How to ignore floating number in scanf("%d")?

Question

If user enters floating number for an integer variable I want to print invalid input. is that possible?

 int a;
 scanf("%d",&a); // if user enters 4.35 print invalid input 

I have tried for characters like this

  if(scanf("%d",&a)==1);
  else printf("invalid input");

But how to do for floating numbers. If user enters 4.35 it truncates to 4 but I want invalid input.

Answer 1

Since the start of a floating point number with any digits before the decimal point looks like an integer, there is no way to detect this with %d alone.

You might consider reading the whole line with fgets() and then analyzing with sscanf():

int a;
int n;
char line[4096];
if (fgets(line, sizeof(line), stdin) != 0 && sscanf(line, "%d%n", &a, &n) == 1)
   ...analyze the character at line[n] for validity...

(And yes, I did mean to compare with 1; the %n conversion specifications are not counted in the return value from sscanf() et al.)

One thing that scanf() does which this code does not do is to skip blank lines before the number is entered. If that matters, you have to code a loop to read up to the (non-empty) line, and then parse the non-empty line. You also need to decide how much trailing junk (if any) on the line is tolerated. Are blanks allowed? Tabs? Alpha characters? Punctuation?

Answer 2

You'll have to read it as a double and then check if it is an integer. The best way to check if it is an integer is to use modf, which returns the decimal portion of the double. If there is one you have an error:

double d;
scanf("%lf", &d);

double temp;
if(modf(d, &temp)){
  // Handle error for invalid input
}

int a = (int)temp;

This will allow integers or floating point numbers with only 0s after the decimal point such as 54.00000. If you want to consider that as invalid as well, you are better off reading character by character and verifying that each character is between 0 and 9 (ascii 48 to 57).

Answer 3

This can not be done with out reading pass the int to see what stopped the scan.

Classic idiom

char buf[100];
if (fgets(buf, sizeo(buf), stdin) == NULL) {
  ; // deal with EOF or I/O error
}
int a;
char ch;
if (1 != sscanf(buf, "%d %c", &a, &ch)) {
  ; // Error: extra non-white space text
}

Answer 4

You can do it using strtol() and strtod() and comparing the end pointers, e.g. this:

#include <stdio.h>
#include <stdlib.h>

int main(void) {
    char buffer[100];
    char * endptr_n;
    char * endptr_d;
    long n;
    double d;

    fgets(buffer, 100, stdin);
    n = strtol(buffer, &endptr_n, 10);
    if ( endptr_n == buffer ) {
        fputs("You didn't enter a number.", stderr);
        return EXIT_FAILURE;
    }

    d = strtod(buffer, &endptr_d);

    if ( *endptr_d == '\0' || *endptr_d == '\n' ) {
        if ( endptr_d == endptr_n ) {
            puts("You entered just a plain integer.");
        } else {
            puts("You entered a floating point number - invalid.");
        }
    } else {
        puts("You entered garbage after the number - invalid.");
    }

    return EXIT_SUCCESS;
}

outputs:

paul@local:~/src/c$ ./testint
2
You entered just a plain integer.
paul@local:~/src/c$ ./testint
2.3
You entered a floating point number - invalid.
paul@local:~/src/c$ ./testint
3e4
You entered a floating point number - invalid.
paul@local:~/src/c$ ./testint
4e-5
You entered a floating point number - invalid.
paul@local:~/src/c$ ./testint
423captainpicard
You entered garbage after the number - invalid.
paul@local:~/src/c$

It doesn't use scanf(), but that's a good thing, and it avoids the need to manually check the input following the integer you read.

Obviously, if the only thing on the line is the number, then a lot of this becomes unnecessary, since you can just call strtol() and check *endptr_n immediately, but if there may be other stuff on the line this is how you can do it, e.g. if you want to accept an integer followed by anything non-numeric, but not a floating point followed by the same thing, you can just remove the if ( *endptr_d == '\0' || *endptr_d == '\n' ) logic.

EDIT: updated the code to show the check to *endptr.

Answer 5

This one is bit easier:

#include <stdio.h>

int main() 
{
    int a;
    long double b;

    scanf("%f",&b);
    a = (int) b;

    a == b ? printf("%d\n",a) : printf("Invalid input!");

    return 0;
} 

Input: 4
Output:

4

Input: 4.35
Output:

 Invalid input

Answer 6

Here's an easy way:

#include <stdio.h>

int main(int argc, char **argv) {
    int d;
    printf("Type something: ");

    // make sure you read %d and the next one is '\n'
    if( scanf("%d", &d) == 1 && getchar() == '\n' ) {
        printf("%d\n", d);
    }

    return 0;
}

.

$ a.exe
Type something: 312312.4214

$ a.exe
Type something: 2312312
2312312

$ a.exe
Type something: 4324.

$

Answer 7

First of all, there is nothing wrong with scanf. When a user enters a float then they actually type in a number dot number. So, code a scanf to detect that data entry.

main()
  {
       char c1[2];
       int num1;

       int nr_nums;

       nr_nums = scanf("%d%1[.e0123456789]", &num1, &c1);

       if (nr_nums == 1) {printf("\ndata = %d", num1);}
       if (nr_nums == 2) {printf("\nInvalid");}

 }

Modified this code per another possible data entry format of 1. or 3e-1 as suggested by a comment.

This code gets to the basics of your requirement. It accepts Integer data entry and detects when a float is entered.

Answer 8

If you have your number represented as a string (when you have used fgets) you can run a for loop through it and compare each character to '.'.

Answer 9

One other option I can see follows:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int main()
{

 char mystring[31];
 scanf("%30s[0-9]\n", mystring);
 int mynumber = atoi(mystring);
 printf("here is your integer: %d", mynumber);
 getchar();
 getchar();
 return 0;











 }