z0mi3ie
z0mi3ie

Reputation: 3

Writing to a binary file from a struct part of a linked list

I'm trying to understand how to properly right data to a binary file. The data is being written but when read back in for testing it is obvious I am not writing it properly. The data is read/written as follows.

Unsigned Char tells how long next string will be, then a string is written, and then an integer ID is written. ( or length studentname student_id ). The following is the printList function which saves the data to a file.

void printList(struct node * nptr, string filename){
  ofstream myfile;
  myfile.open( filename.c_str(), ios::binary );

  while(nptr){

  myfile.write( (char*) &nptr->data.len, sizeof( unsigned char ) );
  myfile.write( (char*) nptr->data.name, len * sizeof(char) );
  myfile.write( (char*) &nptr->data.id, sizeof(int) );

 //myfile.write( (const char*) &nptr->data, sizeof( Student ) );

 nptr = nptr->next;
}

myfile.close();

}

Here is the student struct from which the data is stored in a link list:

struct node{
Student data;
struct node* next;
};

//=== Student struct===//
 struct Student{
    unsigned char len;
char* name;
int id;
};
//====================//

Here is the loadbin file I wrote to read the binary back in, again I am messing something up and I have read and watched a ton of tutorials on this. I really am stumped.

unsigned char len;
char* name;
int id;

int main( int argc, char* argv[] )
{
    if( argc > 1 )
    {
        ifstream f( argv[1] , ios::binary);

        while( !f.eof() )
        {
            f.read( (char*) &len , sizeof( unsigned char ) );

            if( f.eof() )
                break;

        name = new char[len+1];
        name[len+1] = '\0';

        f.read( (char*) name , len * sizeof( char ) );
        f.read( (char*) &id , sizeof( int ) );

        cout << (int)len << " " << name << " " << id 
                                            << "\n";
        }
    }

   else
        cout << "\nToo few arguments.\n";

}

Some more info, I can't change how the file is being read back in. The fix has to come from when I'm writing. I just included that in case it would help.

Upvotes: 0

Views: 1845

Answers (2)

paxdiablo
paxdiablo

Reputation: 882078

For a start, I'm not even sure this should compile, unless you have a len variable floating around somewhere, such as in your last code segment:

myfile.write( (char*) nptr->data.name, len * sizeof(char) ); // len, what len?

Since sizeof(char) is always one and I prefer explicit expressions as much as possible, I'd rewrite the three output lines as:

myfile.write ((const char *)(&(nptr->data.len)), 1);
myfile.write ((const char *)(nptr->data.name), nptr->data.len);
myfile.write ((const char *)(&(nptr->data.id)), sizeof(int));

A full program showing how to use that follows, similar enough that it will help, but different enough so that you can't just hand it in as your own work :-)

#include <iostream>
#include <fstream>

struct Student {
    unsigned char len;
    char *name;
    int id;
};

struct node {
    Student data;
    struct node *next;
};

int main (void) {
    // Create dummy list, two nodes.

    node *temp = new node();
    temp->data.len = 6;
    temp->data.name = (char *)"diablo";
    temp->data.id = 2;
    temp->next = NULL;

    node *shortList = new node();
    shortList->data.len = 3;
    shortList->data.name = (char *)"pax";
    shortList->data.id = 1;
    shortList->next = temp;

    // Write it out.

    std::ofstream myfile;
    myfile.open ("xyzzy.bin", std::ios::binary);
    node *curr = shortList;
    while (curr != NULL) {
        myfile.write ((const char *)(&(curr->data.len)), 1);
        myfile.write ((const char *)(curr->data.name), curr->data.len);
        myfile.write ((const char *)(&(curr->data.id)), sizeof(int));
        curr = curr->next;
    }
    myfile.close ();

    return 0;
}

If you run that program then do a hex dump, you get:

addr  +0 +1 +2 +3 +4 +5 +6 +7 +8 +9 +a +b +c +d +e +f  0123456789abcdef
----  -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --  ----------------
0000  03 70 61 78 01 00 00 00 06 64 69 61 62 6c 6f 02  .pax.....diablo.
0010  00 00 00                                         ...
0013

which I think matches the format you wanted.

Upvotes: 1

gbulmer
gbulmer

Reputation: 4290

Without some clue to the contents of the file, it is hard to debug, but try to ensure things are absolutely clear.

Rather than

  myfile.write( (char*) &nptr->data.len, sizeof( unsigned char ) );
  myfile.write( (char*) nptr->data.name, len * sizeof(char) );
  myfile.write( (char*) &nptr->data.id, sizeof(int) );

it is less error-prone to write (I squeezed out spaces to keep within line length):

  myfile.write((char*)&nptr->data.len, sizeof(nptr->data.len) );
  myfile.write((char*)nptr->data.name, nptr->data.len*sizeof(nptr->data.name[0]));
  myfile.write((char*)&nptr->data.id, sizeof(nptr->data.id) );

and use similar expressions to unpack the data when it is read back. Using sizeof on the variables is more robust than using type names.

Upvotes: 0

Related Questions