I will pose the question as it applies to one method, but know that this solution should be generic enough to apply to any possible method given. Consider the following:
I have a regular old global method:
void MyMethod( int a, int b )
{
cout << a << " , " << b << endl;
}
I have another method that is to call this method.
void Caller( void* method, void* data, int size )
{
// convert method to some function calling convention
// call the method here with the given data
}
This caller should be able to call any method internally without knowing how many parameters it takes and what their data types are. All it really knows about the method is the address to the method and the size, in bytes, of the entire parameter list.
So simply stated, how can I call an arbitrary method and pass it an arbitrary amount of data for it to interpret as parameters?
Essentially, without modifying MyMethod, how do I push the void* data into the registers used as parameters from within Caller? Is this possible? I'm not concerned about safety or portability.
Before Caller is called, I have an array of void*'s that point to the data that can be passed to the internally called method. I'm not sure if this is even the best approach to this problem.
I'm writing a scripting system, in essence, that can call methods from the script. So the methods are stored in a lookup table, where each is given a string name, and have a void* to the actual method to be called. At execution time I know how many parameters the method expects and what types the parameters are (the types are stored as metadata when the method is given an entry in the lookup table). This allows me to convert the string values that are the parameters in the script to the values they actually should be (using a custom conversion system). But the converter returns a void*, because you call it as such:
string s = "123456";
void* result = Converter::Convert( "string*", "int", &s );
I can guarantee that the value stored in result is actually of the requested type (if a converter for this type-pair exists), but have no way of casting to this type, as the type name is merely provided as a string. This makes the converter flexible and really type-indifferent. But it makes handling the values that it returns complicated. So in script I would make a call like this:
MyMethod( 111, 222 )
This would then be parsed, the method name would be used to look up the method address, and the converter would then convert the values it finds into the expected datatypes, but return them as void*. Then a call to Caller would be made, passing in the method address, arguments it has converted, as an array of bytes, and the size of the array of parameter data, in bytes. It is at this point that I need to call that method and pass these parameters. Again, I cannot modify the existing methods it is calling.
I've looked into assembly to pass this data in, but it seems that you have to either make the method naked to read parameters directly in assembly or do something else, and I've never really worked in assembly before. Although if the solution lies in assembly, I'm fine with learning some.
Changing the implementation details slightly, here is how to do what you want
#include <iostream>
#include <boost/any.hpp>
#include <vector>
#include <functional>
#include <map>
#include <string>
using namespace std;
template<class F>
struct ConstructCaller{};
template<class T, int i>
struct TypeAndInt
{
enum{idx = i};
typedef T type;
};
template<class... T>
struct TypeList{};
template<class A, class B>
struct CombineTypeList{};
template<class... T1, class... T2>
struct CombineTypeList<TypeList<T1...>, TypeList<T2...>>
{
typedef TypeList<T1..., T2...> type;
};
template<int idx, class... T>
struct ToTypeAndIntList{
};
template<int idx,class T0, class T1, class... T>
struct ToTypeAndIntList<idx, T0,T1,T...>{
typedef typename CombineTypeList<TypeList<TypeAndInt<T0, idx> >, typename ToTypeAndIntList<idx+1,T1,T...>::type>::type type;
};
template<int idx, class T0>
struct ToTypeAndIntList<idx,T0>{
typedef TypeList < TypeAndInt<T0, idx> > type;
};
template<class... P>
struct ConstructCaller<void(*)(P...)>
{
typedef void(*FuncType)(P...);
FuncType f_;
template<class T>
typename T::type Get(const vector<boost::any>& vec){
return boost::any_cast<typename T::type>(vec.at(T::idx));
}
template<class... TI>
void DoCall(TypeList<TI...>, const vector<boost::any>& vec){
return f_(Get<TI>(vec)...);
}
void operator()(const vector<boost::any>& vec){
typedef typename ToTypeAndIntList<0, P...>::type List_t;
return DoCall(List_t{}, vec);
}
};
std::map < std::string, std::function<void(const std::vector<boost::any>&)>> func_map;
template<class F>
void RegisterFunction(std::string name, F f){
ConstructCaller<F> c;
c.f_ = f;
func_map[name] = c;
}
void MyMethod(int a, int b)
{
cout << a << " , " << b << endl;
}
void MyMethod2(std::string a, int b)
{
cout << a << " , " << b << endl;
}
int main(){
RegisterFunction("MyMethod", &MyMethod);
RegisterFunction("MyMethod2", &MyMethod2);
std::vector<boost::any> vec;
vec.push_back(1);
vec.push_back(2);
func_map["MyMethod"](vec);
vec.clear();
vec.push_back(std::string("Hello World"));
vec.push_back(2);
func_map["MyMethod2"](vec);
}
Note as presented here, this only works with global functions with a void return type.
