Thread: sscanf matching problems

I need to count the number of "timesteps" in a file (format shown below) I cound this by counting the number of lines that have 2 integers and a single double and nothing else in them using the code

Code:

int countTimeSteps(FILE *position)
{
  char line [MAXNUM]; //lines will be much less than 1000 characters, but it can't hurt
  int check,x,y,N_timeSteps;
  double z;
  N_timeSteps = 0;
  
  while (fgets(line, MAXNUM, position) != NULL) //read in file 1 line at a time
  {
    //puts(line);
    check = sscanf(line,"%d%d%lf",&x,&y,&z); 
    if (check != 0 && check != EOF) //if the line contains an integer value at the start
    {
      ++N_timeSteps;
    }
  }
  return N_timeSteps;
}

However, this seems to match every line.... Why is this not working?

Here is part of the file I am reading from:

Code:

5 5 0
 1           0           0           0           0           0           0
 1      5.7993      5.7993           0           0           0           0
 1      5.7993           0     -5.7993           0           0           0
 1      5.7993           0      5.7993           0           0           0
 1      5.7993     -5.7993           0           0           0           0
-1      2.0504           0           0           0           0           0
-1      7.8496      5.7993           0           0           0           0
-1      7.8496           0     -5.7993           0           0           0
-1      7.8496           0      5.7993           0           0           0
-1      7.8496     -5.7993           0           0           0           0
5 5 0.12
 1  2.9581e-06           0           0  1.1595e-06           0           0
 1      5.7993      5.7993           0  1.4193e-06 -6.9289e-08           0
 1      5.7993 -1.3264e-23     -5.7993  1.4193e-06 -3.2084e-24  6.9289e-08
 1      5.7993 -1.3264e-23      5.7993  1.4193e-06 -3.2084e-24 -6.9289e-08
 1      5.7993     -5.7993           0  1.4193e-06  6.9289e-08           0
-1      1.2922           0           0    -0.29463           0           0
-1      6.9823      5.8662           0    -0.33479    0.021945           0
-1      6.9823           0     -5.8662    -0.33479           0   -0.021945
-1      6.9823           0      5.8662    -0.33479           0    0.021945
-1      6.9823     -5.8662           0    -0.33479   -0.021945           0
5 5 0.23
 1  9.5596e-06           0 -1.7237e-23  1.5282e-06           0 -1.6028e-23
 1      5.7993      5.7993 -2.6508e-24  1.6843e-06  1.5394e-08 -6.4124e-24
 1      5.7993  -3.117e-23     -5.7993  1.6843e-06 -1.3807e-40 -1.5394e-08
 1      5.7993  -3.117e-23      5.7993  1.6843e-06 -1.3807e-40  1.5394e-08
 1      5.7993     -5.7993 -2.5042e-24  1.6843e-06 -1.5394e-08 -6.0848e-24
-1    -0.37038           0  1.9004e-18    -0.38423           0  1.5304e-18
-1      5.1619      5.9365  2.5342e-18    -0.39775   0.0021853  2.6822e-18
-1      5.1619           0     -5.9365    -0.39775           0  -0.0021853
-1      5.1619           0      5.9365    -0.39775           0   0.0021853
-1      5.1619     -5.9365  2.5358e-18    -0.39775  -0.0021853  2.6835e-18

The function should return 3, but it is returning the number of lines instead. Why is this happening? I can see why it is matching the second line, perhaps, but not all of them...

So: Is there a function that I can se to check if a line contains ONLY
"[integer] [integer] [double]"? and nothing else?

Think I got it sorted

Code:

int countTimeSteps(FILE *position)
{
  char line [MAXNUM]; //lines will be much less than 1000 characters, but it can't hurt
  int check,N_timeSteps;
  double x;
  N_timeSteps = 0;
  
  while (fgets(line, MAXNUM, position) != NULL) //read in file 1 line at a time
  {
    //puts(line);
    check = sscanf(line,"%lf%lf%lf%lf%lf%lf%lf",&x,&x,&x,&x,&x,&x,&x); 
    if (check == 3) //if the line contains an integer value at the start
    {
      ++N_timeSteps;
    }
  }
  return N_timeSteps;
}

But if anyone has suggestions for a more elegant solution I am all ears

And I would still like an explanation of why the first code was matching all lines in the file - I must be misunderstanding the way sscanf works.

Originally Posted by tabstop

Each of those lines starts with an integer and an integer and a double. (Print out the results and you'll see what I mean.)

Code:

 1  2.9581e-06

int 1, int 2, double .9581e-06

I don't accept 0 as a double. 0.0 is a double. It has to have a decimal point in it.

Now your job is very easy, every line with only one decimal point in it, gets counted.

If you're looking for integers, don't you want to use %d, instead of %lf?

%lf will make the ints into doubles, I was only interested in the return value of sscanf there - if it was 3, it means three numbers, which means another timestep. For long lines it would return 7. The one decimal place idea would probably work, but would be less efficient since it involves counting characters in line. Not much of a difference, but it's there ^_^

You could try this.

Code:

  while (fgets(line, MAXNUM, position) != NULL) //read in file 1 line at a time
  {
    int pos;
    check = sscanf(line,"%d%d%lf%n",&x,&y,&z,&pos);
    if ( check == 3 && line[pos] == '\n' )
    {
      ++N_timeSteps;
    } 
  }

Note that %n is NOT counted as a conversion, so it's still == 3
This assumes an immediate \n at the end of the line. Refine it if you have trailing whitespace.

> probably not the best idea to declare a variable inside a loop though ^_^
Why not?
You don't need it to have any greater scope do you?

How so?

Sure, if the object in question was a bloated C++ class with an expensive ctor and dtor, then you would have a point.

But this is C, and it is an int.
Unless your compiler is particularly brain-damaged, the space for that variable is allocated just ONCE when the function is called. It is not created and destroyed each time around the loop.