Question

As the title says, I need to write a small program to read data from standard input, sort it, and send it to standard output. The program should take 1 argument that tells it how long is one record (in bytes). Here’s how I test it:

printf 'D\x00C\x00\x00B\x00A' | ./binsort 2 | od -c 

The above should output something like:

0000000  \0   A  \0   B   C  \0   D  \0 0000010 

Here’s what I have so far (binsort.c):

#include <stdio.h> #include <stdlib.h> #include <limits.h>  using namespace std;   void print_usage() {         printf('%s\n', 'Usage: '); }   int compare (const void * a, const void * b) // the compare function for qsort... might need some work {   return ( *(int*)a - *(int*)b ); }   int main(int argc, char *argv[]) {         if (argc != 2 || stdin == NULL) // check for argument and piped input         {                 print_usage();                 exit(EXIT_FAILURE);         }          int entry_size = atoi(argv[1]);          if (entry_size <= 0 || entry_size >= INT_MAX) // check for valid range of entry size         {                 print_usage();                 exit(EXIT_FAILURE);         }          char *input = new char[entry_size]; // to hold single record          while (fgets(input, entry_size, stdin) != NULL)         {                 printf('%s', input); // output single record we just read         }          exit(EXIT_SUCCESS); } 

Then compile with g++ binsort.c -o binsort.

The above compiles but doesn’t output the data that printf sent to it. It should output it in 2-byte chunks… like D\0 C\0 \0B \0A… but it doesn’t.

I’m thinking of using qsort to do the sorting on a malloc/realloc-allocated array. However, I never had experience with these, in effect banging my head on this little utility for a few days. Anyone can help?

P.S. People asked if this is a homework assignment… It’s not – developers at my company want to use this to debug output of their project.

Answer 1

Don’t use scanf() and printf(). Those are supposed to be used with text data. Since you’re dealing with binary data, you instead want to use the lower-level fread() and fwrite() functions. Since you don’t know how much total data there is, you’ll have to use a dynamic data structure (such as a resizeable array) to store the input. You can’t process the data on-line — you have to read it all in, sort it, then write it back out. Your sort function is also wrong — you’re only comparing the first 4 bytes of each record, which is wrong if the record size is anything other than 4 bytes. Use memcmp() instead.

This should work:

char *input = NULL; size_t input_size = 0; int num_items = 0; int entry_size;  int compare_func(const void *e1, const void *e2) {   return memcmp(e1, e2, entry_size); }  int main(int argc, char **argv) {    // ...   char *datum = malloc(entry_size);  // check for NULL   input_size = 4096;   input = malloc(input_size);  // check for NULL    while(1)   {     if(fread(datum, 1, entry_size, stdin) < entry_size)       break;     size_t new_size = (num_items + 1) * entry_size;     if(new_size > input_size)     {       input = realloc(input, input_size * 2);  // check for NULL       input_size *= 2;     }     memcpy(input + num_items * entry_size, datum, entry_size);     num_items++;   }    qsort(input, num_items, entry_size, compare_func);    fwrite(input, entry_size, num_items, stdout);    return 0; }