This solution also uses boost::any which can store any type, and from which you can extract the type later. Thus to use it register your functions. Then created a vector of boost::any and put in your arbitrary values into the vector. Then look up the function name and call like in the example main.
Let me know if you have any questions.
Related
I've found this interesting code here on stackoverflow from:
Using a STL map of function pointers
template<typename T,typename... Args>
T searchAndCall(std::string s1, Args&&... args){
// ....
// auto typeCastedFun = reinterpret_cast<T(*)(Args ...)>(mapVal.first);
auto typeCastedFun = (T(*)(Args ...))(mapVal.first);
//compare the types is equal or not
assert(mapVal.second == std::type_index(typeid(typeCastedFun)));
return typeCastedFun(std::forward<Args>(args)...);
}
};
Basically, mapVal is a map of function pointers casted to void(*)(void) that will be casted back to their original type with this function. What I would like to do know is how typeCastedFun will be deduced when you don't specify the template parameters.
For instance, let's suppose that you had:
int f(const MyClass& a, MyClass b) {...}
... if you have:
MyClass first, second;
searchAndCall<int>(first, second);
What Args... parameter will be deduced? if I recall correctly, using the function casted back to a function with a different signature compared to the original one, should yield undefined behavior. Is there any other alternative?
What I would like to do is a way to store the type of the function somewhere and use this information to do the correct cast. Everything in the most efficient way.
Thanks
[edit1]
More specifically, I'm trying to build a kind of generic function dispatcher, able to call functions (templated with an enum class value) with different signatures using a lookup table for efficiency reasons. No boost::any as it internally uses a new
[edit2] Use of macros is not allowed
The key problem is that by taking the calling argument types directly, and attempting to cast the function pointer, you are losing all implicit conversions.
Your function signature has to match exactly, or you will get UB if you try to call it. And there is generally no way to get the signature from the args without manually specifying it at the call site.
One workaround to try would be to add a wrapper lambda which takes standardized args with pre-specified implicit coversions applied, e.g. T -> const T&, and possibly numeric types -> double.
Then, when you look up the function, you can cast it to use these standardized args, and the calling args will be implicitly converted.
This would rule out functions taking rvalue refs and non-const references, but I don't thing this is unreasonable for a function that you don't know the signature of, unless you want to disregard const-correctness completely.
Also, other implicit conversions wouldn't happen, e.g. Derived& -> Base&, or char* -> std::string, and I don't think there would be an easy way to make that happen without creating extra limitations.
Overall, it's definitely a tricky thing to do in c++, and anything you try will be hacky. This way should be decent enough. The performance overhead of one extra function call (which can be inlined), and possibly some extraneous argument conversions will be overshadowed by the unavoidable RTTI checking.
Here is a sample implementation (also here on ideone):
#include <unordered_map>
#include <typeinfo>
#include <typeindex>
#include <string>
#include <type_traits>
#include <iostream>
#include <assert.h>
#include <cxxabi.h>
#include <sstream>
#include <stdexcept>
template <typename Func, Func f>
struct store_func_helper;
// unix-specific
std::string demangle(const std::string& val) {
int status;
char *realname;
std::string strname = realname = abi::__cxa_demangle(val.c_str(), 0, 0, &status);
free(realname);
return strname;
}
// args will be implicitly converted to arg<T>::type before calling function
// default: convert to const Arg&
template <typename Arg, typename snifae=void>
struct arg {
using type = const Arg&;
};
// numeric types: convert to double.
template <typename Arg>
struct arg <Arg, typename std::enable_if<std::is_arithmetic<Arg>::value, void>::type> {
using type = double;
};
// set more special arg types here.
