how does scanf() check if the input is an integer or character?

I am wondering how does the standard C library function scanf() check if the input is an integer or a character when we call scanf("%d",&var) when a character itself is just a number? I know that when it encounters a non-integer it puts it back into the开发者_如何学Go input buffer and returns a -1 but how does it know that the input is not an integer?

You're correct in that each character is really represented as an 8-bit integer. The solution is simple: look at that number, and see if it is in the range 48-57, which is the range of SCII codes for the characters '0' - '9'.

Starting on line 1315 of the scanf() source code we can see this in action. scanf() is actually more complicated, though - it also looks at multi-byte characters to determine the numeric value. Line 1740 is where the magic happens and that character is actually converted into a number. Finally, and possibly this is the most useful, the strtol() function does the looping to perform that conversion.

The input is always a string. If scanf is expecting an integer (because you passed it "%d"), it attempts to convert the string to an integer for you.

In scanf statement you are mentioning %d which is holding integer only so by default it understood that the variable as integer

Basically, the scanf function matches regular expressions based on the conversion specifier you pass. If you specify %d, that tells scanf to match input against a regular expression that's one or more characters between '0' and '9' (optionally with a leading + or - character). It then converts that sequence of characters to the equivalent integer value.

A very simplistic version might look something like this:

while (isdigit(c = fgetc(stream))
  val = val * 10 + valueOf(c); 
ungetc(c, stream);

where isdigit is a standard library function that returns true (non-zero) if the character value represents a decimal digit, and valueOf is a user-defined function that maps the character representing an integer ('0' - '9') to the equivalent integer value (0 - 9) (I'm not aware of a standard library function that does that for individual character values). Why not just subtract '0' from c to get the equivalent integer value? Depending on the encoding, it's not guaranteed that all decimal integer characters will be laid out in order (All The World Is Not ASCII); best to delegate the actual conversion to a function that's aware of the current encoding.

It tries to parse a valid decimal integer from the beginning of the stream, in practice optional + or - and then sequence of one or more digits.

It basically uses a loop like:

while (isdigit((ch=getchar()))
    // ...

You misunderstand the %d format specifier. It does not return the ASCII value of the entered character (you can use %c for that); it returns the integer that the characters represent. So, on an input character of '9', %d returns the value 9 - not the ASCII value of '9'.

(Well, %d actually looks at a sequence of characters, so if the input is '9' followed by '0' followed by ' ', it'll interpret that as 90).