Extracting data from streams

I am currently aghast at the lack of support of reading information from files in C++.
I can't find a good way to read a chunk of data from them.

My current goals are to be able to iterate over the contents of a file using random access/forward iterators (because I need ++, += and []), which, of course, stream iterators doesn't support (pathetic things).
So I thought I'd implement a buffered stream iterator that reads chunks of data from the file and allows for random access to it.
What I'm stuck on, however, is reading that chunks of data from the file.
I have two approaches:

(IndiceType is a typedef for std::size_t.)

Code:

---

IndiceType BufferSize = m_Buffer.size();
m_Buffer.resize(BufferSize == 0 ? 1 * MB : BufferSize * 2);
m_ItEnd = m_Buffer.size();
m_pFile->read((&*m_Buffer.begin() + m_ItCurrent),
        std::streamsize(m_ItEnd - m_ItCurrent));

---

This approach uses a vector and indices to manually resize the vector and read a chunk of data from the file. Unfortunately, I don't know how much data it reads.

(T is a template type for char or wchar_t.)
(BufType is a typedef for std::vector<T>.)

Code:

---

std::istream_iterator<T, T> SrcIt(*m_pFile);
std::istream_iterator<T, T> EndIt;
std::back_insert_iterator<BufType> DstIt(m_Buffer);
int i = 0;
for (;; i++)
{
        *DstIt++ = *SrcIt++;
        if (SrcIt == EndIt || i == 1 * MB)
                break;
}
m_ItEnd += i;

---

Another approach is this. Basically std::copy with a istream iterator and a back_insert iterator. But it's painfully slow, so it's out of the question, even though it works.
Another thing that bugs me is that I can't use std::copy, because I can't get an iterator that is 1 MB bytes away from the begin iterator, since istream_iterator doesn't support operator +.

So any good ideas on how to do this? C++ is worse in this area than C is, really.
It's truly a shame.

Member variables in the class that code uses:

Code:

---

BufType m_Buffer;
IndiceType m_ItBegin;
IndiceType m_ItCurrent;
IndiceType m_ItEnd;
std::basic_istream<T>* m_pFile;
bool m_End; // Is *this == end()?

---

Quote:

---

Originally Posted by Elysia

This approach uses a vector and indices to manually resize the vector and read a chunk of data from the file. Unfortunately, I don't know how much data it reads.

---

Check istream:gcount() after each read operation. That should put you on track.

Gotcha. Thanks.

Code:

---

1 * MB

---

Probably gratuitous syntactical sugar, but it made me lol.

Yes, it's syntactic sugar, because it's easier to read than numbers such as 1024 * 1024:

Code:

---

typedef unsigned int ShortUnitSize_t;
typedef unsigned long long LongUnitSize_t;
const ShortUnitSize_t KB = 1024;
const ShortUnitSize_t MB = KB * 1024;
const ShortUnitSize_t GB = MB * 1024;
const LongUnitSize_t TB = GB * 1024LL;
const LongUnitSize_t EB = TB * 1024LL;

---

I think he means gratuitous in the same way

p = malloc( BUFSIZ * sizeof(char) );

would be gratuitous.

By the way I just thought this is a neat idea.

And on my side the optimizer did it's magic on it.
Gonna steal it, Elysia :D

Heh. Go ahead.
It's a public forum, after all.