I'm trying to write a logic, with usage of command and factory patterns. Based on a data inside the std::variant I want to create an object. It may look like a map of C++ types.
Here is the code: Wandbox
I can hardcode all used types but I want to automate it with variadic templates, how to do it?
#include <cassert>
#include <variant>
#include <type_traits>
#include <memory>
#include <iostream>
struct IFace
{
virtual void foo() = 0;
};
template<typename TKey, typename TValue>
struct Base : IFace
{
using Key = TKey;
using Value = TValue;
};
struct Int : Base<int, Int>
{
void foo() override { std::cout << "Int"; }
};
struct Double : Base<double, Double>
{
void foo() override { std::cout << "Double"; }
};
using Var = std::variant<int, double>;
template<typename ...Args>
std::shared_ptr<IFace> factory(const Var& v)
{
std::shared_ptr<IFace> p;
std::visit([&p](auto&& arg){
using TKey = std::decay_t<decltype(arg)>;
// TODO: use variadic instead of hardcoded types
if constexpr (std::is_same_v<TKey, int>)
{
using TValue = typename Base<TKey, Int>::Value;
p = std::make_shared<TValue>();
std::cout << "int ";
}
else if constexpr (std::is_same_v<TKey, double>)
{
using TValue = typename Base<TKey, Double>::Value;
p = std::make_shared<TValue>();
std::cout << "double ";
}
}, v);
return p;
}
int main()
{
const Var v = 42;
auto p = factory<Int, Double>(v);
assert(p != nullptr);
p->foo();
return 0;
}
This popular little "overloaded" class is always useful for passing lambdas to std::visit:
template<class... Ts> struct overloaded : Ts... { using Ts::operator()...; };
template<class... Ts> overloaded(Ts...) -> overloaded<Ts...>;
Then, instead of using a generic lambda and trying to deduce the right object type from the parameter type, just generate an overload for each Args.
template<typename ...Args>
std::shared_ptr<IFace> factory(const std::variant<typename Args::Key...>& v)
{
return std::visit(overloaded {
[](const typename Args::Key&) -> std::shared_ptr<IFace> { return std::make_shared<Args>(); }...
}, v);
}
Demo: https://godbolt.org/z/nKGf3c
Related
I am trying to implement an std::unordered_map that returns pairs of either double, int or std::string. The keys for the map are std::strings. Below is what I have tried so far:
#include <fstream>
#include <iostream>
#include <string>
#include <sstream>
#include <unordered_map>
#include <utility>
#include <vector>
// A base class for boundary class
class Boundbase {
public:
Boundbase(){};
virtual ~Boundbase(){};
};
// A different map of boundaries for each different data type
template <class dType>
class Boundary : public Boundbase {
std::pair<dType, dType> bpair;
public:
//Constructor
Boundary(const std::string &lbound,
const std::string &ubound) {
setbound(lbound, ubound);
};
//A method to set boundary pair
void setbound(const std::string &lbound,
const std::string &ubound);
// A method to get boundary pair
std::pair<dType, dType> getbound() {return bpair;}
};
// Class to hold the different boundaries
class Boundaries {
std::unordered_map<std::string, Boundbase*> bounds;
public:
//Constructor
Boundaries() {};
// A method to set boundary map
void setboundmap(std::unordered_map<std::string,
std::vector<std::string>> xtb);
// A template to get boundaries.
std::unordered_map<std::string, Boundbase*> getbounds()
{return bounds;}
};
// A method to set covariate boundary
template <class dType> void
Boundary<dType>::setbound(const std::string &lbound,
const std::string &ubound) {
dType val;
std::istringstream isa(lbound);
while(isa >> val) {
bpair.first = val;
}
std::istringstream isb(ubound);
while(isb >> val) {
bpair.second = val;
}
}
// A method to set boundary map
void Boundaries::setboundmap(std::unordered_map<std::string,
std::vector<std::string>> xtb) {
for(auto s : xtb) {
char type = s.second[1][0];
switch(type) {
case 'd': {
std::pair<std::string, Boundbase*> opair;
opair.first = s.first;
opair.second = new Boundary<double>(
s.second[2], s.second[3]);
bounds.insert(opair);
}
break;
case 'i': {
std::pair<std::string, Boundbase*> opair;
opair.first = s.first;
opair.second = new Boundary<int>(
s.second[2], s.second[3]);
bounds.insert(opair);
break;
}
case 'c': {
std::pair<std::string, Boundbase*> opair;
opair.first = s.first;
opair.second = new Boundary<std::string>(
s.second[2], s.second[2]);
bounds.insert(opair);
break;
}
}
}
}
This compiles ok using g++. When I try to run it though ( as follows):
int main() {
Data D;
Boundaries B;
std::ifstream iss("tphinit.txt");
D.read_lines(iss);
auto dbounds = D.get_xtypebound();
B.setboundmap(dbounds);
auto tbounds = B.getbounds();
auto sbound = tbounds["X1"];
std::cout << sbound->bpair.first << ","
<< sbound->bpair.second << std::endl;
}
I get 'class Boundbase' has no member named 'bpair' which is true because I am pointing to the base class and not the derived class. As far as I can tell, trying to get the derived member bpair requires that I use the visitor pattern. Now, it is clear that I am noob so when I had a look at different ways of doing this on SO I was a little in over my head (no reflection on the authors, just on my inexperience).
