parsing input with scanf in C

Question

I've been having a lot of problems trying to figure out how to use scanf. It seems to work fine with integers, being fairly straight forward scanf("%d", &i).

Where I am running into issues is using scanf in loops trying to read input. For example:

do {
  printf("counter: %d: ", counter);
  scanf("%c %c%d", &command, &prefix, &input);

} while (command != 'q');

1) When I enter in a validly structured input like c P101,it seems to loop again before prompting me. This seems to happen even with a single-

scanf("%c", &c)

in a while loop. It'll do the loop twice before prompting me again. What is making it loop twice, and how do I stop it?

2) When I enter in less amount of input that programmatically wouldnt have another character or number such as q, pressing enter seems to prompt me to enter more. How do get scanf to process both single and double character entries?

Answer 1

For question 1, I suspect that you've got a problem with your printf(), since there is no terminatin "\n".

The default behavior of printf is to buffer output until it has a complete line. That is unless you explicitly change the buffering on stdout.

For question 2, you've just hit one of the biggest problems with scanf(). Unless your input exactly matches the scan string that you've specified your results are going to be nothing like what you expect.

If you've got an option you'll have better results (and fewer security issues) by ignoring scanf() and doing your own parsing. eg. use fgets() to read an entire line into a string, and then process the individual fields of the string.

Answer 2

scanf sucks, for the reasons you've already discovered. It's much better to use something to get the whole line and then parse it.

Answer 3

When you enter "c P101" the program actually receives "c P101\n". Most of the conversion specifiers skip leading whitespace including newlines but %c does not. The first time around everything up til the "\n" is read, the second time around the "\n" is read into command, "c" is read into prefix, and "P" is left which is not a number so the conversion fails and "P101\n" is left on the stream. The next time "P" is stored into command, "1" is stored into prefix, and 1 (from the remaining "01") is stored into input with the "\n" still on the stream for next time. You can fix this issue by putting a space at the beginning of the format string which will skip any leading whitespace including newlines.

A similiar thing is happening for the second case, when you enter "q", "q\n" is entered into the stream, the first time around the "q" is read, the second time the "\n" is read, only on the third call is the second "q" read, you can avoid the problem again by adding a space character at the beginning of the format string.

A better way to do this would be to use something like fgets() to process a line at a time and then use sscanf() to do the parsing.

Answer 4

Once you have the string that contains the line. i.e. "C P101", you can use the parsing abilities of sscanf.

See: http://www.cplusplus.com/reference/clibrary/cstdio/sscanf.html