ifstream binary read into unsigned char b[LENGTH]

[Steven T. Hatton]

What is the best way to read data from a file into a fixed size array of
unsigned char? This is a file holding only a SHA1 digest. What I would
really like to do is initialize the ifstream with the buffer allocated to
hold the data, seekg(ios_base::end) and magically have the data appear in
the buffer. I believe something close to this can be done.

I might allocate the array:

const unsigned LENGTH = 256;
unsigned char b[LENGTH];

Then cast it to a std::streambuf

std::streambuf sb(b[LENGTH]);

But can I construct std::ifstream that uses that same memory to hold the
data from its file? How to I prevent reading past the end of my buffer, or
the end of the file data?

 

[Rolf Magnus]

What is the best way to read data from a file into a fixed size array of
unsigned char? This is a file holding only a SHA1 digest. What I would
really like to do is initialize the ifstream with the buffer allocated to
hold the data, seekg(ios_base::end) and magically have the data appear in
the buffer.

Why should a seek read any data?

I believe something close to this can be done.

I might allocate the array:

const unsigned LENGTH = 256;
unsigned char b[LENGTH];

Then cast it to a std::streambuf

std::streambuf sb(b[LENGTH]);

I don't think that you can be sure that the buffer will contain exactly the
data you want. After all, it's a buffer that's meant for internal use.

But can I construct std::ifstream that uses that same memory to hold the
data from its file?

The purpose of streams is text formatting. If you want to read binary data,
just use a filebuf directly.

How to I prevent reading past the end of my buffer, or the end of the file
data?

Your idea seems overly complicated to me. Just use the filebuf::sgetn()
function to read as many bytes as you want and you're done.

 

[Steven T. Hatton]

Why should a seek read any data?

As I understand things, that will (usually?) force the file to be completely
read into memory.

Your idea seems overly complicated to me. Just use the filebuf::sgetn()
function to read as many bytes as you want and you're done.

Will that copy the bytes from one location in memory to another? That's
what I want to avoid. This is basically the same thing as move semantics.
Rather than copying the contents of the buffer, I simply want to take
ownership of it.