I have two integer vectors of equal length. Let's say I want to remove all items in the first vector which are NAN. Obviously, I use the remove_if algorithm. Let's say this removes elements that were at indexes 1,2,5. I then want to remove items from the second vector at these indexes.
What's the most canonical C++ way of doing this?
This can be done using Boost by creating a zip_iterator
and then iterating over the tuple
of iterators from both containers in parallel.
First pass a pair of zip_iterators
to std::remove_if
, and have the predicate inspect the elements of the first vector
for NaN
auto result = std::remove_if(boost::make_zip_iterator(boost::make_tuple(v1.begin(), v2.begin())),
boost::make_zip_iterator(boost::make_tuple(v1.end(), v2.end())),
[](boost::tuple<double, int> const& elem) {
return std::isnan(boost::get<0>(elem));
});
Then use vector::erase
to remove the unneeded elements.
v1.erase(boost::get<0>(result.get_iterator_tuple()), v1.end());
v2.erase(boost::get<1>(result.get_iterator_tuple()), v2.end());
Live demo
The boilerplate required to create the zipped iterator ranges can be further reduced by using boost::combine
and Boost.Range's version of remove_if
.
auto result = boost::remove_if(boost::combine(v1, v2),
[](boost::tuple<double, int> const& elem) {
return std::isnan(boost::get<0>(elem));
});
Live demo
Use a vector<pair<int, int>>
to tie the two vector together. Then, perform your remove based on the first element to and get rid of both at the same time.