program crashes when copying binary file

Hi,

I am trying to copy a binary file (band2) using the following program:
------------------

Code:

---

#include <stdio.h>
#include <stdlib.h>

void main()
{
    FILE *fp2, *fpr2;
    fp2 = fopen("C:\\band2", "rb");
    fpr2 = fopen("C:\\r2", "wb");
    long int i, size = 1249820;
    unsigned char *b2 = malloc (sizeof(*b2));
    unsigned char *r2 = malloc (sizeof(*r2));
    for(i=0, i<size, i++){
        fscanf(fp2, "%c", &b2[i]);
    }
    for(i=0, i<size, i++){
        fprintf(fpr2, "%c", b2[i]);
    }
    free(b2);
    free(b3);
}

---

---------------------
The program crashes every time I run it. Can someone pls point out to me the mistake?

Quote:

---

Originally Posted by bored_guy

Hi,

I am trying to copy a binary file (band2) using the following program:

The program crashes every time I run it. Can someone pls point out to me the mistake?

---

I don't understand the use of *b2 as an array. It's just a pointer to an unsigned char. Same with r2. No idea why you're malloc'ing one of each of them.

fscanf() is a function that converts data into the type you specify. Why use it for binary copying? Won't it muddle up your file's data?

Don't you want just a large buffer, and then read into it from the file, and write from it to the copy of the file, using the binary file read and writer?

You want to use fread() and fwrite().

You are malloc()ing 4 bytes (the sizeof a pointer) and are trying to write 1249820 byte into that location. Yes, Virginia, that will fail.

Since you don't need to do any character manipulation, read and write a chunk of data at a time instead of a character at a time.

As Adak said - you only need one buffer, not two.

Close your files when you are done with them.

the reason im using fscanf and fprintf is cos i need to do some mathematical manipulation with each and every number in the binary file, which are in unsigned char (each number runs anywhere from 0-255). but before all those manipulations come in, i wanted to just try copying the values into new files. Actually I thought a much simpler prog would suffice for copying files:

Code:

---

FILE *fp2, *fp3, *fpr2, *fpr3;
fp2 = fopen("C:\\band2", "rb");
fp3 = fopen("C:\\band3", "rb");
fpr2 = fopen("C:\\r2", "wb");
fpr3 = fopen("C:\\r3", "wb");

long int i, size = 1249820;;
unsigned char a2[size], a3[size];

for(i=0, i<size, i++){
  fscanf(fp2, "&#37;c", &a2[i];
  fscanf(fp3, "%c", &a3[i];
  fprintf(fpr2, "%c", a2[i];
  fprintf(fpr3, "%c", a3[i];
}

//then fclose all files

---

it crashes when i run it, but there are no compilation errors. the funny thing is, when i change 'size' to a smaller number, say 10,000, it runs perfectly. i was told this might be due to some differences in the 'stack' and 'heap' memory types in the computer, which is why im now using malloc instead.

What do you mean when you say "numbers"?

fscanf(%c) and fgetc() are going to be the same thing, except the latter is considerably less heavy on machine resources.

> i was told this might be due to some differences in the 'stack' and 'heap' memory types in the computer
Your default stack size is usually around the 1MB mark.

If you really want to store the whole (and large file), then you'll need
unsigned char *p = malloc( size * sizeof *p );
Your first post was good, except for the lack of a size (you only had 1 char).

Yea im using the word ‘numbers ‘ rather loosely here i think. Basically Im doing some image processing, whereby each of my image file contains 1,249,820 pixels and each pixel is represented by a number ranging from 0-255. So I presume that the data is stored as unsigned char since each number is 1 byte.

Actually, how many data types are there? Ascii, binary, and text? And how do we determine which format a file is in?

I would still read the file in large chunks rather than 1 byte at a time, since it's much faster. After you have the chunk in memory you can do whatever you want with those bytes before you write them to the new file.

Code:

---

unsigned char *b = malloc( size * sizeof(*b) );
if ( b != NULL && fread( b, 1, size, fp ) == size ) {
  // yes, got the whole file in one go
}

---

Whats the difference between fwrite and fprintf? I tried the following variations:

Code:

---


for(i=0; i<size; i++)
{
        fscanf(fp2, "&#37;c", &b2[i]);
        fscanf(fp3, "%c", &b3[i]);
        c[i]=((float)b3[i]-(float)b2[i])/((float)b2[i]+(float)b3[i]+0.000001);
        fprintf(fpndvi, "%f", c[i]);
}

---

and

Code:

---

for(i=0; i<size; i++)
{
        fscanf(fp2, "%c", &b2[i]);
        fscanf(fp3, "%c", &b3[i]);
        c[i]=((float)b3[i]-(float)b2[i])/((float)b2[i]+(float)b3[i]+0.000001);
}
    fwrite(c, sizeof(float), size, fpndvi);

---

Both can run fine, but give vastly different results (that is what my image processing program, ENVI, tells me anw). I am quite certain im writing floating-point data to disk in both cases..

fprintf() prints the floating point value as digits that we humans can read. fwrite stores the raw (aka "binary") floating point data as it is stored internally in the memory, so 1.0 would be represented by 3F 80 00 00 or 00 00 80 3f - which you get depends on the type of processor being used.

Which you should use depends on what the other side is using to read it back in. If the other side is using fscanf("%f" ... ) then you need to use fprintf. If the other side is using fread, then fwrite is the right way to go.

--
Mats