std::ranges::unique() algorithm
- od C++20
- Simplified
- Detailed
// (1)
constexpr ranges::subrange<I>
unique( I first, S last, C comp = {}, Proj proj = {} );
// (2)
constexpr ranges::borrowed_subrange_t<R>
unique( R&& r, C comp = {}, Proj proj = {} );
The type of arguments are generic and have the following constraints:
I
-std::permutable
S
-std::sentinel_for<I>
R
-std::ranges::forward_range
C
:- (1) -
std::indirect_equivalence_relation<std::projected<I, Proj>>
- (2) -
std::indirect_equivalence_relation<std::projected<ranges::iterator_t<R>, Proj>>
- (1) -
Proj
- (none)
The Proj
and C
template argument have the following default types std::identity
, ranges::equal_to
for all overloads.
Additionally, each overload has the following constraints:
- (2) -
std::permutable<ranges::iterator_t<R>>
// (1)
template<
std::permutable I,
std::sentinel_for<I> S,
class Proj = std::identity,
std::indirect_equivalence_relation<std::projected<I, Proj>> C = ranges::equal_to
>
constexpr ranges::subrange<I>
unique( I first, S last, C comp = {}, Proj proj = {} );
// (2)
template<
ranges::forward_range R,
class Proj = std::identity,
std::indirect_equivalence_relation<std::projected<ranges::iterator_t<R>, Proj>> C = ranges::equal_to >
requires std::permutable<ranges::iterator_t<R>>
constexpr ranges::borrowed_subrange_t<R>
unique( R&& r, C comp = {}, Proj proj = {} );
-
(1) Eliminates all except the first element from every consecutive group of equivalent elements from the range [
first
;last
) and returns a subrange [ret
;last
), whereret
is a past-the-end iterator for the new end of the range.Two consecutive elements
*(i - 1) and *i
are considered equivalent ifstd::invoke(comp, std::invoke(proj, *(i - 1)), std::invoke(proj, *i)) == true
, wherei
is an iterator in the range [first + 1
;last
). -
(2) Same as (1), but uses
r
as the range, as if usingranges::begin(r)
asfirst
andranges::end(r)
aslast
.
The function-like entities described on this page are niebloids.
Parameters
first last | The range of elements to process. |
r | The range of elements to process. |
comp | The binary predicate to compare the projected elements with. |
proj | The projection to apply to the elements. |
Return value
Returns
{
ret,
last
}
Where ret
is a past-the-end iterator for the new end of the range.
Complexity
For nonempty ranges, exactly ranges::distance(first, last) - 1
applications of the corresponding predicate comp
and no more that twice as many applications of any projection proj
.
Exceptions
(none)
Possible implementation
unique(1) and unique(2)
struct unique_fn
{
template<std::permutable I, std::sentinel_for<I> S, class Proj = std::identity,
std::indirect_equivalence_relation<std::projected<I, Proj>>
C = ranges::equal_to>
constexpr ranges::subrange<I>
operator()(I first, S last, C comp = {}, Proj proj = {}) const
{
first = ranges::adjacent_find(first, last, comp, proj);
if (first == last)
return {first, first};
auto i {first};
++first;
while (++first != last)
if (!std::invoke(comp, std::invoke(proj, *i), std::invoke(proj, *first)))
*++i = ranges::iter_move(first);
return {++i, first};
}
template<ranges::forward_range R, class Proj = std::identity,
std::indirect_equivalence_relation<std::projected<ranges::iterator_t<R>, Proj>>
C = ranges::equal_to>
requires std::permutable<ranges::iterator_t<R>>
constexpr ranges::borrowed_subrange_t<R>
operator()(R&& r, C comp = {}, Proj proj = {}) const
{
return (*this)(ranges::begin(r), ranges::end(r),
std::move(comp), std::move(proj));
}
};
inline constexpr unique_fn unique {};
Notes
Removing is done by shifting (by means of move assignment) the elements in the range in such a way that the elements that are not to be removed appear in the beginning of the range. Relative order of the elements that remain is preserved and the physical
size of the container is unchanged.
Iterators in [ret
; last
) (if any) are still dereferenceable, but the elements themselves have unspecified values (as per MoveAssignable post-condition).
A call to ranges::unique
is sometimes followed by a call to a container’s erase
member function,
which erases the unspecified values and reduces the physical size of the container to match its new logical size. These two invocations together model the Erase–remove idiom.
Examples
#include <algorithm>
#include <cmath>
#include <complex>
#include <iostream>
#include <vector>
struct id {
int i;
explicit id(int i) : i {i} {}
};
void print(id i, const auto& v)
{
std::cout << i.i << ") ";
std::ranges::for_each(v, [](auto const& e) { std::cout << e << ' '; });
std::cout << '\n';
}
int main()
{
// a vector containing several duplicated elements
std::vector<int> v {1, 2, 1, 1, 3, 3, 3, 4, 5, 4};
print(id {1}, v);
// remove consecutive (adjacent) duplicates
const auto ret = std::ranges::unique(v);
// v now holds {1 2 1 3 4 5 4 x x x}, where 'x' is indeterminate
v.erase(ret.begin(), ret.end());
print(id {2}, v);
// sort followed by unique, to remove all duplicates
std::ranges::sort(v); // {1 1 2 3 4 4 5}
print(id {3}, v);
const auto [first, last] = std::ranges::unique(v.begin(), v.end());
// v now holds {1 2 3 4 5 x x}, where 'x' is indeterminate
v.erase(first, last);
print(id {4}, v);
// unique with custom comparison and projection
std::vector<std::complex<int>> vc { {1, 1}, {-1, 2}, {-2, 3}, {2, 4}, {-3, 5} };
print(id {5}, vc);
const auto ret2 = std::ranges::unique(vc,
// consider two complex nums equal if their real parts are equal by module:
[](int x, int y) { return std::abs(x) == std::abs(y); }, // comp
[](std::complex<int> z) { return z.real(); } // proj
);
vc.erase(ret2.begin(), ret2.end());
print(id {6}, vc);
}
1) 1 2 1 1 3 3 3 4 5 4
2) 1 2 1 3 4 5 4
3) 1 1 2 3 4 4 5
4) 1 2 3 4 5
5) (1,1) (-1,2) (-2,3) (2,4) (-3,5)
6) (1,1) (-2,3) (-3,5)
Hover to see the original license.