So my main question is: Is this the best and simplest way to go about this? I would like to avoid boost::variant if at all possible (mainly for the sake of purity: this cannot be that difficult). A sub-question is whether I have to use the visitor pattern or is there a better/simpler way to get the member pbair?
I will have to perform this lookup many times so I am hoping to make it as fast as possible but using the stl for the sake of simplicity.
Make your values std variants over the 3 types.
Failing that, boost variant.
Std and boost variant really are what you want. You'll end up implementing some subset of its implementation.
Failing that, find a tutorial on how to implement ones of them, or use std any. Failing that, dynamic casts around an otherwise useless wrapper type with a virtual dtor stored in a unique ptr, or do manual RTTI with try get methods.
This just gets increasingly ugly and/or inefficient however.
Boost variant, and std variant from it, was implemented for a reason, and that reason was solving the exact problem you are describing in an efficient manner.
#include <tuple>
#include <utility>
#include <string>
template<class...Ts>
struct destroy_helper {
std::tuple<Ts*...> data;
destroy_helper( std::tuple<Ts*...> d ):data(d){}
template<class T>
static void destroy(T* t){ t->~T(); }
template<std::size_t I>
void operator()(std::integral_constant<std::size_t, I>)const {
destroy( std::get<I>( data ) );
}
};
struct construct_helper {
template<class T, class...Args>
void operator()(T* target, Args&&...args)const {
::new( (void*)target ) T(std::forward<Args>(args)...);
}
};
template<std::size_t...Is>
struct indexes {};
template<std::size_t N, std::size_t...Is>
struct make_indexes:make_indexes<N-1, N-1, Is...> {};
template<std::size_t...Is>
struct make_indexes<0, Is...>{
using type=indexes<Is...>;
};
template<std::size_t N>
using make_indexes_t = typename make_indexes<N>::type;
template<class F>
void magic_switch( std::size_t i, indexes<>, F&& f ) {}
template<std::size_t I0, std::size_t...Is, class F>
void magic_switch( std::size_t i, indexes<I0,Is...>, F&& f )
{
if (i==I0) {
f( std::integral_constant<std::size_t, I0>{} );
return;
}
magic_switch( i, indexes<Is...>{}, std::forward<F>(f) );
}
template<class T0>
constexpr T0 max_of( T0 t0 ) {
return t0;
}
template<class T0, class T1, class...Ts>
constexpr T0 max_of( T0 t0, T1 t1, Ts... ts ) {
return (t1 > t0)?max_of(t1, ts...):max_of(t0, ts...);
}
template<class...Ts>
struct Variant{
using Data=typename std::aligned_storage< max_of(sizeof(Ts)...), max_of(alignof(Ts)...)>::type;
std::size_t m_index=-1;
Data m_data;
template<std::size_t I>
using alternative_t=typename std::tuple_element<I, std::tuple<Ts...>>::type;
using pointers=std::tuple<Ts*...>;
using cpointers=std::tuple<Ts const*...>;
template<class T> T& get(){ return *reinterpret_cast<T*>(&m_data); }
template<class T> T const& get() const { return *reinterpret_cast<T*>(&m_data); }
template<std::size_t I>
alternative_t<I>& get(){ return std::get<I>(get_pointers()); }
template<std::size_t I>
alternative_t<I> const& get()const{ return std::get<I>(get_pointers()); }
pointers get_pointers(){
return pointers( (Ts*)&m_data... );
}
cpointers get_pointers()const{
return cpointers( (Ts const*)&m_data... );
}
std::size_t alternative()const{return m_index;}
void destroy() {
if (m_index == -1)
return;
magic_switch(m_index, make_indexes_t<sizeof...(Ts)>{}, destroy_helper<Ts...>(get_pointers()));
}
template<std::size_t I, class...Args>
void emplace(Args&&...args) {
destroy();
construct_helper{}( std::get<I>(get_pointers()), std::forward<Args>(args)... );
m_index = I;
}
Variant()=default;
Variant(Variant const&)=delete;//todo
Variant&operator=(Variant const&)=delete;//todo
Variant(Variant &&)=delete;//todo
Variant&operator=(Variant &&)=delete;//todo
~Variant(){destroy();}
};
int main() {
Variant<int, double, std::string> bob;
bob.emplace<0>( 7 );
bob.emplace<1>( 3.14 );
bob.emplace<2>( "hello world" );
}
here is a really simple variant interface.
