fread() in fstream?

Hi,

Is there any function in <fstream> that reads n bytes from a file into a variable (of any type like int, char, etc.). I mean something like fread() in C. I found several functions in my book (like read() or readsome()), but they all read in a buffer of char*. What do I do if I want to read in an int?

Thanx...

Originally Posted by MathFan

Hi,

Is there any function in <fstream> that reads n bytes from a file into a variable (of any type like int, char, etc.). I mean something like fread() in C. I found several functions in my book (like read() or readsome()), but they all read in a buffer of char*. What do I do if I want to read in an int?

Thanx...

Well, like using fread you can always read data into a memory area occupied by an int through the use of a cast.

my bad.. you already know about those..

They will read in anything (char will accept any character) then in your program if you need the data to be of type int.. just peform a typecast

But if a char is 8 bits and int is 16 bits, how can I convert one into another?

one of the joys of c++ is that the compiler will take care of the memory managment for ye'. feel free to typecast to anything you want basically. it's all good. (as long as you are typecasting to a larger container.. going in reverse may lead to truncation)

OK, thanx... heh, I think I must brush up some C++ basics But, isn't it a rather strange way of doing it? First reading something into a character, then casting it? Wouldn't it be easier to implement read(void*, int) in iostream?

I think using char (or string) is probably the best way to handle program input. char or string will accept anything.. unlike int, float, double, short, long which will literally go nuts if you try to input something that is not of that type.. example: if you tried to input "akj:LKJ:LKAH" into a type float.. your program would go bezerk. So, accept input using char or string.. typecast as necessary.. and successfully handle any invalid input.

Here also is a recent post on error checking

Whooooooah guys, I think we're heading in TOTALLY the wrong direction. Do NOT read into a character and then typecast to an int.

What hk_mp5kdw meant is:

Code:

int myInt;
file.read((char*)&myInt, sizeof(int));

You take the address of the int and interpret it as char*, and read sizeof(int) bytes into the block of memory beginning at that address.

I may be wrong about this, but I have the impression that you can't perform pointer arithmetic on void*, i.e. thePointer++. As such, I suspect the void* may be internally typecasted to char* or unsigned char* anyway in fread(). Anyway, since char and unsigned char are supposed to be 1 byte, they're convenient types for bytewise operations such as binary-mode file i/o with arbitrarily sized blocks of data - if you want to write a block of data to a file, the function doesn't care whether it's a char or int or MyBestFriend object; all it cares is that you pass it the location of the object (i.e. its address), and its size, and it will then write the information byte by byte into a file.