Sorting a vector of custom objects

Question

How does one go about sorting a vector containing custom (i.e. user defined) objects.
Probably, standard STL algorithm sort along with a predicate (a fu

Answer 1

In your class, you may overload the "<" operator.

class MyClass
{
  bool operator <(const MyClass& rhs)
  {
    return this->key < rhs.key;
  }
}

Answer 2

To sort a vector you can use the sort() algorithm in .

sort(vec.begin(),vec.end(),less<int>());

The third parameter used can be greater or less or any function or object can also be used. However the default operator is < if you leave third parameter empty.

// using function as comp
std::sort (myvector.begin()+4, myvector.end(), myfunction);
bool myfunction (int i,int j) { return (i<j); }

// using object as comp
std::sort (myvector.begin(), myvector.end(), myobject);

Answer 3

typedef struct Freqamp{
    double freq;
    double amp;
}FREQAMP;

bool struct_cmp_by_freq(FREQAMP a, FREQAMP b)
{
    return a.freq < b.freq;
}

main()
{
    vector <FREQAMP> temp;
    FREQAMP freqAMP;

    freqAMP.freq = 330;
    freqAMP.amp = 117.56;
    temp.push_back(freqAMP);

    freqAMP.freq = 450;
    freqAMP.amp = 99.56;
    temp.push_back(freqAMP);

    freqAMP.freq = 110;
    freqAMP.amp = 106.56;
    temp.push_back(freqAMP);

    sort(temp.begin(),temp.end(), struct_cmp_by_freq);
}

if compare is false, it will do "swap".

Answer 4

A simple example using std::sort

struct MyStruct
{
    int key;
    std::string stringValue;

    MyStruct(int k, const std::string& s) : key(k), stringValue(s) {}
};

struct less_than_key
{
    inline bool operator() (const MyStruct& struct1, const MyStruct& struct2)
    {
        return (struct1.key < struct2.key);
    }
};

std::vector < MyStruct > vec;

vec.push_back(MyStruct(4, "test"));
vec.push_back(MyStruct(3, "a"));
vec.push_back(MyStruct(2, "is"));
vec.push_back(MyStruct(1, "this"));

std::sort(vec.begin(), vec.end(), less_than_key());

---

Edit: As Kirill V. Lyadvinsky pointed out, instead of supplying a sort predicate, you can implement the operator< for MyStruct:

struct MyStruct
{
    int key;
    std::string stringValue;

    MyStruct(int k, const std::string& s) : key(k), stringValue(s) {}

    bool operator < (const MyStruct& str) const
    {
        return (key < str.key);
    }
};

Using this method means you can simply sort the vector as follows:

std::sort(vec.begin(), vec.end());

Edit2: As Kappa suggests you can also sort the vector in the descending order by overloading a > operator and changing call of sort a bit:

struct MyStruct
{
    int key;
    std::string stringValue;

    MyStruct(int k, const std::string& s) : key(k), stringValue(s) {}

    bool operator > (const MyStruct& str) const
    {
        return (key > str.key);
    }
};

And you should call sort as:

std::sort(vec.begin(), vec.end(),greater<MyStruct>());

Answer 5

Sorting such a vector or any other applicable (mutable input iterator) range of custom objects of type X can be achieved using various methods, especially including the use of standard library algorithms like

• sort,
• stable_sort,
• partial_sort or
• partial_sort_copy.

Since most of the techniques, to obtain relative ordering of X elements, have already been posted, I'll start by some notes on "why" and "when" to use the various approaches.

The "best" approach will depend on different factors:

1. Is sorting ranges of X objects a common or a rare task (will such ranges be sorted a mutiple different places in the program or by library users)?
2. Is the required sorting "natural" (expected) or are there multiple ways the type could be compared to itself?
3. Is performance an issue or should sorting ranges of X objects be foolproof?

If sorting ranges of X is a common task and the achieved sorting is to be expected (i.e. X just wraps a single fundamental value) then on would probably go for overloading operator< since it enables sorting without any fuzz (like correctly passing proper comparators) and repeatedly yields expected results.