The hard part is turning a runtime index into which of the compile time indexes you want to use. I call that the magic switch problem.
You might also want to implement apply visitor.
...
Or...
template<class T>
struct Derived;
struct Base {
virtual ~Base() {}
template<class T>
friend T* get(Base* base) {
Derived<T>* self = dynamic_cast<T*>(base);
return self?&self.t:nullptr;
}
template<class T>
friend T const* get(Base const* base) {
Derived<T> const* self = dynamic_cast<T const*>(base);
return self?&self.t:nullptr;
}
};
template<class T>
struct Derived:Base {
Derived(T in):t(std::move(in)){}
T t;
};
std::unordered_map<std::string, std::unique_ptr<Base>> map;
map["hello"] = std::unique_ptr<Base>( new Derived<int>(-1) );
map["world"] = std::unique_ptr<Base>( new Derived<double>(3.14) );
int* phello = get<int>(map["hello"]);
if (phello) std::cout << *hello << "\n";
double* pworld = get<double>(map["world"]);
if (pworld) std::cout << *world << "\n";
which is a seriously bargain-basement std::any.
I want to create class which should containe map with function pointers (subscribers). But that functions can be with different signature. My code looks like this but it not completed and I am not sure if that is right. Can somebody help me please how to correct append pointers to map and invoke them in myMainClass::start()?
myMainClass.h
#pragma once
#include "iostream";
#include "mySubscriber.h"
struct myMainClass {
myMainClass() {}
~myMainClass() {}
bool callback1(int iData) {
std::cout << "callback 1 with iData " << iData << std::endl;
}
bool callback2(std::string sData) {
std::cout << "callback 2 with sData " << sData << std::endl;
}
bool callback3(int iData, std::string sData) {
std::cout << "callback 1 with iData " << iData << ", sData " << sData << std::endl;
}
// SHOULD BE SOMETHING LIKE THIS
bool start() {
mySubscriber ss;
ss.subscribe("callback1", callback1);
ss.subscribe("callback2", callback2);
ss.getSubscribe("callback1")(5);
ss.getSubscribe("callback2")("test");
}
};
mySubscriber.h
#pragma once
#include "map";
#include "string";
#include "functional";
class mySubscriber {
typedef std::function<void()> func;
std::map<std::string, func*> _subscribes;
public:
mySubscriber() : _subscribes{} {}
~mySubscriber() {
_subscribes.clear();
}
/*
* append or change function pointer
*/
void subscribe(std::string fName, func* f) {
auto find = _subscribes.find(fName);
if (find != _subscribes.end())
{
find->second = f;
}
else
{
_subscribes.emplace(fName, f);
}
}
/*
* get subscribe function
*/
func* getSubscribe(std::string fName) {
auto find = _subscribes.find(fName);
if (find != _subscribes.end())
{
return find->second;
}
return NULL;
}
};
At first some general hints:
Avoid raw pointer usage as far as possible, even for internals! Use std::unique_ptr or std::shared_ptr instead!
Reducing a data conglomerate to a standard container, indexing it via a dynamic data type like std::string and use it in a plain void std::function context results (almost?) always in type erasure and a loss of according type safe outer access. In fact, this even has nothing to do with further differences between plain functions and member methods in the first place.
A possible first solution approach:
This is a minimal working example that should cover your quite dynamic requirements. For me it compiles and runs well with MS VS 2017 (C++17). I tried to use your original structs as far as possible.
