Question

I have a program that accept orders by reading commands from a file.
In this file some commands are “float string”, like “1.0”,”2.0″, but they are invalid, what the program need is integer, like “1”,”2″. So, how can I make the program understand the commands like “1.0” is invalid? Is there any neat way to do this?

    char buf[CMDSIZE];
    if(fgets(buf, CMDSIZE, stdin))      //buf likes this: "1.0 \n"
    {
        *prio = 1; *command = -1; *ratio =1.0;
        // I need to make sure that command is not "1.0" or something like this              
        sscanf(buf, "%d", command);
        switch(*command){....blahblah......}
    }

Thank you.

Answer 1

It’s easier to use strtol.

This will parse a base-10 integer. The pointer e will point to the first character after the integer. You can check to make sure it’s a NUL byte and signal an error otherwise. (You also have to check that the input isn’t empty.)

If you want to allow spaces / newlines after the number, you can do that too. Note that strtol eats leading whitespace — but not trailing whitespace.

long v;
char *e;
v = strtol(buf, &e, 10);
if (!buf[0] || *e) {
    error();
}
// v has number

Footnote: Checking for overflow and underflow with strtol is a little weird. You have to set errno to 0 first, call strtol, then check if the result is LONG_MIN or LONG_MAX and if errno is set to ERANGE.