// Functions stored in the map are first wrapped in a lambda with this signature.
template <typename Ret, typename... Arg>
using func_type = Ret(*)(typename arg<Arg>::type...);
class func_map {
template <typename Func, Func f>
friend class store_func_helper;
public:
template <typename Func, Func f>
void store(const std::string& name){
store_func_helper<Func, f>::call(this, name );
}
template<typename Ret, typename... Args>
Ret call(std::string func, Args... args){
using new_func_type = func_type<Ret, Args...>;
auto& mapVal = m_func_map.at(func);
if (mapVal.second != std::type_index(typeid(new_func_type))){
std::ostringstream ss;
ss << "Error calling function " << func << ", function type: "
<< demangle(mapVal.second.name())
<< ", attempted to call with " << demangle(typeid(new_func_type).name());
throw std::runtime_error(ss.str());
}
auto typeCastedFun = (new_func_type)(mapVal.first);
//args will be implicitly converted to match standardized args
return typeCastedFun(std::forward<Args>(args)...);
};
private:
std::unordered_map<std::string, std::pair<void(*)(),std::type_index> > m_func_map;
};
#define FUNC_MAP_STORE(map, func) (map).store<decltype(&func),&func>(#func);
template <typename Ret, typename... Args, Ret(*f)(Args...)>
struct store_func_helper<Ret(*)(Args...), f> {
static void call (func_map* map, const std::string& name) {
using new_func_type = func_type<Ret, Args...>;
// add a wrapper function, which takes standardized args.
new_func_type lambda = [](typename arg<Args>::type... args) -> Ret {
return (*f)(args...);
};
map->m_func_map.insert(std::make_pair(
name,
std::make_pair((void(*)()) lambda, std::type_index(typeid(lambda)))
));
}
};
//examples
long add (int i, long j){
return i + j;
}
int total_size(std::string arg1, const std::string& arg2) {
return arg1.size() + arg2.size();
}
int main() {
func_map map;
FUNC_MAP_STORE(map, total_size);
FUNC_MAP_STORE(map, add);
std::string arg1="hello", arg2="world";
std::cout << "total_size: " << map.call<int>("total_size", arg1, arg2) << std::endl;
std::cout << "add: " << map.call<long>("add", 3, 4) << std::endl;
}
I'm trying to programming in C++ a framework where the user can indicates a set of functions inside its program where he wants to apply a memoization strategy.
So let's suppose that we have 5 functions in our program f1...f5 and we want to avoid the (expensive) re-computation for the functions f1 and f3 if we already called them with the same input. Notice that each function can have different return and argument types.
I found this solution for the problem, but you can use only double and int.
MY SOLUTION
Ok I wrote this solution for my problem, but I don't know if it's efficient, typesafe or can be written in any more elegant way.
template <typename ReturnType, typename... Args>
function<ReturnType(Args...)> memoize(function<ReturnType(Args...)> func)
{
return ([=](Args... args) mutable {
static map<tuple<Args...>, ReturnType> cache;
tuple<Args...> t(args...);
auto result = cache.insert(make_pair(t, ReturnType{}));
if (result.second) {
// insertion succeeded so the value wasn't cached already
result.first->second = func(args...);
}
return result.first->second;
});
}
struct MultiMemoizator
{
map<string, boost::any> multiCache;
template <typename ReturnType, typename... Args>
void addFunction(string name, function < ReturnType(Args...)> func) {
function < ReturnType(Args...)> cachedFunc = memoize(func);
boost::any anyCachedFunc = cachedFunc;
auto result = multiCache.insert(pair<string, boost::any>(name,anyCachedFunc));
if (!result.second)
cout << "ERROR: key " + name + " was already inserted" << endl;
}
template <typename ReturnType, typename... Args>
ReturnType callFunction(string name, Args... args) {
auto it = multiCache.find(name);
if (it == multiCache.end())
throw KeyNotFound(name);
boost::any anyCachedFunc = it->second;
function < ReturnType(Args...)> cachedFunc = boost::any_cast<function<ReturnType(Args...)>> (anyCachedFunc);
return cachedFunc(args...);
}
};
And this is a possible main:
int main()
{
function<int(int)> intFun = [](int i) {return ++i; };
function<string(string)> stringFun = [](string s) {
return "Hello "+s;
};
MultiMemoizator mem;
mem.addFunction("intFun",intFun);
mem.addFunction("stringFun", stringFun);
try
{
cout << mem.callFunction<int, int>("intFun", 1)<<endl;//print 2
cout << mem.callFunction<string, string>("stringFun", " World!") << endl;//print Hello World!
