64

Is there any specific reason for not having std::copy_if algorithm in C++ ? I know I can use std::remove_copy_if to achieve the required behavior. I think it is coming in C++0x, but a simple copy_if which takes a range, a output iterator and a functor would have been nice. Was it just simply missed out or is there some other reason behind it?

Naveen
  • 74,600
  • 47
  • 176
  • 233

6 Answers6

42

According to Stroustrup's "The C++ Programming Language" it was just an over-sight.

(as a citation, the same question answered in boost mail-lists: copy_if)

sbk
  • 9,212
  • 4
  • 32
  • 40
  • 10
    As an update, the C++11 standard has corrected this oversight by adding a new `copy_if` algorithm: http://en.cppreference.com/w/cpp/algorithm/copy – Bret Kuhns Sep 23 '12 at 16:46
  • Is there a variant that passes the iterator to the the unary predicate instead of passing the result of dereferencing the iterator? I don't see one, but that seems like another oversight. – Levi Morrison Oct 08 '18 at 00:24
28

Stroustrup says they forgot it. It's in C++11.

However, you can use remove_copy_if (which really should be called copy_if_not) along with not1 instead.

rlbond
  • 65,341
  • 56
  • 178
  • 228
10

Just for completeness, in case someone googles his/her way to this question, it should be mentioned that now (post C++11) there is a copy if algorithm. It behaves as expected (copies the elements in a range, for which some predicate returns true, to another range).

A typical use case would be

std::vector<int> foo{ 25, 15, 5, -5, -15 };
std::vector<int> bar;

// copy only positive numbers:
auto it = std::copy_if (foo.begin(), foo.end(), std::back_inserter(bar), 
            [](int i){return !(i<0);
          });
Nikos Athanasiou
  • 29,616
  • 15
  • 87
  • 153
8

Multiple sources indicate that it was left out of STL by accident.

However, I am not sure if that's a fact or a self-perpetuating myth. I'd appreciate if anyone would point out a source more credible than a link to a random post on the Internet.

Community
  • 1
  • 1
Alex B
  • 82,554
  • 44
  • 203
  • 280
  • 4
    Here's a non-random post on the internet, selected on the basis that it claims to be an email from Stroustrup to the Boost mailing list: http://lists.boost.org/Archives/boost/2001/01/8030.php. Of course it could be a fraud, or it could be that Stroustrup himself has bought the myth. I guess it's possible that Stepanov generally thought it best to have `remove_copy_if`, and deliberately excluded `copy_if` as redundant. But seriously, it's clearly some kind of mistake to have `remove_copy_if` but not `copy_if`, if only one of taste :-) – Steve Jessop Apr 17 '10 at 02:45
7

It's dead easy to write your own:

template <class InputIterator, class OutputIterator, class Predicate>
OutputIterator copy_if(InputIterator first, InputIterator last,
                       OutputIterator result, Predicate pred)
{
  return std::remove_copy_if(first,last,result,std::not1(pred));
}

Edit: This version works with all predicates:

template <class InputIterator, class OutputIterator, class Predicate>
OutputIterator copy_if(InputIterator first, InputIterator last,
                       OutputIterator result, Predicate pred)
{
  while(first!=last)
  {
    if(pred(*first))
        *result++ = *first;
    ++first;
  }
  return result;
}
alex tingle
  • 6,920
  • 3
  • 25
  • 29
  • 2
    This isn't actually correct, as noted in *Effective STL* item 36, because it only works on adaptable functors. – rlbond Sep 19 '09 at 18:00
  • looks like it's not that dead easy, it seems that your second attempt returns the first value matching the predicate - besides, IMHO, you're a little off-topic here – rotoglup Sep 26 '09 at 10:08
  • rotoglup - I'm afraid that you don't know what you're talking about. 'result' is an OutputIterator. I did test this code you know. – alex tingle Sep 26 '09 at 10:14
  • 4
    Don't you have to iterate the OutputIterator? `*(result++) = *first;` – Peter Kovacs Jan 26 '10 at 02:03
  • @Peter Kovacs - nope. I thought that too at first, but if you check you'll find that you don't need to explicitly increment it. – alex tingle Jan 27 '10 at 15:43
  • 1
    -1: If `result` is an `OutputIterator` then why do you assign `*first` to it? Your code is wrong and @Peter Kovacs is correct. – Troubadour Apr 28 '11 at 22:28
  • 3
    @Peter & Troubadour - of course you are both correct. Mea culpa. The code I wrote did actually work, but presumably that was only because of some crazy intersection between the implementation details and my test code. I have corrected it and +1'd you. – alex tingle May 03 '11 at 19:53
2

Just for completeness I will add that boost has boost::algorithm::copy_if for those of you who cannot use c++11's version (like me) in boost/algorithm/cxx11/copy_if.hpp which will use std::copy_if when:

#if __cplusplus >= 201103L
//  Use the C++11 versions of copy_if if it is available
using std::copy_if;         // Section 25.3.1
#else

Example:

#include <boost/algorithm/cxx11/copy_if.hpp>
#include <boost/assign/list_of.hpp> // for 'list_of()'
#include <boost/foreach.hpp>

#include <iostream>
#include <vector>
#include <iterator>

struct Odd
{
  bool operator()(int n)
  {
    return n & 1;
  }
};

int main()
{
  std::vector<int> v = boost::assign::list_of(0)(1)(2)(3)(4);
  BOOST_FOREACH(int i, v)
    std::cout << i << ' ' ;

  std::vector<int> out;
  boost::algorithm::copy_if(v.begin(), v.end(), std::back_inserter(out), Odd());

  std::cout << std::endl;

  BOOST_FOREACH(int i, out)
    std::cout << i << ' ' ;

}

Output:

0 1 2 3 4 
1 3 
Patryk
  • 22,602
  • 44
  • 128
  • 244