Parsing header of file - c++

Question

I am working with images. I would like to extract width and height of it from a header.
The width is represented at location
size
4B width
4B height
they are represented at specific index.
I tried parse it and extract it with code
ifstream f(name, ios::binary | ios:: in ); // reading a file in binary
ostringstream ob;
if ( f.fail()){ // fail test
return false;
}
f.seekg (0, f.end);
int length = f.tellg(); // length of the file
memory = new char[length]; // allocate array of chars
f.read (memory, length); // read the content of the file into an array
f.seekg (0, f.beg); // point back at the beginning of the file.
Each of them has 4B , so using for loop
for ( int i = index ; i <4 ;i++){
cout << hex<< memory[i];
}
or i even tried it converting it into number using
string a;
for ( int i = index; i < 4 ;i++{
a+=memory[i];
}
cout << atoi( a.c_str() ) << endl;
Should output a number , but it output some unreadable format.

Answer 1

You're seeking to the beginning of the file after reading it, rather than before. You need to switch round the read() and the seekg():
f.seekg (0, f.beg); // point back at the beginning of the file.
f.read (memory, length); // read the content of the file into an array

Answer 2

You don't give very much information to go on so a little guesswork here but this is how I might approach reading the file:
The main features being, don't use manual memory allocation, use a std::vector (its what its for). Also copy the data from the char array into variables cast from the correct type. This ensures alignment is correct (never cast into the char array). Another approach might be to read directly from the file into the correctly typed variables cast into char*.
int main(int, char* argv[])
{
// first parameter needs to be file name
std::string name = argv[1] ? argv[1]:"";
std::ifstream ifs(name, std::ios::binary|std::ios::ate); // open at end
if(!ifs)
{
std::cerr << std::strerror(errno) << '\n';
return EXIT_FAILURE;
}
if(ifs.tellg() < 8) // too small
{
std::cerr << "Bad image file, too short" << '\n';
return EXIT_FAILURE;
}
// Don't allocate memory manually, use a container
std::vector<char> image(ifs.tellg()); // big enough for whole file
ifs.seekg(0); // back to beginning
if(!ifs.read(image.data(), image.size()))
{
std::cerr << std::strerror(errno) << '\n';
return EXIT_FAILURE;
}
// copy raw data into variables
std::uint32_t width; // 4 bytes wide integer
std::uint32_t height; // 4 bytes wide integer
std::copy(&image[4], &image[ 8], (char*)&width); // offset 4 bytes
std::copy(&image[8], &image[12], (char*)&height); // offset 8 bytes
// at this point a lot depends on the system architecture and
// how the number is stored in the file. The documentation
// should tell you if it is little-endian or big-endian
// you may have to do manual jiggery-pokery
// to change endienness
std::cout << "width : " << width << '\n';
std::cout << "height: " << height << '\n';
}

Answer 3

For a quick answer of the problem stated in the title, and by looking at the input file you specify, you could simply read three lines, skip the first and then extract one integer each from the following two lines.
So something like
int width, height;
std::string input;
std::istringstream is;
std::getline(stream, input); // One line for the "size" string
std::getline(stream, input); // One line for the width
is.str(input)
is >> width;
std::getline(stream, input); // One line for the height
is.str(input)
is >> height;
If you have multiple entries like this in the file, then do the above in a loop.
If the file doesn't actually contain the texts you show, just the numbers, then it's even simpler:
int width, height;
stream >> width >> height;
Or if you have multiple entries
while (stream >> width >> height) { ... }

Answer 4

std::cout will print data passed as characters if the data is char, so try casting before printing.
for ( int i = 0 ; i < 4 ; i++) {
cout << hex<< (int)(unsigned char)memory[index + i];
}
If the number in the file is little endian, it can be converted to integer like this:
// Use #include <cstdint> for using type uint32_t and uint8_t
uint32_t num = 0; // the converted number will be here
for ( int i = 0 ; i < 4 ; i++) {
num |= (uint8_t)memory[index + i] << (uint32_t)(8 * i);
}