Validate double number

Hi gurus

I need to validate an input from the user and to make sure is a valid number, I've seemed multiple threads on integer but I really need to validate a double

123.34 or 345.23 this kind of number and make sure is a valid number


Thanks a bunch gurus

Code:

---

if (strtod(buf, NULL) == 0 && buf[0] != '0'); /* Invalid number - not a double! */

---

Code:

---

if (strtod(buf) == 0 && buf[0] != '0') /* Invalid number - not a double! */ ;

---

fixed. :)

Hehehe. Typically I just put an empty if statement since I have nothing to put in the if. Just to demonstrate :)

What about -123.34?

well, bad example. I guess the only one that would fail would be -0.

Thanks gurus

Is there a generic validation in case the user goes as follow

123.45
123.00
123
-123.45

Thanks for battling with me here

Elysia's example will work for all those values. It's a generic validation. Do you understand why / how it works?

It will only fail for -0, but who in their right mind types that? A little additional logic can fix that, though. Also forgot that strtod takes two arguments. Updated reply.

Quote:

---

Originally Posted by robwhit

Code:

---

if (strtod(buf) == 0 && buf[0] != '0') /* Invalid number - not a double! */ ;

---

fixed. :)

---

Aha! Checked the docs:

Code:

---

if (strtod(buf, NULL) == 0 && buf[0] != '0') /* Invalid number - not a double! */ ;

---

edit: ARRGHGH! beaten.

If you are really picky about detecting -0, this will work:

Code:

---

if ( strtod(buf, NULL) == 0 && (buf[0] != '0' && (buf[0] != '\0' && buf[0] != '-' && buf[1] != '0')) ) /* Not a double! */;

---

Not really T but i case you fill like a brushing your teaching skills with me andexplain I can alway search for it and learn how it works

thanks again gurus, I though C was going to be easy!!!

What makes you think it won't accept -0? Or any negative number?

What the code does is first convert the inputted string (buf) to a double with a call to strtod. Then it checks if the result is 0. Strtod returns 0 in any of the following situations: the string is not a double, or not a number, or the user entered 0. So then I proceed to check the first element in the inputted string (buf[0]) and check if it's '0', too. If it is, then the user entered 0. If it isn't '0', then the user entered a non-number.

The more complicated example first checks if buf[0] (first input character) is not '0' and also checks if buf[0] is not '-' and the second character, buf[1] is not '0'. That would mean something like, if the first two characters aren't "-0".
So in essence, if the converted double is 0 AND the first character entered is NOT '0' AND the first two entered character aren't "-0" THEN it's not a valid double.

Quote:

---

Originally Posted by robwhit

What makes you think it won't accept -0? Or any negative number?

---

The first example won't accept -0 because it's the same as 0, so strtod(buf, NULL) == 0 is true BUT buf[0] is NOT 0, it's actually -, so it assumes it's not a double and fails.
The second example fixes that.
It does accept negative numbers, however.

So ely (Buf) is my user value entry? what does the Null do, now that you are on the roll

Thanks a bunch again

I assume buf is your buffer - the string you read from the user:

Code:

---

char buf[255];
fgets(buf, sizeof(buf), stdin);

---

Null as in NULL?
Like if (p == NULL)?

Ah I see. Thanks for the explanation. I was thinking of just strtod.

However, if I may:

Code:

---

if ( strtod(buf, NULL) == 0 && (buf[0] != '0' && ((buf[0] != '\0') && buf[0] != '-' && buf[1] != '0')) ) /* Not a double! */;

---

?

Ely

does the same function works for floats also as well?

Quote:

---

Originally Posted by robwhit

Ah I see. Thanks for the explanation. I was thinking of just strtod.

However, if I may:

Code:

---

if ( strtod(buf, NULL) == 0 && (buf[0] != '\0') && (buf[0] != '0' && (buf[0] != '-' && buf[1] != '0')) ) /* Not a double! */;

---

?

---

Yes! That might be safer!

Quote:

---

Originally Posted by guaro555

Ely

does the same function works for floats also as well?

---

Yes, it works for floats as well, just change strtod to strtof.

Well, just call me a fly in the ointment.

This will fail the test and be called valid: 0DLFKD

It converts to zero since it's garbage, and the first char is zero, so it would be accepted.

I know, I know. I'm just stirring the pot.

Todd :)

Fine! :p
How about this?

Code:

---

if (d == 0 && strcmp(buf, "0") != 0 && strcmp(buf, "-0") != 0) /* Not a double! */;

---

You really are doing your best to thwart my optimization, aren't you? :p

Really, to get the most out of strtod you have to take advantage of all of it's arguments. Knowing where strtod stopped its conversion helps you decide what is acceptable and not acceptable input. Dave Sinkula has illustrated the proper technique in a series of snippets.

My understanding is that strtod works for floats just fine:

float fltpt = (float) strtod( example, &endpt );

http://www.daniweb.com/code/snippet597.html

I suppose... I'm used to atof/atod/atoi, which only takes one. Hmmm.