If sorting is a common task or likely to be required in different contexts, but there are multiple criteria which can be used to sort X objects, I'd go for Functors (overloaded operator() functions of custom classes) or function pointers (i.e. one functor/function for lexical ordering and another one for natural ordering).

If sorting ranges of type X is uncommon or unlikely in other contexts I tend to use lambdas instead of cluttering any namespace with more functions or types.

This is especially true if the sorting is not "clear" or "natural" in some way. You can easily get the logic behind the ordering when looking at a lambda that is applied in-place whereas operator< is opague at first sight and you'd have to look the definition up to know what ordering logic will be applied.

Note however, that a single operator< definition is a single point of failure whereas multiple lambas are multiple points of failure and require a more caution.

If the definition of operator< isn't available where the sorting is done / the sort template is compiled, the compiler might be forced to make a function call when comparing objects, instead of inlining the ordering logic which might be a severe drawback (at least when link time optimization/code generation is not applied).

Ways to achieve comparability of class X in order to use standard library sorting algorithms

Let std::vector<X> vec_X; and std::vector<Y> vec_Y;

1. Overload T::operator<(T) or operator<(T, T) and use standard library templates that do not expect a comparison function.

Either overload member operator<:

struct X {
  int i{}; 
  bool operator<(X const &r) const { return i < r.i; } 
};
// ...
std::sort(vec_X.begin(), vec_X.end());

or free operator<:

struct Y {
  int j{}; 
};
bool operator<(Y const &l, Y const &r) { return l.j < r.j; }
// ...
std::sort(vec_Y.begin(), vec_Y.end());

2. Use a function pointer with a custom comparison function as sorting function parameter.

struct X {
  int i{};  
};
bool X_less(X const &l, X const &r) { return l.i < r.i; }
// ...
std::sort(vec_X.begin(), vec_X.end(), &X_less);

3. Create a bool operator()(T, T) overload for a custom type which can be passed as comparison functor.

struct X {
  int i{};  
  int j{};
};
struct less_X_i
{
    bool operator()(X const &l, X const &r) const { return l.i < r.i; }
};
struct less_X_j
{
    bool operator()(X const &l, X const &r) const { return l.j < r.j; }
};
// sort by i
std::sort(vec_X.begin(), vec_X.end(), less_X_i{});
// or sort by j
std::sort(vec_X.begin(), vec_X.end(), less_X_j{});

Those function object definitions can be written a little more generic using C++11 and templates:

struct less_i
{ 
    template<class T, class U>
    bool operator()(T&& l, U&& r) const { return std::forward<T>(l).i < std::forward<U>(r).i; }
};

which can be used to sort any type with member i supporting <.

4. Pass an anonymus closure (lambda) as comparison parameter to the sorting functions.

struct X {
  int i{}, j{};
};
std::sort(vec_X.begin(), vec_X.end(), [](X const &l, X const &r) { return l.i < r.i; });

Where C++14 enables a even more generic lambda expression:

std::sort(a.begin(), a.end(), [](auto && l, auto && r) { return l.i < r.i; });

which could be wrapped in a macro

#define COMPARATOR(code) [](auto && l, auto && r) -> bool { return code ; }

making ordinary comparator creation quite smooth:

// sort by i
std::sort(v.begin(), v.end(), COMPARATOR(l.i < r.i));
// sort by j
std::sort(v.begin(), v.end(), COMPARATOR(l.j < r.j));

Answer 6

I was curious if there is any measurable impact on performance between the various ways one can call std::sort, so I've created this simple test:

$ cat sort.cpp
#include<algorithm>
#include<iostream>
#include<vector>
#include<chrono>

#define COMPILER_BARRIER() asm volatile("" ::: "memory");

typedef unsigned long int ulint;

using namespace std;

struct S {
  int x;
  int y;
};

