Thread: Advanced File IO

Are you opening the file in binary mode? Should be something like:

Code:

int size;
ifstream in( "data.enc", ios::binary );
in.read( ( char* )&size, sizeof( int ) );

lol

Anyway, ok, here is some of my code.

My encoder outputting the file size as 4 bytes:

Code:

   /////////////get file size and send to decoder///////////////////
    unsigned long count = 0;
    INPUT_FILE.get();
    while(!INPUT_FILE.eof())
    {
        count++;       //use tellg instead?
        INPUT_FILE.get();
    }
    INPUT_FILE.clear();
    INPUT_FILE.seekg(0);
    
    unsigned long temp = 255;
    for(int i = 0; i < 4; i++)
    {
        unsigned long x = (count & temp);
        x = x/int(pow(float(256), float(i)));
        OUTPUT_FILE << char(x);
        temp = temp * 256;
    }

//count (32 bits long) is divided up into four bytes and sent out

yeah its kind of ugly but it works. My problem supposedly originates when my decoder tries to read these four bytes:

Code:

    ifstream INPUT_FILE;
    ofstream OUTPUT_FILE;
    
    openFiles(INPUT_FILE, OUTPUT_FILE);
    
    /////////////////////get file size/////////////////////
    unsigned long count = 0;
    unsigned long temp = 256;
          
    for(int i = 0; i < 4; i++)
    {
        int t;    
        t = INPUT_FILE.get(); //doesn't work either if i use a char and get(char)
        if(t < 0)
            t+=256;
        unsigned long x = t;
        x = x*int(pow(float(temp), float(i)));   
        count += x;
    }

In some cases this won't work. The encoder put my file size (26) into byte form, which is the substitution ascii character. ifstream refuses to read this character, but I've got to have a way to read all 256 ascii characters.


Thanks Sebastiani. I'll try it.

ifstream is really not meant for binary stuff, in my opinion. I feel much more at home with FILE * usually, or streambufs. That said:

1) Your measurement of the file size. seekg() to the end, and tellg(), instead of reading the whole file off of disk. That'll be a whole heck of a lot cheaper than reading the entire file and discarding it. Better, if you can (and it's not always possible), don't care about the file size. Read the file straight through, beginning to end. The benefit here is you can then read non-seekable files, such as pipes.

2) Try read(), although I'm not sure why get() doesn't work (but my gut says I ran into this years ago, I just forget what the issue was!) Since you know you're going to read 4 bytes, request 4 bytes, not 1,1,1,1.