compare.c

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <string.h>
#include <sys/errno.h>

#define BUFSIZE (1024*1024)

/* when you want to know what went wrong, or how it's different */

unsigned long long blocks_read; /* blocks read */
unsigned long long differences;

/* change this to take BUFSIZE from environment */
/* env options? bs= , copy to enviroment, update argv */

int main ( int argc , char * argv[])
{
  int numfiles;
  ssize_t bytes; /* bytes read, by read for file1 */
  int * files;
  {
    typedef char comparebuf[BUFSIZE];
    comparebuf * buffers;
    int i;

    numfiles=argc-1; /* number of files to compare with */
    if ( numfiles > getdtablesize() )
      fprintf(stderr,"Number of files, %d, exceeds, descriptor table size=%d\n"
              ,numfiles,getdtablesize());

    files=calloc(numfiles,sizeof(int));
    if ( files == NULL ) {
      perror("calloc, files:");
      exit(10);
    } else { /* open files to compare with, handles in files[] */
      int arg; /* position in argument list */
      files[0]=open( argv[1], O_RDONLY|O_NONBLOCK, 0123 /*file mode not used */ );
      if ( files[0] == -1 ) {  /* files[0] == first file */
        perror("open file1:");
        exit(11);
      }
      for( i = 1,arg=2; i<numfiles ; i++,arg++ ) {
        files[i]=open(argv[arg], O_RDONLY|O_NONBLOCK, 0 );
        if ( files[i] == -1 ) {
          perror("opening files:");
          exit(arg+10);
        }
      }
    }
    buffers=calloc(numfiles,sizeof(comparebuf));
    if ( buffers == NULL ) {
      perror("calloc, buffers:");
      exit(10);
    }

    blocks_read=0;
    differences=0;
    do {
      extern int errno;
      ssize_t rr;
      bytes=0;
      {
        register size_t nbytes=sizeof(comparebuf);
        char * buffer=buffers[0];
      TryAgain:
        rr=read(files[0],buffer,nbytes);
        if ( rr == -1 ) {   /* error ? */
          if ( errno ==  EAGAIN )
            goto TryAgain;
          perror("file1:");
          exit(11);
        } else if ( rr < nbytes && rr > 0 ) {  /* short read, try again */
          bytes+= rr;
          nbytes= nbytes - rr ;
          buffer= buffer + rr ;
          goto TryAgain;
        } else  /* rr == 0 || rr == nbytes,  EOF or complete read */
          bytes+= rr;
      }
      if ( bytes == sizeof(comparebuf) )
        blocks_read++;

      /* at this point, bytes, would should be == to sizeof(comparebuf), except
       * for, possibly,the last read of the file, and then, of course,0 for EOF
       */

      for(i=1;i<numfiles;i++) {
        ssize_t ibytes; /* bytes read, by read for file i */
        ibytes=0;
        {
          register size_t fbytes;
          char * buffer;

          fbytes=bytes;
          if ( fbytes <= 0 )
            fbytes= sizeof(comparebuf);
          buffer=buffers[i];
        FilesTryAgain:
          rr=read(files[i],buffer, fbytes);
          if ( rr == -1 && errno == EAGAIN )
              goto FilesTryAgain;
          else if ( rr == -1 ) {
            fprintf(stderr,"Error file %d %lld blocks read.\n",i+1,
                    blocks_read);
            perror("read files:");
            exit(i+11);
          } else if ( rr < fbytes && rr > 0) {
            /* short read, try to read the rest */
            ibytes+= rr;
            fbytes= fbytes - rr;
            buffer= buffer + rr;
            goto FilesTryAgain;
          } else
            ibytes+= rr;
        }
        if ( ( bytes == 0 && ibytes > 0 )
             || ( bytes != 0 && ( ibytes > bytes ))) {
          printf("File %d too long.\n",i+1);
          differences++;
          continue;
        } else if ( ibytes < bytes ) {
          printf("File %d EOF,     block %lld.\n",i+1,blocks_read);
          differences++;
        } else if (memcmp(buffers[0], buffers[i],bytes) ) {
          printf("File %d differs, block %lld.\n",i+1,blocks_read);
          differences++;
        }
      } /* end for loop */



#if 1
      if ( blocks_read % 128 == 0 ) {
        register int g;
        g = blocks_read / 1024 ;
        printf("  Thru Block %lld, GB %i.\n",blocks_read,g);
      }
#endif
    } while ( bytes > 0 );
    printf("%lld blocks read,  %lld differences.\n",blocks_read,differences);
    free(buffers);
    for(i=0;i<numfiles;i++)
      close(files[i]);
    free(files);
    return (differences != 0 ); /* so, == 0, returns 0 */
  }
}