std::set_difference(3)
π std::set_difference(3) π
okay so hi bestie!! welcome to std::set_difference from the C++ Standard Library β the algorithm that answers "what's in list A that is not in list B" when both lists are already sorted. it is literally set math but make it fashion, and honestly? she ate π
β¨
πΈ NAME πΈ
std::set_difference β yep, that's the whole name. no plot twist, no secret subtitle, literally just std::set_difference π
β¨ Synopsis β¨
Defined in header <algorithm> β go add that #include, bestie, do not be shy π
π¦ Overload (1) π¦
template< class InputIt1, class InputIt2, class OutputIt >
OutputIt set_difference( InputIt1 first1, InputIt1 last1,
InputIt2 first2, InputIt2 last2,
OutputIt d_first );
(1) β and it's constexpr since C++20, we love a glow-up β
π¦ Overload (2) π¦
template< class ExecutionPolicy, class ForwardIt1, class ForwardIt2, class ForwardIt3 >
ForwardIt3 set_difference( ExecutionPolicy&& policy,
ForwardIt1 first1, ForwardIt1 last1,
ForwardIt2 first2, ForwardIt2 last2,
ForwardIt3 d_first );
(2) β this one's been around since C++17, and she's the parallel one π
π¦ Overload (3) π¦
template< class InputIt1, class InputIt2, class OutputIt, class Compare >
OutputIt set_difference( InputIt1 first1, InputIt1 last1,
InputIt2 first2, InputIt2 last2,
OutputIt d_first, Compare comp );
(3) β constexpr since C++20 too, custom comparator edition π
π¦ Overload (4) π¦
template< class ExecutionPolicy, class ForwardIt1, class ForwardIt2, class ForwardIt3, class Compare >
ForwardIt3 set_difference( ExecutionPolicy&& policy,
ForwardIt1 first1, ForwardIt1 last1,
ForwardIt2 first2, ForwardIt2 last2,
ForwardIt3 d_first, Compare comp );
(4) β since C++17, parallel and with a comparator. the full combo meal π
π« What it actually does π«
so here's the vibe: it copies the elements from the sorted range [first1, last1) which are not found in the sorted range [first2, last2) to the range beginning at d_first. and yes β the output range is also sorted. no cap, order is preserved π«Ά
now the multiplicity math, because this is the part everyone forgets and then gets confused about π
if [first1, last1) contains m elements that are equivalent to each other, and [first2, last2) contains n elements that are equivalent to them, then the final std::max(m - n, 0) elements will be copied from [first1, last1) to the output range, preserving order.
so it's not "remove all duplicates", it's "remove as many as the second range can cancel out". iconic honestly β¨
π₯ The behaviour notes (read these, fr fr) π₯
- if
[first1, last1)or[first2, last2)is not sorted with respect tooperator<(until C++20) /std::less{}(since C++20), the behavior is undefined. π like, undefined undefined. do not test her. - if
[first1, last1)or[first2, last2)is not sorted with respect tocomp, the behavior is undefined. same energy, custom comparator flavour π 2, 4. same as (1, 3), but executed according topolicy. these overloads participate in overload resolution only if:
std::is_execution_policy_v<std::decay_t<ExecutionPolicy>>istrue(until C++20)std::is_execution_policy_v<std::remove_cvref_t<ExecutionPolicy>>istrue(since C++20)
and last but very much not least:
if the output range overlaps with[first1, last1)or[first2, last2), the behavior is undefined. π¨ overlapping? don't. seriously. that's the UB corner, we don't go there π
π Parameters π
first1,last1β the range of elements to examine πfirst2,last2β the range of elements to search for πd_firstβ the beginning of the output range β¨policyβ the execution policy to use. see execution policy for details. πcompβ comparison function object (i.e. an object that satisfies the requirements ofCompare) which returnstrueif the first argument is less than (i.e. is ordered before) the second.
the signature of the comparison function should be equivalent to the following:
bool cmp(const Type1& a, const Type2& b);
important details, bestie, don't skip π«Ά β while the signature does not need to have const&, the function must not modify the objects passed to it, and must be able to accept all values of type (possibly const) Type1 and Type2 regardless of value category (thus, Type1& is not allowed, nor is Type1 unless for Type1 a move is equivalent to a copy (since C++11)). the types Type1 and Type2 must be such that objects of types InputIt1 and InputIt2 can be dereferenced and then implicitly converted to both Type1 and Type2.
π§ Type requirements π§
InputIt1,InputIt2must meet the requirements of LegacyInputIterator.OutputItmust meet the requirements of LegacyOutputIterator.ForwardIt1,ForwardIt2,ForwardIt3must meet the requirements of LegacyForwardIterator.Comparemust meet the requirements of Compare.