#include <variant>
#include <set>
#include <string>
#include <iostream>
struct myMainClass {
myMainClass() {}
~myMainClass() {}
bool callback1(int iData) {
std::cout << "callback 1 with iData " << iData << std::endl;
return true;
}
bool callback2(std::string sData) {
std::cout << "callback 2 with sData " << sData << std::endl;
return true;
}
bool callback3(int iData, std::string sData) {
std::cout << "callback 1 with iData " << iData << ", sData " << sData << std::endl;
return true;
}
template <typename T> class CallbackBaseTmpl;
template <typename Ret, typename ...Args>
class CallbackBaseTmpl<Ret(Args...)>
{
public:
using Signature = Ret(Args...);
CallbackBaseTmpl(const std::function<Signature>& func) : m_function(func) {}
CallbackBaseTmpl(std::function<Signature>&& func) :
m_function(std::move(func)) {}
inline Ret Func(Args&&... args) { return m_function(std::forward<Args>(args)...); }
private:
std::function<Signature> m_function;
};
class Callback1Type : public CallbackBaseTmpl<bool(int)>
{
using CallbackBaseTmpl::CallbackBaseTmpl;
};
class Callback2Type : public CallbackBaseTmpl<bool(std::string)>
{
using CallbackBaseTmpl::CallbackBaseTmpl;
};
class Callback3Type : public CallbackBaseTmpl<bool(int, std::string)>
{
using CallbackBaseTmpl::CallbackBaseTmpl;
};
using CompoundCallbackType = std::variant<Callback1Type, Callback2Type, Callback3Type>;
class CallbackHolder
{
public:
CallbackHolder(const CompoundCallbackType& callbackImpl) : m_callbacksImpl(callbackImpl) {}
inline auto getIndex() const { return m_callbacksImpl.index(); }
inline CompoundCallbackType& getImpl() const { return m_callbacksImpl; }
private:
mutable CompoundCallbackType m_callbacksImpl;
};
class CallbacksContainer
{
public:
template <typename VariantType>
bool subscribe(const VariantType& compoundType)
{
return subscribe(CallbackHolder(compoundType));
}
bool subscribe(const CallbackHolder& cHolder)
{
auto res = m_containerImpl.insert(cHolder);
return res.second;
}
template <typename CallbackType, typename... Args>
auto getSubscribe(Args&&... args)
{
// linear search - can be optimized
for (auto& implEntry : m_containerImpl)
{
bool isWanted = std::visit([&args...](auto&& arg) {
using T = std::decay_t<decltype(arg)>;
if constexpr (std::is_same_v<T, CallbackType>)
return true;
else
return false;
}, implEntry.getImpl());
if (isWanted)
return std::get<CallbackType>(implEntry.getImpl()).Func(std::forward<Args>(args)...);
}
throw std::logic_error("Cannot access element");
}
private:
struct CustomComparer {
bool operator() (const CallbackHolder& lhs, const CallbackHolder& rhs) const
{
// Each variant entry allowed only once in the set
return lhs.getIndex() < rhs.getIndex();
}
};
std::set<CallbackHolder, CustomComparer> m_containerImpl;
};
bool start() {
CallbacksContainer ms;
ms.subscribe(Callback1Type(std::bind(&myMainClass::callback1, this, std::placeholders::_1)));
ms.subscribe(Callback2Type(std::bind(&myMainClass::callback2, this, std::placeholders::_1)));
ms.getSubscribe<Callback1Type>(5);
ms.getSubscribe<Callback2Type>("TEST");
ms.subscribe(Callback3Type(std::bind(&myMainClass::callback3, this, std::placeholders::_1, std::placeholders::_2)));
ms.getSubscribe<Callback3Type>(2, "");
return true;
}
};
Explanation: I replaced your original map with an std::set as a kind of registry container so there are still no duplicates allowed. Some efforts are required via Wrappers to achieve the desired final access scheme.
You can easily change the desired registered functions for a type in a dynamic but always very type safe way now. Feel free to extend this scheme for your own purposes. Likely, there are several parts that can be optimized, shortened or extended. Maybe there's also a nice way to avoid this mutable inside the CallbackHolder. The (non-grave for a few functions) linear search within the set can be avoided via an actual typeid sorting and specialized according finding for instance.