Ah you caught my post before I saw my error....

the strcmp won't work because there might be more characters afterwards.

lol we are such a waste of bandwidth.

Code:

---

char* pEnd;
double d = strtod(buf, &pEnd);
if (pEnd - buf == 0) /* "Not a double!" */;

---

That should work fine, too.

Quote:

---

Originally Posted by robwhit

Ah you caught my post before I saw my error....

the strcmp won't work because there might be more characters afterwards.

lol we are such a waste of bandwidth.

---

Not sure I get what you mean with strcmp won't work since d will ONLY be 0 if the user entered "0" or "-0" or a non-string, so what's the error?

I'm assuming it's part of a stream.

I'm beginning to have brain problems... :p

Still unsure :o
How do you expect it to fail? Example, please?
Unless it's part of your "problems"? ;)

for example, you read in a buffer from a file:

char buf[128];
fread(buf, sizeof(buf), 1, file);

then you try to see if it contains a double.

oops, what if there is more data than the double? It will not have a \0, and so it will not compare equal.

Quote:

---

Originally Posted by Todd Burch

Well, just call me a fly in the ointment.

This will fail the test and be called valid: 0DLFKD

It converts to zero since it's garbage, and the first char is zero, so it would be accepted.

I know, I know. I'm just stirring the pot.

Todd :)

---

T

If I understood the statement correctly you have a few && in there so the first option is true as is garbage like you said and the second is true because is 0, so still not a valid double

That is correct. I'm working on some logic to validate a double now. Stay tuned. ;)

Todd

I think this works pretty good. It uses a regular expression to validate a double. It most likely does NOT illustrate best practices for regular expression use in C.

Code:

---

#include <stdio.h>
#include <regex.h>

int main (int argc, const char * argv[]) {
        char data[80] ;
        int rc ;
        regex_t * myregex ;
        regmatch_t arrayOfMatches[2];
       
        // Compile the regular expression
        rc = regcomp( myregex,
                "^ *[-+]?([0-9]+(\.[0-9]*)?|\.[0-9]+)([eE][-+]?[0-9]+)? *$",
                REG_EXTENDED | REG_NOSUB ) ;
        printf("RC from regcomp() = &#37;d\n", rc);
       
       
        printf("Enter a double value\n");
        scanf("%s",data) ;
       
        // Compare the entered value to the regex
        if (!regexec(myregex, data, 1 , arrayOfMatches, 0 ) ) {
                printf("double %s is valid.\n", data) ;
        }
        else printf("double %s is not valid\n", data );
        return 0;
}

---

Todd

Well, I think my regex is correct, but it reports 123e as being valid. Methinks the default RE support on a Mac might not be working properly.

Todd

It also reports --123 as being valid. I suppose this might be why folks recommend the perl compatible regular expression (pcre) library over the default lib.

Todd

Never mind, unless someone can explain the bogus results. It also says 123ee-4 is valid.

Right, so what's wrong with using this:

Code:

---

int main()
{
  char buf[40];
  char *endPt;
  double x;
  fgets(buf, sizeof(buf), stdin);
  buf[strlen(buf)-1] = '\0;'
  x = strtod(buf, &endPt);
  if (endPt != '\0') printf("Not a valid number\n");
  else printf("x = %f\n", x);
  return 0;
}

---

It's fairly efficient compared to using regular expressions, and it validates that the entire string is accepted by strtod.

--
Mats

Quote:

---

Originally Posted by matsp

Right, so what's wrong with using this:

Code:

---

int main()
{
  char buf[40];
  char *endPt;
  double x;
  fgets(buf, sizeof(buf), stdin);
  buf[strlen(buf)-1] = '\0;'
  x = strtod(buf, &endPt);
  if (endPt != '\0') printf("Not a valid number\n");
  else printf("x = %f\n", x);
  return 0;
}

---

It's fairly efficient compared to using regular expressions, and it validates that the entire string is accepted by strtod.

--
Mats

---

LOL. Other than it doesn't work, nothing! (I fixed the typo with the semi colon). I entered 123.45 and it says it's not valid.

Todd

This fixes it.

Code:

---

int main()
{
  char buf[40];
  char *endPt;
  double x;
  fgets(buf, sizeof(buf), stdin);
  buf[strlen(buf)-1] = '\0';
  x = strtod(buf, &endPt);
  if (*endPt != '\0') printf("Not a valid number\n");
  else printf("x = %f\n", x);
  return 0;
}

---

Ah, yes, the star was missing. I never actually TRIED it, I just wrote it "from the hip".

--
Mats

I think it is a clever approach - leveraging the second parm to strtod(). I found one case, that of just pressing enter, that didn't work. So, thanks to matsp, here's the final version. ;)

Code:

---

int main()
{
  char buf[40];
  char *endPt;
  double x;
  fgets(buf, sizeof(buf), stdin);
  buf[strlen(buf)-1] = '\0' ;
  x = strtod(buf, &endPt);
  if (*endPt != '\0' || strlen(buf)==0 ) printf("Not a valid number\n");
  else printf("x = %f\n", x);
  return 0;
}

---