RRQ/  | 01/02 |  Filename  |   0  |    Mode    |   0  |
WRQ  -------------------------------------------------

*/

in between i have to insert 1 byte zero bits value . But i am not able to add that value. Also if i add a 1 zero bits byte.. which actually means a string terminating character than how to get proper strlen value.

If any one can help me with this...

enter code here


const char opcode_read[2] ={'0','1'};
const char opcode_write[2] ={'0','2'};
const char opcode_data[2] ={'0','3'};
const char opcode_acknowledge[2] ={'0','4'};
const char opcode_error[2] ={'0','5'}; 
const char mode_netascii[] = "netascii\0";
char blk_read_request[100];
char file_name[] = "rfc0791.txt\0";

memcpy(blk_read_request, opcode_read, 2);
memcpy(&blk_read_request[2], file_name, strlen(file_name) + 1);
memcpy(&blk_read_request[2 + strlen(file_name)], mode_netascii, strlen(mode_netascii) + 1);

for (int i = 0; i < strlen(blk_read_request); i++) {
    cout << i << " : " << blk_read_request[i] << " " << std::bitset<CHAR_BIT > (blk_read_request[i]) << "\n";
}
+2  A: 

This is automatic in C++ strings.

This:

const char mode_netascii[] = "netascii";

Is equivalent to this:

const char mode_netascii[] = { 'n', 'e', 't', 'a', 's', 'c', 'i', 'i', '\0' };
Kip
A: 
char *total_send_string = malloc(2+strlen(file_name)+sizeof(Mode)+3);
memset(total_send_string, 0, strlen(total_send_string);
//Then copy each member of the total send packet into the correct offset
//Now you have the correct result
Paul Nathan
+3  A: 

If you want to transfer the null across, not only must you memcpy with "strlen(filename) + 1", but you'll also need to update the destination pointer for the subsequent memcpys accordingly.

memcpy(&blk_read_request[2 + strlen(file_name) + 1], mode_netascii, strlen(mode_netascii) + 1);

Note the extra +1 in there.

For the record, being C++, you might want want to consider using a higher-level class than "char *" which can handle embedded null characters.

For example, the following prints "8", and opens up std::stringstream as a better way of forming your packets:

#include <string>
#include <iostream>

int main()
{
    std::string x = "foo";
    x += '\0';
    x += "bar";
    x += '\0';

    std::cout << x.length() << std::endl;
}
Martin
+2  A: 

If you need to deal with data that will (or might) include null bytes then you'll generally want to avoid using C-style strings and the functions that deal with them. Use character/byte arrays and keep a length associated with the length of the data in the buffers.

In C++ vectors of bytes (or char) would work great for this.

Also in C++, the std::string type can contain null characters just fine; however, I'd suggest avoiding their use and sticking with std::vector<char> since it would be all too easy to fall into some bug traps of passing the result of string::c_str() to something that expects null terminated strings.

Michael Burr