π Return value π
iterator past the end of the constructed range. that's your "here's where I stopped writing" pointer, go use it for erase or whatever π
β Complexity β
given N1 as std::distance(first1, last1) and N2 as std::distance(first2, last2):
1, 2. at most 2Β·(N1+N2)β1 comparisons using operator< (until C++20) / std::less{} (since C++20). 3, 4. at most 2Β·(N1+N2)β1 applications of the comparison function comp.
linear-ish, we love a clean bound π
π Exceptions π
the overloads with a template parameter named ExecutionPolicy report errors as follows:
- if execution of a function invoked as part of the algorithm throws an exception and
ExecutionPolicyis one of the standard policies,std::terminateis called. for any otherExecutionPolicy, the behavior is implementation-defined. - if the algorithm fails to allocate memory,
std::bad_allocis thrown.
so like, she can absolutely take you out. respectfully π
π Possible implementation π
πΈ set_difference (1) πΈ
template<class InputIt1, class InputIt2, class OutputIt>
OutputIt set_difference(InputIt1 first1, InputIt1 last1,
InputIt2 first2, InputIt2 last2, OutputIt d_first)
{
while (first1 != last1)
{
if (first2 == last2)
return std::copy(first1, last1, d_first);
if (*first1 < *first2)
*d_first++ = *first1++;
else
{
if (! (*first2 < *first1))
++first1;
++first2;
}
}
return d_first;
}
πΈ set_difference (3) πΈ
template<class InputIt1, class InputIt2, class OutputIt, class Compare>
OutputIt set_difference(InputIt1 first1, InputIt1 last1,
InputIt2 first2, InputIt2 last2, OutputIt d_first, Compare comp)
{
while (first1 != last1)
{
if (first2 == last2)
return std::copy(first1, last1, d_first);
if (comp(*first1, *first2))
*d_first++ = *first1++;
else
{
if (!comp(*first2, *first1))
++first1;
++first2;
}
}
return d_first;
}
π Example π
// Run this code
#include <algorithm>
#include <iostream>
#include <iterator>
#include <vector>
template<typename T>
std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
{
os << '{';
for (auto n{v.size()}; const auto& e : v)
os << e << (--n ? ", " : "");
return os << '}';
}
struct Order // a struct with very interesting data
{
int order_id{};
friend std::ostream& operator<<(std::ostream& os, const Order& ord)
{
return os << ord.order_id;
}
};
int main()
{
const std::vector<int> v1{1, 2, 5, 5, 5, 9};
const std::vector<int> v2{2, 5, 7};
std::vector<int> diff;
std::set_difference(v1.begin(), v1.end(), v2.begin(), v2.end(),
std::inserter(diff, diff.begin()));
std::cout << v1 << " β " << v2 << " == " << diff << "\n\n";
// we want to know which orders "cut" between old and new states:
std::vector<Order> old_orders{{1}, {2}, {5}, {9}};
std::vector<Order> new_orders{{2}, {5}, {7}};
std::vector<Order> cut_orders;
std::set_difference(old_orders.begin(), old_orders.end(),
new_orders.begin(), new_orders.end(),
std::back_inserter(cut_orders),
[](auto& a, auto& b) { return a.order_id < b.order_id; });
std::cout << "old orders: " << old_orders << '\n'
<< "new orders: " << new_orders << '\n'
<< "cut orders: " << cut_orders << '\n';
}
π Output π
{1, 2, 5, 5, 5, 9} β {2, 5, 7} == {1, 5, 5, 9}
old orders: {1, 2, 5, 9}
new orders: {2, 5, 7}
cut orders: {1, 9}
see how the 5s survive? that's the std::max(m - n, 0) tea right there. v1 had three 5s, v2 had one, so two 5s made it through π«Άβ¨
π¦ Defect reports π¦
the following behavior-changing defect reports were applied retroactively to previously published C++ standards.
LWG 291β applied to C++98 β behavior as published: it was unspecified how to handle equivalent elements in the input ranges β correct behavior: specified π
goodbye ambiguity, hello clarity π
π See also π
includesβ returnstrueif one sequence is a subsequence of another (function template)set_symmetric_differenceβ computes the symmetric difference between two sets (function template)ranges::set_difference(C++20) β computes the difference between two sets (niebloid)
and that's the whole page, bestie!! π source: http://cppreference.com, version 2024.06.10 β date on the tin and everything, mwah πππ¦πΈβ¨
π¦ where this page comes from
- nixpkgs unstable: stdman 2024.07.05
- opensuse-leap 16.0: stdman 2024.07.05-bp160.1.9 Β· MIT
- π girlified by deepseek-ai/DeepSeek-V4.1-Flash-1017 on : a language model's rewrite of the original, under its terms