cout << mem.callFunction<string, string>("TrumpIsADickHead", " World!") << endl;//KeyNotFound thrown
}
catch (boost::bad_any_cast e)
{
cout << "Bad function calling: "<<e.what()<<endl;
return 1;
}
catch (KeyNotFound e)
{
cout << e.what()<<endl;
return 1;
}
}
How about something like this:
template <typename result_t, typename... args_t>
class Memoizer
{
public:
typedef result_t (*function_t)(args_t...);
Memoizer(function_t func) : m_func(func) {}
result_t operator() (args_t... args)
{
auto args_tuple = make_tuple(args...);
auto it = m_results.find(args_tuple);
if (it != m_results.end())
return it->second;
result_t result = m_func(args...);
m_results.insert(make_pair(args_tuple, result));
return result;
}
protected:
function_t m_func;
map<tuple<args_t...>, result_t> m_results;
};
Usage is like this:
// could create make_memoizer like make_tuple to eliminate the template arguments
Memoizer<double, double> memo(fabs);
cout << memo(-123.456);
cout << memo(-123.456); // not recomputed
It's pretty hard to guess at how you're planning to use the functions, with or without memoisation, but for the container-of-various-function<>s aspect you just need a common base class:
#include <iostream>
#include <vector>
#include <functional>
struct Any_Function
{
virtual ~Any_Function() {}
};
template <typename Ret, typename... Args>
struct Function : Any_Function, std::function<Ret(Args...)>
{
template <typename T>
Function(T& f)
: std::function<Ret(Args...)>(f)
{ }
};
int main()
{
std::vector<Any_Function*> fun_vect;
auto* p = new Function<int, double, double, int> { [](double i, double j, int z) {
return int(i + j + z);
} };
fun_vect.push_back(p);
}
The problem with this is how to make it type-safe. Look at this code:
MultiMemoizator mm;
std::string name = "identity";
mm.addFunction(name, identity);
auto result = mm.callFunction(name, 1);
Is the last line correct? Does callFunction have the right number of parameters with the right types? And what is the return type?
The compiler has no way to know that: it has no way of understanding that name is "identity" and even if it did, no way to associate that with the type of the function. And this is not specific to C++, any statically-typed language is going to have the same problem.
One solution (which is basically the one given in Tony D's answer) is to tell the compiler the function signature when you call the function. And if you say it wrong, a runtime error occurs. That could look something like this (you only need to explicitly specify the return type, since the number and type of parameters is inferred):
auto result = mm.callFunction<int>(name, 1);
But this is inelegant and error-prone.
Depending on your exact requirements, what might work better is to use "smart" keys, instead of strings: the key has the function signature embedded in its type, so you don't have to worry about specifying it correctly. That could look something like:
Key<int(int)> identityKey;
mm.addFunction(identityKey, identity);
auto result = mm.callFunction(identityKey, 1);
This way, the types are checked at compile time (both for addFunction and callFunction), which should give you exactly what you want.
I haven't actually implemented this in C++, but I don't see any reason why it should be hard or impossible. Especially since doing something very similar in C# is simple.
you can use vector of functions with signature like void someFunction(void *r, ...) where r is a pointer to result and ... is variadic argument list. Warning: unpacking argument list is really inconvenient and looks more like a hack.
At first glance, how about defining a type that has template arguments that differ for each function, i.e.:
template <class RetType, class ArgType>
class AbstractFunction {
//etc.
}
have the AbstractFunction take a function pointer to the functions f1-f5 with template specializations different for each function. You can then have a generic run_memoized() function, either as a member function of AbstractFunction or a templated function that takes an AbstractFunction as an argument and maintains a memo as it runs it.
The hardest part will be if the functions f1-f5 have more than one argument, in which case you'll need to do some funky things with arglists as template parameters but I think C++14 has some features that might make this possible. An alternative is to rewrite f1-f5 so that they all take a single struct as an argument rather than multiple arguments.
EDIT: Having seen your problem 1, the problem you're running into is that you want to have a data structure whose values are memoized functions, each of which could have different arguments.
I, personally, would solve this just by making the data structure use void* to represent the individual memoized functions, and then in the callFunction() method use an unsafe type cast from void* to the templated MemoizedFunction type you need (you may need to allocate MemoizedFunctions with the "new" operator so that you can convert them to and from void*s.)
If the lack of type safety here irks you, good for you, in that case it may be a reasonable option just to make hand-written helper methods for each of f1-f5 and have callFunction() dispatch one of those functions based on the input string. This will let you use compile-time type checking.
EDIT #2: If you are going to use this approach, you need to change the API for callFunction() slightly so that callFunction has template args matching the return and argument types of the function, for example:
int result = callFunction<int, arglist(double, float)>("double_and_float_to_int", 3.5, 4);
and if the user of this API ever types the argument type or return types incorrectly when using callFunction... pray for their soul because things will explode in very ugly ways.
