working with byte arrays

jkirkerx

I understand now that the data I receive from a socket is a byte array. But I don't understand how to pass the byte array to a function for further processing. Plus I'm not sure if I should leave it as a byte array, or try to convert it to a string, so I can extract data from it. I'm confused as to whether this buffer is a pointer to data somewhere else, and If I need to make a copy of it, before passing it to a function. When I pass it to a function, the data is lost. Kind of tired of going around in circles here on this, and I need road map to head in the right direction. I make my receive buffer char recvbuf[256]; I get data 0x05 '' 0x54 'T' 0x53 'S' 0x65 'e' And I try to pass it to a function for processing. Unless just extracting the info I need does not require a function. Suggestion? void _process(char *pbuffer) { }

Chuck OToole

I can't speak for your application nor what data you will be receiving from the socket. But if the data is a "protocol", that is, some well defined sequence of bytes with fields, flags, data, etc, then it is most likely that the information is "binary", that is, not all the data maps nicely into printable or viewable characters, all zero bytes comes to mind, your 0x05 example too. C and C++ "string" variables, whether CString or std::string and their manipulation functions are not really tolerant of nulls intersperced in the data. C# strings do allow nulls in the data and I have seen many a C / C++ programmer get totally confused by the idea of nulls in strings. So, I'd say save yourself some future grief and use a byte array. Passing a byte array is no different than passing a char array other than some folks incorrectly equate "char array" with "string" in all cases.

Albert Holguin

Don't convert it to a string... that's a bad idea... extract whatever data bytes according to whatever API you happen to be using, then cast the converted bytes to whatever type they're supposed to be. As far as whether you need to copy the data before passing it, it depends on where you defined your recvbuff[] and whether it'll be used again, if the buffer is a local variable within your receive method, then it'll automatically go out of scope when the method is finished, so you should copy the data... also, if the buffer is to be used repeatedly (receiving multiple packets), then the data should be copied over to another buffer before moving on, otherwise it might be overwritten before it's used downstream.

jkirkerx

Great Information! In this case, the data being received are byte stream? Header stuff, ServerName:WWW;Custered:false;NamedPipe:\\pipename;Version:1.5.02.23;empty space Taking into consideration the thread below yours, I will make a copy of the buffer, and pass the copy to the function to fish out the server name, and version. Thanks for looking at my post, your advice puts back on track. jkirkerx

jkirkerx

The research I did made it sound like std:string was the swiss army knife or manipulation. I used memset to copy the data, because I could not figure out how to use strcpy, the copy has the first 2 bytes, and stopped. I still can't figure out why when I pass the data, I only get the first 2 bytes. You wisdom gives me a map to use, so I can focus on the copy and pass, then I have to tackle the data extraction. Thank you very much jkirkerx

Lost User

You seem to be working at this from the wrong direction. The first thing you need to do is work out (or ask the sender) the exact format of the data you are receiving. All data read from sockets is presented as a stream of bytes, and without knowing the significance, format and values of each byte or group of bytes, there is no way you can do anything with the information you receive. Never assume that data is presented as a string unless you are certain that is how it has been sent.

Unrequited desire is character building. OriginalGriff

Lost User

Yeah, if you know the data is a char array, pass it to the func as you outineline in your post, and let it do the processing. It is bette to do this, than to proces the data at the level of the code that manages socket communication, it is a more logical split to treat communication at one level, and data processing at another.

============================== Nothing to say.

jkirkerx

The sender is microsoft's sql browser. I discovered the return packet data using etherreal, and sending various request to it, until I got the reply I was looking for. Then I wrote the socket program. I have the data packets coming back from all the sql servers on the lan, and they all send the same format back to me. I just need to extract the server name and version number of sql out of the byte array.

jkirkerx

I write in vb, and without a doubt, I would pass the data to another function for processing, or just store the data, and process it later. In straight c++, most of the stuff seems pretty easy, but there are some really low level stuff you can get into that gets tricky. I've never asked anyone to write code for me, but I'm really stuck trying to figure out how to pass the data to a another function. I will post my code, and the byte stream I get back, and perhaps someone could can spot the mistake I made, it might be something really small that I didn't get.

BOOL CA_SQLServer_Scan::_socket_Enumerate_SQLServers( void )
{
int iResult;
WSADATA wsaData;

char recvbuf\[256\];
int rv = 0; 
int recvbuflen = 256;
int bytesReceived = 0;
				    
iResult = WSAStartup(MAKEWORD(2,2), &wsaData);
if (iResult != NO\_ERROR) {
    return FALSE;
}		
//////////////////////////////////////////////////////////////////////////////////////
// Send Data Socket
SOCKET sUDPSocket = INVALID\_SOCKET;
struct sockaddr\_in sTargetDevice;
char cBuffer\[\] = {0x02};

sUDPSocket = socket(AF\_INET, SOCK\_DGRAM, IPPROTO\_UDP);
if (sUDPSocket == INVALID\_SOCKET) {
    int iSocketError = WSAGetLastError();
	goto wsaCleanup;
}

ZeroMemory( &sTargetDevice, sizeof(sTargetDevice));
sTargetDevice.sin\_family = AF\_INET;
sTargetDevice.sin\_addr.s\_addr = inet\_addr( "192.168.3.255" );
sTargetDevice.sin\_port = htons( BROADCAST\_SND\_PORT );

