Intersection.hpp

Go to the documentation of this file.

#ifndef SINGLEPP_INTERSECTION_HPP
#define SINGLEPP_INTERSECTION_HPP
 
#include "defs.hpp"
 
#include <vector>
#include <algorithm>
#include <cstdint>
#include <numeric>
#include <unordered_map>
 
namespace singlepp {
 
template<typename Index_ = DefaultIndex>
using Intersection = std::vector<std::pair<Index_, Index_> >;
 
template<typename Index_, typename Id_>
Intersection<Index_> intersect_genes(Index_ test_nrow, const Id_* test_id, Index_ ref_nrow, const Id_* ref_id) {
    std::unordered_map<Id_, Index_> ref_found;
    for (Index_ i = 0; i < ref_nrow; ++i) {
        auto current = ref_id[i];
        auto tfIt = ref_found.find(current);
        if (tfIt == ref_found.end()) { // only using the first occurrence of each ID in ref_id.
            ref_found[current] = i;
        }
    }
 
    Intersection<Index_> output;
    for (Index_ i = 0; i < test_nrow; ++i) {
        auto current = test_id[i];
        auto tfIt = ref_found.find(current);
        if (tfIt != ref_found.end()) {
            output.emplace_back(i, tfIt->second);
            ref_found.erase(tfIt); // only using the first occurrence of each ID in test_id; the next will not enter this clause.
        }
    }
 
    // This is implicitly sorted by the test indices... not that it really
    // matters, as subset_to_markers() doesn't care that it's unsorted.
    return output;
}
 
}
 
#endif