EDIT #3: You can to some extent do the type checking you need at runtime using std::type_info and storing the typeid() of the argument type and return type in your MemoizedFunction so that you can check whether the template arguments in callFunction() are correct before calling - so you can prevent the explosion above. But this will add a bit of overhead every time you call the function (you could wrap this in a IF_DEBUG_MODE macro to only add this overhead during testing and not in production.)
In the code I register one or multiple function pointer in a manager class.
In this class I have a map that maps the argument types of the function to said function. It may look like so: std::map< std::vector<std::type_index> , void*>
template<typename Ret, typename... Args>
void Register(Ret(*function)(Args...)) {
void* v = (void*)function;
// recursively build type vector and add to the map
}
At runtime the code gets calls (from an external script) with an arbitrary number of arguments. These arguments can be read as primitive data types or as custom types that will be specified at compile time.
With every call from the script, I have to find out which function to call, and then call it. The former is easy and already solved (filling a vector with type_index in a loop), but I can't think of a solution for the latter.
My first approach was using variadic templates in recursion with an added template argument for each read type - but this turned out to be impossible since templates are constructed at compile time, and the arbitrary number of arguments is read at runtime.
Without variadic templates however, I don't see any possibility to achieve this. I considered boost::any instead of void*, but I didn't see how that would solve the need to cast back to the original type. I also thought of using std::function but that would be a templated type, so it could not be stored in a map for functions with different arguments.
(If it's unclear what I'm asking, think of LuaBinds possibility to register overloaded functions. I tried to understand how it's implemented there (without variadic templates, pre-C++11), but to no avail.)
Suppose you had the arguments in a vector of some kind, and a known function (fully).
You can call this. Call the function that does this invoke.
Next, work out how to do this for template<class... Args>. Augment invoke.
So you have written:
typedef std::vector<run_time_stuff> run_time_args;
template<class... Args>
void invoke( void(*func)(Args...), run_time_args rta )
at this point. Note that we know the types of the argument. I do not claim the above is easy to write, but I have faith you can figure it out.
Now we wrap things up:
template<class...Args>
std::function<void(run_time_args)> make_invoker(void(*func)(Args...)){
return [func](run_time_args rta){
invoke(func, rta);
};
}
and now instead of void* you store std::function<void(run_time_args)> -- invokers. When you add the function pointers to the mechanism you use make_invoker instead of casting to void*.
Basically, at the point where we have the type info, we store how to use it. Then where we want to use it, we use the stored code!
Writing invoke is another problem. It will probably involve the indexes trick.
Suppose we support two kinds of arguments -- double and int. The arguments at run time are then loaded into a std::vector< boost::variant<double, int> > as our run_time_args.
Next, let us extend the above invoke function to return an error in the case of parameter type mismatch.
enum class invoke_result {
everything_ok,
error_parameter_count_mismatch,
parameter_type_mismatch,
};
typedef boost::variant<int,double> c;
typedef std::vector<run_time_stuff> run_time_args;
template<class... Args>
invoke_result invoke( void(*func)(Args...), run_time_args rta );
now some boilerplate for the indexes trick:
template<unsigned...Is>struct indexes{typedef indexes type;};
template<unsigned Max,unsigned...Is>struct make_indexes:make_indexes<Max-1, Max-1,Is...>{};
template<unsigned...Is>struct make_indexes<0,Is...>:indexes<Is...>{};
template<unsigned Max>using make_indexes_t=typename make_indexes<Max>::type;
With that, we can write an invoker:
namespace helpers{
template<unsigned...Is, class... Args>
invoke_result invoke( indexes<Is...>, void(*func)(Args...), run_time_args rta ) {
typedef void* pvoid;
if (rta.size() < sizeof...(Is))
return invoke_result::error_parameter_count_mismatch;
pvoid check_array[] = { ((void*)boost::get<Args>( rta[Is] ))... };
for( pvoid p : check_array )
if (!p)
return invoke_result::error_parameter_type_mismatch;
func( (*boost::get<Args>(rts[Is]))... );
}
}
template<class... Args>
invoke_result invoke( void(*func)(Args...), run_time_args rta ) {
return helpers::invoke( make_indexes_t< sizeof...(Args) >{}, func, rta );
}
And that should work when func's args exactly match the ones passed in inside run_time_args.
Note that I was fast and loose with failing to std::move that std::vector around. And that the above doesn't support implicit type conversion. And I didn't compile any of the above code, so it is probably littered with typos.
I was messing around with variadic templates a few weeks ago and came up with a solution that might help you.