// Try a Connectionless Send Data first
iResult = sendto(sUDPSocket, cBuffer, 1, 0, (SOCKADDR \*) & sTargetDevice, sizeof (sTargetDevice));
if (iResult == SOCKET\_ERROR) {
    goto closeSocket;        
}
    
// Shutdown the send socket for some reason
iResult = shutdown(sUDPSocket, SD\_SEND);
if (iResult == SOCKET\_ERROR) {
    goto closeSocket;
}
// End of Send Data Connection
//////////////////////////////////////////////////////////////////////////////////////
// Enumerate Return Packets	
do {
    bytesReceived = recv(sUDPSocket, recvbuf, recvbuflen, 0);
    if ( bytesReceived > 0 ) {            
		char buf\[256\];
		recvbuf\[bytesReceived\] = '\\0';
		memcpy(buf, recvbuf, bytesReceived);			
		\_process\_SQL\_BufferData(buf, bytesReceived);						
	}
    else if ( bytesReceived == 0 ) {
        printf("Connection closed\\n");
		goto closeSocket;
	}		
	else {
        printf("recv failed: %d\\n", WSAGetLastError());
		goto closeSocket;
	}

Lost User

jkirkerx wrote:

I just need to extract the server name and version number of sql out of the byte array.

Given that you have figured out the format and content of the messages, what is the problem?

Unrequited desire is character building. OriginalGriff

Goto_Label_

Does this code help point you in the right direction? rcvBuffer is passed to the function processData

#include #include void processData(char &c)
{
char* array = &c;
std::cout << array << std::endl;
}

int main()
{

char\* rcvBuffer = "Hello world";
char& c = rcvBuffer\[0\];
processData(c);

return 0;
}

Lost User

assuming this is the format of all messages that you need to process, it looks like the message format is :

Addr Content
0 05
1 0084 (as bytes 8400)
3 "ServerName;DELLC521-01;InstanceName;SQLEXPRESS;IsClustered;No;Version;10.50.1600.1;;"

Byte zero contains some flag or identifier (no idea what)
Bytes 1 & 2 contain a 16-bit integer giving the length of the following data
Bytes 3 - n contain the message data, which in this case is a semi-colon delimited string.

So given the above information your message processor can copy the characters (using the length information) into a proper null-terminated string buffer. That string may then be split into tokens and each required value can be found by comparing the keywords with known tokens. Something like:

int msgLength = buffer[2] << 8 + buffer[1];
char* szWords = new char[msgLength + 1];
strncpy(szWords, buffer + 3, msgLength);
// remainder of code left as an exercise for the reader
// use strtok() to extract each token
// strcmp() to look for keywords
//
// alternatively use std::string to parse and extract

Unrequited desire is character building. OriginalGriff I'm sitting here giving you a standing ovation - Len Goodman

App_

:thumbsup:

App_

is done like this:

#include <iostream>
using namespace std;

void charray(char test[]) {
strncpy(test,"test",sizeof(test));
}

int main() {
char test[256]={' '};
charray(test);
printf("Text in %s!\n", test);
return 0;
}

jkirkerx

I experimented with that Friday night, and only the first byte copied over in strncpy. I think the 2nd byte bombed in the copy. The same thing happens when I pass the byte array to a function. It must be the 2nd byte, so perhaps I need to cut off the header of the byte array, and then do the copy and pass.

jkirkerx

Give me an hour to digest the knowledge you presented, looks very helpful and promising. perhaps in words, get the message length the packet declared make a new byte array copy the message only to the new byte array, leaving the packet header behind then pass the message to the function, and extract the ; delimited data.

jkirkerx

Didn't work at first, then I hardcoded the message length just for testing, and everything worked, including passing the string to the function. I need to figure out the length calculation. Working on that now, almost there, wow

int msgLength = 80;
char* szWords = new char[msgLength + 1];
strncpy(szWords, recvbuf + 3, msgLength);

This is the szWords,

szWords 0x003e8f10 "ServerName;DELLC521-01;InstanceName;SQLEXPRESS;IsClustered;No;Version;10.50.1600Íýýýý««««««««þîþ"

Lost User

You cannot use sizeof() in the function above, because the array size is not known at that point.

Unrequited desire is character building. OriginalGriff I'm sitting here giving you a standing ovation - Len Goodman

Lost User

jkirkerx wrote:

I need to figure out the length calculation.

I gave you the code for that in my previous message; and I missed a line of code, so it should read:

int msgLength = buffer[2] << 8 + buffer[1];
char* szWords = new char[msgLength + 1];
strncpy(szWords, buffer + 3, msgLength);
szWords[msgLength] = '\0';

Unrequited desire is character building. OriginalGriff I'm sitting here giving you a standing ovation - Len Goodman

jkirkerx

So I still need to terminate the string with \0 Having trouble with the length, I did quick cheat with

int msgLength = bytesReceived - 3;
//int msgLength = recvbuf[2] << 8 + recvbuf[1];
char* szWords = new char[msgLength + 1];
strncpy(szWords, recvbuf + 3, msgLength);

int msgLength with is an integer, I keep getting 0, is it because recvbuf[2] is returning the string, but wait, it's a byte representing a number. hmm