#define BODY { return s1.x*s2.y < s2.x*s1.y; }

bool operator<( const S& s1, const S& s2 ) BODY
bool Sgreater_func( const S& s1, const S& s2 ) BODY

struct Sgreater {
  bool operator()( const S& s1, const S& s2 ) const BODY
};

void sort_by_operator(vector<S> & v){
  sort(v.begin(), v.end());
}

void sort_by_lambda(vector<S> & v){
  sort(v.begin(), v.end(), []( const S& s1, const S& s2 ) BODY );
}

void sort_by_functor(vector<S> &v){
  sort(v.begin(), v.end(), Sgreater());
}

void sort_by_function(vector<S> &v){
  sort(v.begin(), v.end(), &Sgreater_func);
}

const int N = 10000000;
vector<S> random_vector;

ulint run(void foo(vector<S> &v)){
  vector<S> tmp(random_vector);
  foo(tmp);
  ulint checksum = 0;
  for(int i=0;i<tmp.size();++i){
     checksum += i *tmp[i].x ^ tmp[i].y;
  }
  return checksum;
}

void measure(void foo(vector<S> & v)){

ulint check_sum = 0;

  // warm up
  const int WARMUP_ROUNDS = 3;
  const int TEST_ROUNDS = 10;

  for(int t=WARMUP_ROUNDS;t--;){
    COMPILER_BARRIER();
    check_sum += run(foo);
    COMPILER_BARRIER();
  }

  for(int t=TEST_ROUNDS;t--;){
    COMPILER_BARRIER();
    auto start = std::chrono::high_resolution_clock::now();
    COMPILER_BARRIER();
    check_sum += run(foo);
    COMPILER_BARRIER();
    auto end = std::chrono::high_resolution_clock::now();
    COMPILER_BARRIER();
    auto duration_ns = std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();

    cout << "Took " << duration_ns << "s to complete round" << endl;
  }

  cout << "Checksum: " << check_sum << endl;
}

#define M(x) \
  cout << "Measure " #x " on " << N << " items:" << endl;\
  measure(x);

int main(){
  random_vector.reserve(N);

  for(int i=0;i<N;++i){
    random_vector.push_back(S{rand(), rand()});
  }

  M(sort_by_operator);
  M(sort_by_lambda);
  M(sort_by_functor);
  M(sort_by_function);
  return 0;
}

What it does is it creates a random vector, and then measures how much time is required to copy it and sort the copy of it (and compute some checksum to avoid too vigorous dead code elimination).

I was compiling with g++ (GCC) 7.2.1 20170829 (Red Hat 7.2.1-1)

$ g++ -O2 -o sort sort.cpp && ./sort

Here are results:

Measure sort_by_operator on 10000000 items:
Took 0.994285s to complete round
Took 0.990162s to complete round
Took 0.992103s to complete round
Took 0.989638s to complete round
Took 0.98105s to complete round
Took 0.991913s to complete round
Took 0.992176s to complete round
Took 0.981706s to complete round
Took 0.99021s to complete round
Took 0.988841s to complete round
Checksum: 18446656212269526361
Measure sort_by_lambda on 10000000 items:
Took 0.974274s to complete round
Took 0.97298s to complete round
Took 0.964506s to complete round
Took 0.96899s to complete round
Took 0.965773s to complete round
Took 0.96457s to complete round
Took 0.974286s to complete round
Took 0.975524s to complete round
Took 0.966238s to complete round
Took 0.964676s to complete round
Checksum: 18446656212269526361
Measure sort_by_functor on 10000000 items:
Took 0.964359s to complete round
Took 0.979619s to complete round
Took 0.974027s to complete round
Took 0.964671s to complete round
Took 0.964764s to complete round
Took 0.966491s to complete round
Took 0.964706s to complete round
Took 0.965115s to complete round
Took 0.964352s to complete round
Took 0.968954s to complete round
Checksum: 18446656212269526361
Measure sort_by_function on 10000000 items:
Took 1.29942s to complete round
Took 1.3029s to complete round
Took 1.29931s to complete round
Took 1.29946s to complete round
Took 1.29837s to complete round
Took 1.30132s to complete round
Took 1.3023s to complete round
Took 1.30997s to complete round
Took 1.30819s to complete round
Took 1.3003s to complete round
Checksum: 18446656212269526361

Looks like all the options except for passing function pointer are very similar, and passing a function pointer causes +30% penalty.

It also looks like the operator< version is ~1% slower (I repeated the test multiple times and the effect persists), which is a bit strange as it suggests that the generated code is different (I lack skill to analyze --save-temps output).