DELEGATE.H
template <typename ReturnType, typename ...Args>
class BaseDelegate
{
public:
BaseDelegate()
: m_delegate(nullptr)
{
}
virtual ReturnType Call(Args... args) = 0;
BaseDelegate* m_delegate;
};
template <typename ReturnType = void, typename ...Args>
class Delegate : public BaseDelegate<ReturnType, Args...>
{
public:
template <typename ClassType>
class Callee : public BaseDelegate
{
public:
typedef ReturnType (ClassType::*FncPtr)(Args...);
public:
Callee(ClassType* type, FncPtr function)
: m_type(type)
, m_function(function)
{
}
~Callee()
{
}
ReturnType Call(Args... args)
{
return (m_type->*m_function)(args...);
}
protected:
ClassType* m_type;
FncPtr m_function;
};
public:
template<typename T>
void RegisterCallback(T* type, ReturnType (T::*function)(Args...))
{
m_delegate = new Callee<T>(type, function);
}
ReturnType Call(Args... args)
{
return m_delegate->Call(args...);
}
};
MAIN.CPP
class Foo
{
public:
int Method(int iVal)
{
return iVal * 2;
}
};
int main(int argc, const char* args)
{
Foo foo;
typedef Delegate<int, int> MyDelegate;
MyDelegate m_delegate;
m_delegate.RegisterCallback(&foo, &Foo::Method);
int retVal = m_delegate.Call(10);
return 0;
}
Not sure if your requirements will allow this, but you could possibly just use std::function and std::bind.
The below solution makes the following assumptions:
You know the functions you want to call and their arguments
The functions can have any signature, and any number of arguments
You want to use type erasure to be able to store these functions and arguments, and call them all at a later point in time
Here is a working example:
#include <iostream>
#include <functional>
#include <list>
// list of all bound functions
std::list<std::function<void()>> funcs;
// add a function and its arguments to the list
template<typename Ret, typename... Args, typename... UArgs>
void Register(Ret(*Func)(Args...), UArgs... args)
{
funcs.push_back(std::bind(Func, args...));
}
// call all the bound functions
void CallAll()
{
for (auto& f : funcs)
f();
}
////////////////////////////
// some example functions
////////////////////////////
void foo(int i, double d)
{
std::cout << __func__ << "(" << i << ", " << d << ")" << std::endl;
}
void bar(int i, double d, char c, std::string s)
{
std::cout << __func__ << "(" << i << ", " << d << ", " << c << ", " << s << ")" << std::endl;
}
int main()
{
Register(&foo, 1, 2);
Register(&bar, 7, 3.14, 'c', "Hello world");
CallAll();
}
I would like to create template class which could store function pointer and arguments for a this function so the function can be later invoked with this arguments.
I would like to write this universally and not to depend on argument types or number.
Here is a scatch of the idea with the use of variadic templates of c++11:
template<class T, typename... Params>
class LazyEvaluation {
private:
// Function to be invoked later
T (*f)(Params...);
// Params for function f
Params... storedParams; // This line is not compilable!
bool evaluated;
T result;
public:
// Constructor remembers function pointer and parameters
LazyEvaluation(T (*f)(Params...),Params... params)
: f(f),
storedParams(params) //this line also cannot be compiled
{}
// Method which can be called later to evaluate stored function with stored arguments
operator T&() {
// if not evaluated then evaluate
if (! evaluated) {
result = f(storedParams...);
evaluated = true;
}
return result;
}
}
I would like to have at least the public interface of this class type safe if it is possible. Although getting this work at least somehow is more important.
I've managed to save the variable number of arguments somehow. But I wasn't able to pass them to the function f. I will write it to answers, but I would like you to think about your own solutions before you see my ugly not working attempt.
I am tring to compile the code above with Microsoft Visual C++ Compiler Nov 2012 CTP (v120_CTP_Nov2012), but it would be best if a compiler independent solution would exist.
Thank you
Here is how I tried to solve it:
The parametr pack can be recursivle expanded and each parametr saved. Function store is supposed to do it. It uses one (two times overloaded) helper function.
template<typename T>
void storeHelperFunction(void*& memory, T last) {
*((T*)memory) = last;
memory = (void*)((char*)memory + sizeof(T));
}
template<typename T, typename... Params>
void storeHelperFunction(void*& memory, T first, Params... rest) {
storeHelperFunction(memory, first);
storeHelperFunction(memory, rest...);
}
template<typename... Params>
void store(void* memory, Params... args) {
// Copy of pointer to memory was done when passing it to this function
storeHelperFunction(memory, args...);
}
Function store takes a pointer to memory where the varialbe number of arguments is supposed to be saved.
