Reading from stdin

Question

What are the possible ways for reading user input using read() system call in Unix. How can we read from stdin byte by byte using read()?

Solution 1

You can do something like this to read 10 bytes:

char buffer[10];
read(STDIN_FILENO, buffer, 10);

remember read() doesn't add '\0' to terminate to make it string (just gives raw buffer).

To read 1 byte at a time:

char ch;
while(read(STDIN_FILENO, &ch, 1) > 0)
{
 //do stuff
}

and don't forget to #include <unistd.h>, STDIN_FILENO defined as macro in this file.

There are three standard POSIX file descriptors, corresponding to the three standard streams, which presumably every process should expect to have:

Integer value   Name
       0        Standard input (stdin)
       1        Standard output (stdout)
       2        Standard error (stderr)

So instead STDIN_FILENO you can use 0.

Edit:
In Linux System you can find this using following command:

$ sudo grep 'STDIN_FILENO' /usr/include/* -R | grep 'define'
/usr/include/unistd.h:#define   STDIN_FILENO    0   /* Standard input.  */

Notice the comment /* Standard input. */

Solution 2

From the man read:

#include <unistd.h>
ssize_t read(int fd, void *buf, size_t count);

Input parameters:

• int fd file descriptor is an integer and not a file pointer. The file descriptor for stdin is 0

• void *buf pointer to buffer to store characters read by the read function

• size_t count maximum number of characters to read

So you can read character by character with the following code:

char buf[1];

while(read(0, buf, sizeof(buf))>0) {
   // read() here read from stdin charachter by character
   // the buf[0] contains the character got by read()
   ....
}

Comments

[Grijesh Chauhan]

your should read in read() manual

[Mats Petersson]

Read will do this just fine, but depending on exactly what you want to do, you may find that you have to do more than just call read - can you post your code, and explain exactly which part you have problems with?

[Mike]

I agree with Mats, what excatly are you looking for here? Where is the problem? There are also a lot of different examples of how to do this on SO, do you do any searching before asking the question?

[Grijesh Chauhan]

You could use int fileno(FILE *stream) first

[Mike]

@Grijesh - You can take it one step further, stdin is going to be 0, so you can just read(0,... Don't need to search around for stdin's file number on *NIX systems

[Grijesh Chauhan]

@Mike Yes I stdin = 0, stdout = 1, stderr = 2 and good is its same for all OS.

[Grijesh Chauhan]

@MOHAMED wait don't delete your answer., after some time I would add some details.

[MOHAMED]

@GrijeshChauhan I m not going to do it I updated it based on your remarks

[Grijesh Chauhan]

@MOHAMED Go ahead MOHAMED to improve your answer: add two points (1) you can just read using char c; read(0, &c, 1) , (2) read() function doesn't terminate input buffer with \0 with fgets() do. So in case if someone wants to use buffer as string he has to explicitly add null like: no_byte = read(0, buffer, sizeof(buffer)-1); buffer[no_bytes] = '\0'; This is the point where new guys do mistakes. like in this question read second error...go ahead this is what I wanted to add.

[weefwefwqg3]

why in the manpage, it uses the word "shall attempt". Is there any case that read will not read exactly the number of bytes specified by the 3rd argument? linux.die.net/man/3/read

[weefwefwqg3]

why in the manpage, it uses the word "shall attempt". Is there any case that read will not read exactly the number of bytes specified by the 3rd argument? linux.die.net/man/3/read

[camerondm9]

@weefwefwqg3 read() will read fewer bytes than requested if it is interrupted by a signal, or if there are fewer bytes available than requested, etc.