Thread: scanf problem

First of all, instead of posting your (short) code as an attachment, post it inline with code-tags. It makes it easier to see (don't have to click again, which makes the likelyhood of an answer better).

Code:

char c = 0;

scanf("%d", &c);

 if((c==32)||(c<48)||(c>57)) // ASCII table
 {
          printf("Invalid input\n");
 }

The %d format is for reading integers, your data is a char. That will lead to some "undefined behaviour", which is a bad thing. For one thing, it will overwrite whatever is after teh char variable with the remaining bytes to the sizeof(int) [usually 4 - so 3 bytes of "overwritten data"].

You should use %c for reading chars. Or use getchar().

Second, it is redundant to check if your input is "space" (32) or less than 48 - since the former is true when the latter is true.

I would also prefer to see

Code:

if (c < '0' || c > '9')
...

It saves someone who isn't perfectly up to date with the ascii table to understand what is being checked, as well as making it portable to machines that doesn't specifically use the ASCII character set [this is less of an issue on modern systems, as EBCDIC and other non-ASCII character sets are fairly rare nowadays].

--
Mats

why not use isprint() ?

Originally Posted by zacs7

why not use isprint() ?

You mean "isdigit()" I suppose?

--
Mats

isprint(), isdigit() an dother "is" functions take a character as an argument [1], and tell you if the argument "is a digit" or "is a printable character" [so, it's not a control character for example]. There are about 10 or so different functions that do similar things. It still means that you need to use getchar() or some such to read the character in.

The "is" functions are excellent for checking if your input "is valid" - if it's a digit, an alpha, alphanumeric, hex digit, etc, etc. They also work on "any character set".

[1] Technically, the function allows an integer as input - this is so that it's safe to use this funciton on EOF [which is -1]. However, for all intents and purposes, see the input as a char.

--
Mats

>You mean "isdigit()" I suppose?
Yes... my bad. At least you got the point, and in the end that's all that really matters eh?

Just trying to justify my error

I'm not sure what you are actually asking: I presume that what you are really looking for is either:
1. A way to read an integer, checking each character to see if the character is valid.
2. Check if what you read (using integer format to scanf()) was accepted by scanf().

The second option is the easier one:
The function scanf() will return "the number of correctly entered fields", so if you do

Code:

  int n;
  int r;
  r = scanf("%d", &n);
  if (r == 1) {
     // Something got accepted as a number. 
  } else {
     // Error, nothing accepted. 
  }

The above code will not accept "abc", but WILL accept "1234abc" as 1234, leaving "abc" in the input buffer. Another scanf() trying to read an integer from the input buffer of "abc" will not accept the further input of "abc".

Alternatively, you use getchar() to read each character, and you "build" the number up yourself. This is possibly what you are supposed to do.

Code:

   int c;
   int n;
   do {
      c = getchar();
      if (isdigit(c))   {
          // form a number
      } else if (c == '\n') {
         // done - but may be "empty" number. 
      } else {
          // Error - something that wasn't a digit entered.
      } 
    } while (moretodo);

This will ensure that ALL the input was accepted, and that there was no input beyond the "number".

Note that BOTH will require you to also remove any "garbage" from the input buffer if the input isn't accepted.

--
Mats

So, the decision is what constitutes "non-integer number".

You will in one form or another, need to convert a string of characters into a number - but scanf() will in the "simple" case do that for you. However, that assumes we accept "1234abc" as a valid input, as there's no trivial check we can do to detect this particular error.

If we have to print an error for "1234abc", then the most immediate solution is to read a character at a time - this character will have the ASCII code of one digit for each input, so although the data would be 1-byte, and have a range of -128..127 (or unsigned char 0..255), but the data you will get is a single digit '0'..'9' each time. So you will need to do a conversion into an integer - consider this as the same task as if you where given a number at a time and had to come up with what the whole number is. It is not very hard to do, but involves a little loop.

Once you have an integer, you can easily check if it's above 3000 and print an error.

--
Mats