The pointer can point to some dynamicly allocated memory or beter to the structure which size is equal to sizeof...(Params).
Such structure which has exactly any desiared size can be constructed using template metaprogramming:
template <int N>
struct allocatorStruct {
char byte1;
allocatorStruct<N-1> next;
};
template <>
struct allocatorStruct<1> {};
I am not sure what the standart says or how the other compilers than the microsoft one compile it. But using my compiler the sizeof(allocatorStruct) is equal to N for any N which is greater or equal to 1.
Hence allocatorStruct<sizeof...(Params)> has the same size as Params.
Another way to create something which has the same size as Params is to use a type char [sizeof...(Params)]. This has the disadvantage that the compiler passes only pointer to this array when you try to pass such array as argument.
That is why it is better to use allocatorStruct<sizeof...(Params)>.
And now the main idea:
When saving the function we can cast it to: T (*)(allocatorStruct<sizeof...(Params)>).
When saving the arguments for the function we can save them to struct of the type allocatorStruct<sizeof...(Params)>.
The size of the arguments is the same. Although the function pointer lies about the type of the function the function pointed to will get its data correctly.
At least I hoped. Depending on the calling convention I expected that the passed arguments can be reordered or wrong because of the difference between left to right saving arguments and right to left passing. But it wasn't the case. Using __cdecl calling convention only first argument was passed and the other was lost. With other calling conventions the program stoped working.
I didn't spend much time debugging it and looking to data in memory(on stack). Is it at least right way to go?
Simply use a lambda expression
// Some function.
int add(int a, int b) {
return a + b;
}
auto lazyFunc = [] { return add(1, 2); };
std::cout << lazyFunc() << std::endl; // Evaluate function and output result.
If you really want to create a class that only evaluates the function once (lazily), using variadic templates, you could do something like in the following code.
I also made the class as such that you don't have to create a new instance every time the parameters change. I use a std::tuple to store the given arguments, and compare against previously given arguments. If the arguments differ, then the function will be reevaluated.
Functions are passed around and stored using a std::function wrapper so I don't have to work with raw function pointers (yuck).
#include <iostream>
#include <functional>
#include <utility>
#include <tuple>
template <typename T>
class LazyEvaluation {};
template <typename ReturnType, typename... Params>
class LazyEvaluation<ReturnType(Params...)> {
private:
std::function<ReturnType(Params...)> func_;
ReturnType result;
std::tuple<Params...> oldParams; // Contains the previous arguments.
public:
explicit LazyEvaluation(std::function<ReturnType(Params...)> func)
: func_(std::move(func)) {}
template <typename... Args>
ReturnType operator() (Args&&... args) {
auto newParams = std::make_tuple(std::forward<Args>(args)...);
// Check if new arguments.
if (newParams != oldParams) {
result = func_(std::forward<Args>(args)...);
oldParams = newParams;
std::cout << "Function evaluated" << std::endl;
}
std::cout << "Returned result" << std::endl;
return result;
}
};
int main() {
auto f = [] (int a, int b) {
return a + b;
};
// Specify function type as template parameter.
