Question

I’d really like your help with the following problem.

I need to read N integers (in binary format) from two two binary files, and to write to another output file all the common integers(again in a binary format) in O(nlogn) time.

Here’s a sketch of what I want to do:

1-I know that I can’t use fscanf with %d, since these are not text files, but since I know that there are N numbers in each file, I want to run in a ‘for’ loop N times and

for (i=0; i<N; i++) {
    fread(&int1, 4, 1, file1);
    fread(&int2, 4, 1, file2);
    arr1[i]=int1;
    arr2[i]=int2;
} 

Where arr1, arr2 are int arrays of N size each, Is this correct? If I know that there are N integers where every integer represented by 4 bits, I would finish at the end of the file, Am I?

2- I want to use qsort for sorting each array with nlogn steps, but then how should I compare between the two with still nlogn steps? I didn’t get to any smart way.

Answer 1

Your code is mostly correct, except it will not behave correctly when it reaches end-of-file. fread does not detect EOF until you try to read past the end, so you should test the return value in case it reaches EOF or an error occurs, e.g.:

if(fread(..., file1) < 1 ||
   fread(..., file2) < 1)
{
    // Error or EOF occurred
    break;
}

You could also just slurp the entire arrays in a single fread call instead of looping, since the data is binary:

int arr1[20], arr2[20];
if(fread(arr1, sizeof(arr1[0]), 20, file1) < 20 ||
   fread(arr2, sizeof(arr2[0]), 20, file2) < 20)
{
    // Error or early EOF occurred
}

For your second question, just place all of the integers into a single flat array. You can either do this by reading them directly into a single array, or copy them into one array after you’ve read in all of the data. Then qsort the combined array.