Update due to feedback:
If strings as keys are required and a maximum degree of freedom should be given, i.e. any callback type should be providable without the necessity for compile time registration, this solution might be an alternative:
#include <map>
#include <string>
#include <iostream>
#include <functional>
#include <memory>
struct myMainClass {
myMainClass() {}
~myMainClass() {}
bool callback1(int iData) {
std::cout << "callback 1 with iData " << iData << std::endl;
return true;
}
bool callback2(std::string sData) {
std::cout << "callback 2 with sData " << sData << std::endl;
return true;
}
bool callback3(int iData, std::string sData) {
std::cout << "callback 1 with iData " << iData << ", sData " << sData << std::endl;
return true;
}
class ICallback
{
public:
virtual ~ICallback() = default;
};
template <typename T> class TypedCallback;
template <typename Ret, typename ...Args>
class TypedCallback<Ret(Args...)> : public ICallback
{
public:
using Signature = Ret(Args...);
TypedCallback(const std::function<Signature>& func) : m_function(func) {}
TypedCallback(std::function<Signature>&& func) :
m_function(std::move(func)) {}
inline Ret Func(Args&&... args) { return m_function(std::forward<Args>(args)...); }
private:
std::function<Signature> m_function;
};
class CallbacksContainer
{
private:
template <typename T> struct CallTraits {};
template <typename C, typename Ret, typename... Args>
struct CallTraits<Ret(C::*)(Args...)>
{
using Signature = Ret(Args...);
using ReturnType = Ret;
};
template <typename C, typename Ret, typename... Args>
struct CallTraits<Ret(C::*)(Args...) const>
{
using Signature = Ret(Args...);
using ReturnType = Ret;
};
template <typename F>
struct FuncTraits
{
using FuncClass = std::decay_t<F>;
using OperatorSignature = decltype(&FuncClass::operator());
using signature = typename CallTraits<OperatorSignature>::Signature;
using returnType = typename CallTraits<OperatorSignature>::ReturnType;
};
template <typename Ret, typename... Args>
struct FuncTraits<Ret(Args...)>
{
using Signature = Ret(Args...);
using ReturnType = Ret;
};
template <typename Ret, typename... Args>
struct FuncTraits<Ret(*)(Args...)>
{
using Signature = Ret(Args...);
using ReturnType = Ret;
};
template <typename Ret, typename... Args>
struct FuncTraits<Ret(&)(Args...)>
{
using Signature = Ret(Args...);
using ReturnType = Ret;
};
public:
template <typename T>
bool subscribe(const std::string& key, T&& func)
{
auto res = m_subscriptions.try_emplace(
key, std::make_unique<TypedCallback<typename FuncTraits<T>::signature>>(std::forward<T>(func)));
return res.second;
}
template <typename Ret, typename... Args>
auto getSubscribe(const std::string& key, Args&&... args) const
{
using Signature = Ret(Args...);
const auto& entry = m_subscriptions.at(key);
auto rp = entry.get();
auto typedCB = dynamic_cast<TypedCallback<Signature>*>(rp);
if (typedCB == nullptr)
{
// TODO: Possible further check if functor can be used due to convertible types, for instance
// with an acyclic visitor?
std::logic_error("Wrong callback signature provided.");
}
return typedCB->Func(std::forward<Args>(args)...);
}
private:
std::map<std::string, std::unique_ptr<ICallback>> m_subscriptions;
};
bool start() {
CallbacksContainer ms;
// Usage with non static member methods
ms.subscribe("callback1", [this](int x) { return callback1(x); });
ms.subscribe("callback2", [this](std::string x) { return callback2(x); });
ms.subscribe("callback3", [this](int x, std::string str) { return callback3(x, str); });
// Usage with lambda
ms.subscribe("callback4", [](int y) { return y != 0; });
// Usage with std::function itself
ms.subscribe("callback5", std::function<bool(int)>([](int y) { return y != 0; }));
// Getters - Unfortunately, exact types are required. Maybe acyclic visitor could help here?
ms.getSubscribe<bool>("callback1", 1);
ms.getSubscribe<bool>("callback2", std::string("TEST"));
ms.getSubscribe<bool>("callback3", 1, std::string("TEST"));
ms.getSubscribe<bool>("callback4", 1);
return true;
}
};
PROs:
No static/compile time method signature registration required -> no variants
At least with C++20, method subscription will be an easy going here, added some helper traits to make things a bit easier here already
Only one underlying map used
CONs:
Less type-safety at some points and the dynamic_cast might be a bit slow but might be improved in terms of performance via a simple type index comparison
The getSubscribe() method has to be used with care. Exact types are required here (the former dynamically registered ones) and it doesn't unfortunately support common signature conversion ways. I see currently no way to get rid of this problem with pre C++20 features. Maybe some tricks with a generic acyclic visitor pattern or SFINAE magic + visitor might help here but that breaks the mould by far here I think. If that arises as a real issue, one can still use the chained parameter scheme in doubt, that guarantees type safety on its own.
You have to somehow turn the memberfunction pointers to regular old function pointers, in order to store them in the same container. You have three options that I can come up with:
#include <functional>
struct Foo {
void foo(int x, int y, int z) {}
/*
Putting the instance as the first parameter is crucial, because the
first argument to a member function call is an implicit this. If instance
is not the first parameter the compiler has to shift around the argument
list, otherwise it's a direct forwarding call.
