how to use std::copy for int** - c++

Question

My2DClass::My2DClass(const int r, const int c, int mat[3][3]):m_r(r),m_c(c)
{
matrix = new int*[r];
for (int i = 0; i < r; i++)
matrix[i] = new int[c];
for (int i = 0; i < m_r; i++) {
for (int j = 0; j < m_c; j++) {
matrix[i][j] = mat[i][j];
}
}
//std::copy(&mat[0][0], &mat[0][0] + m_r * m_c, &matrix[0][0]);
}
How to use std::copy() for int**? The commented line throws an exception at runtime.

Answer 1

You are allocating an array of pointers to arrays, so a single call to std::copy() will not work. You would have to call std::copy() on every array individually. So really, only your inner-most loop can be replaced with std::copy(), eg:
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c)
{
matrix = new int*[r];
for (int i = 0; i < r; i++)
matrix[i] = new int[c];
for (int i = 0; i < m_r; i++) {
std::copy(&mat[i][0], &mat[i][m_c], matrix[i]);
// or: std::copy_n(&mat[i][0], m_c, matrix[i]);
}
}
If you consolidate the 2 remaining loops into a single loop, eg:
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c)
{
matrix = new int*[r];
for (int i = 0; i < r; ++i) {
matrix[i] = new int[c];
std::copy(&mat[i][0], &mat[i][c], matrix[i]);
// or: std::copy_n(&mat[i][0], c, matrix[i]);
}
}
Then you could replace that loop with std::for_each(), eg:
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c)
{
matrix = new int*[r];
std::for_each(matrix, matrix + r,
[=](int* &arr){
arr = new int[c];
std::copy(&mat[i][0], &mat[i][c], arr);
// or: std::copy_n(&mat[i][0], c, arr);
}
);
}
Though, you really should avoid using new[] manually at all, consider using std::vector instead, which would greatly simplify management of your arrays, eg:
// std::vector<std::vector<int>> matrix;
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c), matrix(r)
{
for(int i = 0; i < r; ++i) {
matrix[i].assign(&mat[i][0], &mat[i][c]);
}
}
Otherwise, consider flattening your matrix into a single 1-dimensional array instead. Especially since the input mat is a flat array in memory anyway. Then you can do a single std::copy() from one to the other, eg:
// int *matrix;
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c), matrix(new int[r*c])
{
std::copy(&mat[0][0], &mat[r][c], matrix);
}
Or, using std::vector instead:
// std::vector<int> matrix;
My2DClass::My2DClass(const int r, const int c, int mat[3][3])
: m_r(r), m_c(c), matrix(&mat[0][0], &mat[r][c])
{
}
Either way, you can convert 2-dimensional indexes into 1-dimensional indexes using this formula:
(r * m_c) + c
For example:
int& My2DClass::operator()(int r, int c)
{
return matrix[(r * m_c) + c];
}