C将字符串/字符输入转换为浮点数

Question

I'm writing a program in C, and one of the requirements is that the user input a number as a string value, and then the program must convert it to floating point format, and then do various things to that value. I can't seem to figure out how to simply convert the input string into a number. How would I write the first chunk of code like that? I know there are posts similar to this, but they haven't been helpful. I need to do this fairly simply, and I'd like to not have to #include anything besides ...

#include <stdio.h>

int main(int argc,char* argv[]) { 
  /*having issues with this in particular...*/

  int number;    
  int newNumber;
  int i;
  printf("Enter a number in decimal...");
  scanf("%d",&number);

  /*                        */

  printf("%d in binary is: ",number);
  for(i = 31;i >= 0;i--) {
    newNumber = (number>>i);
    if(newNumber & 1) {
      printf("1");
    }
    else {
  printf("0");
}
   }  


  return 0;
}

Thanks!

Answer 1

You can use strtod() . Read the number into a buffer as a string (say with fgets() ), and then do:

double x;
char *endptr;
errno = 0;
x = strtod(buffer, &endptr);
if (endptr == buffer) {
    //... parse error or empty input ...
} else if (*endptr != '\n' && *endptr != '\0') {
    //... parse error extraneous data ...
} else if ((x == HUGE_VAL || x == -HUGE_VAL) && errno != 0) {
    //... error overflow ...
} else if (x <= DBL_MIN && x >= -DBL_MIN) {
    if (errno != 0) {
        //... error underflow (detected) ....
    } else {
        // ... underflow still possible, but is undiagnosed ...
    }
}

Error checking is done by checking both the value returned for x , and if errno got set. Parse errors is done by checking the endptr . The C standard says underflow detection is implementation defined (C.11 §7.22.1.3 ¶10).

Answer 2

Following up on the comment I made:

Your code appears to "work" - in that it reads the input and produces output. You might consider adding \\n at the end of your printf statements. What exactly isn't working for you? Consider using 8*sizeof(int)-1 rather than 31 . You don't know ahead of time how big an int is on your system.

With a tiny change your code works very nicely for me:

#include <stdio.h>

int main(int argc,char* argv[]) {
  /*having issues with this in particular...*/

  int number;
  int newNumber;
  int i;
  printf("Enter a number in decimal...\n"); // <<< added a newline
  scanf("%d",&number);

  /*                        */

  printf("%d in binary is: ",number);
  for(i = sizeof(int)*8 - 1;i >= 0;i--) {  // make sure `i` starts with the right size
    newNumber = (number>>i);
    if(newNumber & 1) {
      printf("1");
    }
    else {
      printf("0");
    }

  }

  printf("\n");  // <<< print a newline

  return 0;
}

When I run it:

Enter a number in decimal...
123
123 in binary is: 00000000000000000000000001111011

note - you do have to input an integer for this to work. If you need your code to be more robust to user "error" in the input, your best bet is to read the input as a string, then do some more parsing. For example:

char buffer[100];
printf("enter an integer:\n");
fgets(buffer, 100, stdin);
// now you can look for spaces, letters, anything you want to skip, in the string
// you could even use sscanf().
// scanf() is a terribly fragile function to use, especially with "free form user input".
if(sscanf(buffer, "%d", &number) == 1) {
  // successfully converted number... run your code
} else {
  printf("unable to convert input to a number!\n%s\n", buffer);
  return 0;
}

another note re-reading your question, you said "program has to convert to a floating point number. This means you should do

float fnumber;
sscanf(buffer, "%f", &fnumber);

or

double dnumber;
sscanf(buffer, "%lf", &dnumber);

to do the conversion. But then, if you want to print as a binary, you need to cast the number from floating point to unsigned integer - a bit shifting operation is not well defined for a floating point number. So

unsigned int *intPtr, number;
intPtr = (unsigned int*) *fnumber; // this is hairy but it works
number = *intPtr;

and now use number as before - you will be able to print out the binary equivalent of the floating point number. There are people who will complain that the above is not "true to the standard". They might prefer it if you created a union:

union f2i
{
  float fvalue;
  unsigned int ivalue;
} Values;

Then assign the input to Values.fvalue , and use Values.ivalue to print out the binary number. It is still a hack...

Answer 3

Simple scanf example tested with GCC 4.7.3

$ gcc -Wall -Wextra -pedantic -std=c99 cstring-double.c

#include <stdio.h>

int main() {
  double v;
  int err;

  err = scanf("%lf", &v);
  if (1 == err) {
    printf("%lf\n", v);
  } else {
    printf("read failed\n"); }

  return 0; }

Answer 4

Just use atof. cplusplus documentation