*/
static void callback(void* instance, int x, int y, int z) {
return static_cast<Foo*>(instance)->foo(x, y, z);
}
};
int main() {
Foo foo;
void (*f0)(void*, int, int, int){&Foo::callback};
/*
Capturing lambda cannot decay to function pointer, have to use
std::function or smth. similar
*/
std::function<void(int, int, int)> f1{
[&](int x, int y, int z) { return foo.foo(x, y, z); }};
auto f2 = std::mem_fn(&Foo::foo);
f0(&foo, 1, 2, 3);
f1(1, 2, 3);
f2(&foo, 1, 2, 3);
}
Here's a godbolt with the generated assembly https://godbolt.org/z/K9eM4E
I am trying to implement a composite pattern for std::functions with use of template classes, where each composite class processes the return values of its children.
So the pattern classes might look something like this:
class AbstractClass {
public:
virtual void process() = 0;
};
template<typename ReturnType>
class PrimitiveClass : public AbstractClass {
public:
ReturnType process() {
// please note, that the result is not returned by the return statement
return this->func(); //this is just for simplicity
}
private:
std::function<ReturnType()> func;
}
template<typename ReturnType, typename ...Args>
class CompositeClass : public AbstractClass {
public:
ReturnType process() {
// --> This is where I want to process all children first and then pass their return values to this->func
// the following code is kind of a pseudo code:
for(auto it = vector.begin(); it != vector.end(); ++it {
results.add((**it).process())
}
return this->func(results)
}
private:
std::function<ReturnType(Args...)> func;
std::vector<std::shared_ptr<AbstractClass>> children;
};
So for example, I have a CompositeClass with a std::function<int(int, double, bool) and the argument types of that function are also the ReturnTypes of its children. And I want to pass the return values of the children to above-mentioned std::function
Can anyone think of a way, how I can achieve this?
If I understand what you want (and if I'm not wrong)...
(1) to solve the problem of the no-covariant returned value from process() (see comment from Igor Tandetnik) you need a template abstract class to express the correct return value; by example
template <typename T>
struct abstClass
{ virtual T process() const = 0; };
(2) so your CompositeClass (renamed nodeClass, in my following example) inherit from abstClass<ReturnType>
(3) your PrimitiveClass is useless because you can manage the case (reference to a function without arguments) as a CompositeClass with zero Args
(4) you need a leafClass to handle basic values
(5) in CompositeClass (nodeClass), children, instead of a std::vector of shared_ptr<AbstractClass> (that can't do what do you want), can be a
std::tuple<std::shared_ptr<abstClass<Args>>...> children;
Given these points, I propose the following solution (that, unfortunately, is C++14 because use std::index_sequence and std::make_index_sequence that are available starting from C++14; but if you need a C++11 solution, isn't difficult write substitutes for they)
#include <tuple>
#include <memory>
#include <iostream>
#include <functional>
template <typename T>
struct abstClass
{ virtual T process() const = 0; };
template <typename T>
class leafClass : public abstClass<T>
{
private:
T value;
public:
leafClass (T && v0) : value { std::forward<T>(v0) }
{ }
T process () const
{ return value; };
};
template <typename RetT, typename ... ArgTs>
class nodeClass : public abstClass<RetT>
{
private:
using funcT = std::function<RetT(ArgTs...)>;
template <typename T>
using shrPAC = std::shared_ptr<abstClass<T>>;
funcT func;
std::tuple<shrPAC<ArgTs>...> childrens;
template <std::size_t ... Is>
RetT processH (std::index_sequence<Is...> const &) const
{ return func(std::get<Is>(childrens)->process()...); }
public:
nodeClass (funcT && f0, shrPAC<ArgTs> && ... as)
: func { std::forward<funcT>(f0) },
childrens { std::forward<shrPAC<ArgTs>>(as)... }
{ }
RetT process () const
{ return processH(std::make_index_sequence<sizeof...(ArgTs)>{}); }
};
int main ()
{
auto func0 = [](int i, double d, bool b) { return int( b ? i+d : i-d ); };
auto shpLci = std::make_shared<leafClass<int>>(1);
auto shpLcd = std::make_shared<leafClass<double>>(2.2);
auto shpNb = std::make_shared<nodeClass<bool>>([](){ return true; });
auto shpNc0 = std::make_shared<nodeClass<int, int, double, bool>>
(func0, shpLci, shpLcd, shpNb);
auto shpNc1 = std::make_shared<nodeClass<int, int, double, bool>>
(func0, shpNc0, shpLcd, shpNb);
auto shpNc2 = std::make_shared<nodeClass<int, int, double, bool>>
(func0, shpNc1, shpLcd, shpNb);
std::cout << shpNc0->process() << std::endl; // print 3
std::cout << shpNc1->process() << std::endl; // print 5
std::cout << shpNc2->process() << std::endl; // print 7
}
I want to implement a polymorphic visitor using lambdas without implementing a class. I already have a foundation but am struggling with the type deduction for the parameters of my lambdas.
