Question

I have a quick little question involving the read() command.

I am really rusty in C, and unfortunately, my current assignment has me programming in C.
We need to read stdin using the read() and not fgets and the like.

So I have a simple while loop:

int n, i;
char buffer[257], *input;

while((n=read(fileno(stdin), buffer, 256)) > 0)
                {
                        buffer[n] ='\0';
                        if(buffer[n] = '\n') break;
                        write(input, buffer, strlen(buffer));
                }

So I am not sure how to make this loop stop at the press of enter (end of line) so that’s why I have the break code, though I don’t know if that’s done correctly.

The whole objective I am trying to accomplish is to put the input from stdin into the pointer ‘input’

(I am really bad at understanding pointers so bear with me 🙂 )

My problem is that I am getting segmentation faults when I press enter.
I wish I could use fgets because then all this would be solved with a simple

input = fgets(buffer, 256, stdin);

Darn homework 🙁
Anyways, if any of you could point me in the right direction, I would really appreciate it. Thank you!

Answer 1

You’re going about it all wrong. Firstly, write is a system call that’s used to write to a open file descriptor. Your input is nothing like that – a file descriptor is gained by calling the open syscall first, or referring to one of the always-open files, e.g. stdin, stdout or stderr. Not to mention that a file descriptor is an int on Linux.

Also, you should remember that the assumption of your input on stdin ending with a newline doesn’t have to be right all the time. Your program may receive some content on standard input that doesn’t contain any newlines, for example contents of some file (redirected to stdin), or the input from keyboard may simply end with a Ctrl-D (EOF) instead of the “return” key.

On top of all that, an uninitialized pointer doesn’t refer to any valid memory. The declaration of char *input doesn’t give you any right to write/read the memory referred to by input unless you allocate some memory that this pointer will point to first.

Copying strings in C is achieved by using functions declared in <string.h>. They operate on C-strings, e.g. sequences of characters terminated by \0. In this case, read() doesn’t terminate its output by a \0, in which case you’ll want to use the memcpy function instead. Why don’t you simply try reading directly into input, though? It doesn’t make much sense to read to buffer first, simply to copy the data to another chunk of memory later on.

Also, the specification of read states that it reads up to count bytes from the file descriptor. Which means that if your buffer is declared as char[257], then you should call read(stdin,buffer,257), not 256. It would probably also be more suitable to simply declare the buffer as a 256-element array.

I hope that I’ve straightened out some stuff for you.