Let's say I have some legacy code base that decided to use type tags for a polymorphic type like so:
enum class ClassType
{
BaseType = 0, TypeA, TypeB
};
class BaseType
{
public:
virtual ~BaseType() {}
ClassType getType() const
{ return type; }
protected:
ClassType type;
};
class TypeA : public BaseType
{
public:
static const ClassType Type = ClassType::TypeA;
explicit TypeA(int val) : val(val)
{ type = ClassType::TypeA; }
virtual ~TypeA() {}
int val;
};
class TypeB : public BaseType
{
public:
static const ClassType Type = ClassType::TypeB;
explicit TypeB(std::string s) : s(s)
{ type = ClassType::TypeB; }
virtual ~TypeB() {}
std::string s;
};
What I want to achieve is a visitor similar to the std::variant visitors that would then look like this:
std::vector<BaseType*> elements;
elements.emplace_back(new TypeA(1));
elements.emplace_back(new TypeB("hello"));
for (auto elem : elements)
{
visit(elem,
[](TypeA* typeA) {
std::cout << "Found TypeA element, val=" << typeA->val << std::endl;
},
[](TypeB* typeB) {
std::cout << "Found TypeB element, s=" << typeB->s << std::endl;
}
);
}
My so far failing approach for implementing such a visit<>() function was the following code:
template <typename T>
struct identity
{
typedef T type;
};
template <typename T>
void apply_(BaseType* b, typename identity<std::function<void(T*)>&>::type visitor)
{
if (b->getType() != T::Type)
return;
T* t = dynamic_cast<T*>(b);
if (t) visitor(t);
}
template <typename... Ts>
void visit(BaseType* b, Ts... visitors) {
std::initializer_list<int>{ (apply_(b, visitors), 0)... };
}
The compiler complains that it cannot deduce the template parameter T for my apply_ function.
How can I declare the correct template and function signature of apply_ to correctly capture lambdas and maybe even other callables? Or is something like this even possible at all?
Here's an (incomplete) solution that works with any function object that has an unary, non-overloaded, non-templated operator(). Firstly, let's create an helper type alias to retrieve the type of the first argument:
template <typename>
struct deduce_arg_type;
template <typename Return, typename X, typename T>
struct deduce_arg_type<Return(X::*)(T) const>
{
using type = T;
};
template <typename F>
using arg_type = typename deduce_arg_type<decltype(&F::operator())>::type;
Then, we can use a fold expression in a variadic template to call any function object for which dynamic_cast succeeds:
template <typename Base, typename... Fs>
void visit(Base* ptr, Fs&&... fs)
{
const auto attempt = [&](auto&& f)
{
using f_type = std::decay_t<decltype(f)>;
using p_type = arg_type<f_type>;
if(auto cp = dynamic_cast<p_type>(ptr); cp != nullptr)
{
std::forward<decltype(f)>(f)(cp);
}
};
(attempt(std::forward<Fs>(fs)), ...);
}
Usage example:
int main()
{
std::vector<std::unique_ptr<Base>> v;
v.emplace_back(std::make_unique<A>());
v.emplace_back(std::make_unique<B>());
v.emplace_back(std::make_unique<C>());
for(const auto& p : v)
{
visit(p.get(), [](const A*){ std::cout << "A"; },
[](const B*){ std::cout << "B"; },
[](const C*){ std::cout << "C"; });
}
}
ABC
live example on wandbox
Assuming that you cannot change the virtual classes, you may do the following:
template <typename F>
decltype(auto) visitBaseType(BaseType& base, F&& f)
{
switch (base.getType())
{
case ClassType::BaseType: return f(base);
case ClassType::TypeA: return f(dynamic_cast<TypeA&>(base));
case ClassType::TypeB: return f(dynamic_cast<TypeB&>(base));
}
throw std::runtime_error("Bad type");
}
template<class... Ts> struct overloaded : Ts... {
using Ts::operator()...;
overloaded(Ts... ts) : Ts(ts)... {}
};
template<class... Ts> overloaded(Ts...) -> overloaded<Ts...>;
template <typename ... Fs>
decltype(auto) visit(BaseType& base, Fs&&... fs)
{
return visitBaseType(base, overloaded(fs...));
}
Demo
I don't always say this, but this may be a job for the Boost.Preprocessor. You have a list of class types that corresponds to a list of enums, each instance identifies itself via getType(). So we can use that:
#include <boost/preprocessor/seq/for_each.hpp>
#define CLASS_LIST (TypeA) (TypeB)
// just take one visitor
template <class Visitor>
void visit(Base* ptr, Visitor f) {
switch (ptr->getType()) {
#define CASE_ST(r, data, elem) case elem: f(static_cast<elem*>(ptr)); break;
BOOST_PP_SEQ_FOR_EACH(CASE_ST, ~, CLASS_LIST)
#undef CASE_ST
default: f(ptr); // in case you want an "else"
// this is optional
}
}
That will preprocess into:
switch (ptr->getType()) {
case TypeA: f(static_cast<TypeA*>(ptr)); break;
case TypeB: f(static_cast<TypeB*>(ptr)); break;
default: f(ptr);
}
I have the following code which allows me to instantiate and then call a list of void() functions.
(I am using https://github.com/philsquared/Catch for unit testing if you wish to compile and run this code).
#include "catch.hpp"
#include <functional>
#include <vector>
class ChainOfResponsibility : public std::vector<std::function<void()> >, public std::function<void()>
{
public:
void operator()() const
{
for(std::vector<std::function<void()> >::const_iterator it = begin(); it != end(); ++it) {
(*it)();
}
}
};
TEST_CASE("ChainOfResponsibility calls its members when invoked")
{
bool test_function_called = false;
std::function<void()> test_function = [&]()
{
test_function_called = true;
};
ChainOfResponsibility object_under_test;
object_under_test.push_back(test_function);
object_under_test();
REQUIRE(test_function_called);
}
My question is how do I template the ChainOfResponsibility class to accept functions with a different (but consistent) signature?
For example, consider a ChainOfResponsibility<void(int)> or a ChainOfResponsibility<ReturnClass(Argument1Class, Argument2Class)>.
For the sake of argument, lets say that the second example returns the value returned by the last member in the chain, or the default value for ReturnClass if the chain is empty.
Also, if the STL already contains a template class that achieves this, then I would prefer to use it over my home-grown class.
Your specific "discard all the intermediate results" is also fairly simple, but I think it's a bad idea.
template<typename Ret, typename ... Args>
class ChainOfResponsibility
{
std::vector<std::function<Ret(Args...)> > chain;
public:
Ret operator()(Args ... args) const
{
Ret value;
for(auto & func : chain) {
value = func(args...);
}
return value;
}
};
void has to be treated on it's own
template<typename ... Args>
class ChainOfResponsibility<void, Args...>
{
std::vector<std::function<void(Args...)> > chain;
public:
void operator()(Args ... args) const
{
for(auto & func : chain) {
func(args...);
}
}
};
Note that deriving from std:: types is a bad idea, especially std::function, which is a type-erasing callable, not "the base of all callables". You can simply provide an operator()
options for improving the non-void case:
// fold the results
template <typename BinaryFunction>
Ret operator()(Args ... args, BinaryFunction add, Ret init) const
{
for(auto & func : chain) {
init = add(init, func(args...));
}
return init;
}
// return a vector
template <typename BinaryFunction>
std::vector<Ret> operator()(Args ... args) const
{
std::vector<Ret> results(chain.size());
for(auto & func : chain) {
results.push_back(func(args...));
}
return results;
}
You don't need to use the std::function as a base class, using std::vector is sufficent. The template ChainOfResponsibility can use the same template paramter list as the std::function like follows:
#include <iostream>
#include <string>
#include <functional>
#include <vector>
template<typename>
class ChainOfResponsibility;
template<typename R, typename... Args>
class ChainOfResponsibility<R(Args...)> :
public std::vector<std::function<R(Args...)>> {
public:
R operator()(const Args&... args) {
R result {};
for(auto it = this->begin(); it != this->end(); ++it)
result = (*it)(args...);
return result;
}
};
int main() {
ChainOfResponsibility<std::string(int, int)> tester;
tester.push_back([](int a, int b)->std::string {
return std::to_string(a + b);
});
std::cout << tester(4, 2) << std::endl;
}
Anyway, using std::vector only is good enoug for the problem you described. If the content of the overloaded operator() is nothing special, you can change my example above as follows:
int main() {
std::vector<std::function<std::string(int, int)>> tester;
tester.push_back([](int a, int b)->std::string {
return std::to_string(a + b);
});
std::string result;
for(auto& test_fn : tester)
result = test_fn(4, 2);
std::cout << result << std::endl;
}
You also can write a function template instead of overloading the operator():
template<typename R, typename... Args>
R perform(const std::vector<std::function<R(Args...)>>& functions,
const Args&... args) {
R result {};
for(auto& test_fn : functions)
result = test_fn(4, 2);
return result;
}