// E.g. ReturnType(Param1Type, Param2Type, ..., ParamNType)
LazyEvaluation<int(int, int)> ld(f);
std::cout << ld(1, 2) << std::endl;
std::cout << ld(1, 2) << std::endl;
std::cout << ld(3, 4) << std::endl;
}
Output:
Function evaluated
Returned result
3
Returned result
3
Function evaluated
Returned result
7
Given the standard machinery for forming variadic index packs:
template <std::size_t... I> struct index_sequence {};
template <std::size_t N, std::size_t... I>
struct make_index_sequence : public make_index_sequence<N-1, N-1, I...> {};
template <std::size_t... I>
struct make_index_sequence<0, I...> : public index_sequence<I...> {};
and to call functions with unpacked tuple arguments:
template <typename Function, typename... Types, std::size_t... I>
auto apply_(Function&& f, const std::tuple<Types...>& t, index_sequence<I...>)
-> decltype(std::forward<Function>(f)(std::get<I>(t)...)) {
return std::forward<Function>(f)(std::get<I>(t)...);
}
template <typename Function, typename... Types>
auto apply(Function&& f, const std::tuple<Types...>& t)
-> decltype(apply_(f, t, make_index_sequence<sizeof...(Types)>())) {
return apply_(f, t, make_index_sequence<sizeof...(Types)>());
}
This is fairly straightforward:
template<typename Function, typename... Params>
class LazyEvaluation {
private:
typedef decltype(std::declval<Function>()(std::declval<Params>()...)) result_type;
// Function to be invoked later
Function f;
// Params for function f
std::tuple<Params...> storedParams;
mutable bool evaluated;
union {
std::aligned_storage<sizeof(result_type)> space;
mutable result_type result;
};
// Method which can be called later to evaluate stored function with stored arguments
void evaluate() const {
// if not evaluated then evaluate
if (! evaluated) {
new (&result) result_type{apply(f, storedParams)};
evaluated = true;
}
}
public:
// Constructor remembers function pointer and parameters
LazyEvaluation(Function f, Params... params)
: f(std::move(f)),
storedParams(std::move(params)...),
evaluated(false)
{}
~LazyEvaluation() {
if (evaluated)
result.~result_type();
}
operator result_type&() {
evaluate();
return result;
}
operator const result_type& () const {
evaluate();
return result;
}
};
template <typename Function, typename... Params>
LazyEvaluation<Function, Params...>
make_lazy(Function&& f, Params&&... params) {
return {std::forward<Function>(f), std::forward<Params>(params)...};
}
I've used a union and placement new to store the result of evaluation so that it doesn't need to be a default-constructible type, and some mutable tricks so that a const LazyEvaluator can be converted as well as a non-const instance.
I wrote a program in C++ & boost. Is it possible to write a template class producing functors from functions with an unknown number of arguments, e.g. my_call<func>(vector<variant>), where fun can be bool fun(string) or bool fun(int, int, string), etc.?
First, it is important to recognize that boost::variant<> is a class template that requires the list of all the possible types it can hold. So, you won't have just a vector<variant>, but rather a vector<variant<string, double>>, or vector<variant<int, double, string, my_class>>, and you won't be able to mix them.
This made me think you might want to use boost::any rather than boost::variant<>. Thus, I present here a solution that works with boost::variant and can be slightly modified to use boost::any, so you can pick the version you prefer.
To begin with, I must admit that the solution is simple to use but not so simple to understand, so I will have to introduce some machinery first. This machinery is common to both the variant-based and the any-based solution.
//=============================================================================
// META-FUNCTIONS FOR CREATING INDEX LISTS
// The structure that encapsulates index lists
template <size_t... Is>
struct index_list
{
};
// Collects internal details for generating index ranges [MIN, MAX)
namespace detail
{
// Declare primary template for index range builder
template <size_t MIN, size_t N, size_t... Is>
struct range_builder;
// Base step
template <size_t MIN, size_t... Is>
struct range_builder<MIN, MIN, Is...>
{
typedef index_list<Is...> type;
};
// Induction step
template <size_t MIN, size_t N, size_t... Is>
struct range_builder : public range_builder<MIN, N - 1, N - 1, Is...>
{
};
}
// Meta-function that returns a [MIN, MAX) index range
template<size_t MIN, size_t MAX>
using index_range = typename detail::range_builder<MIN, MAX>::type;
The meta-class index_range allows defining compile-time sequences of integers. An interesting proposal have been made by Jonathan Wakely to standardize this kind of construct, so that this whole machinery would not be needed. For the moment, however, we have to hand code this as done above.
Now that we can build compile-time integer sequences, we can exploit variadic templates and argument unpacking to create a dispatching mechanism that translates a vector of variant arguments into a regular argument list. Notice how the concrete variant<> type must be provided as a template argument. This will not be needed for the solution based on any.
// Headers needed for the implementation of the dispatcher
#include <vector>
#include <functional>
#include <boost/variant.hpp>
// Just for convenience
using namespace std;
using boost::variant;
//============================================================================
// DISPATCHER IMPLEMENTATION
// Call dispatching mechanism: notice how the underlying variant type
// must be provided as a template argument (the first one)
template<typename VT, typename R, typename... Args>
struct dispatcher
{
template<typename F>
dispatcher(F f) : _f(f) { }
// The call operator which performs the variant dispatch
R operator () (vector<VT> const& v)
{
if (v.size() != sizeof...(Args))
{
// Wrong number of arguments provided!
return false;
}
// Delegates to internal function call: needed for deducing
// a sequence of integers to be used for unpacking.
index_range<0, sizeof...(Args)> indexes;
return do_call(v, indexes);
}
private:
// The heart of the dispatching mechanism
template<size_t... Is>
R do_call(vector<VT> const& v, index_list<Is...> indexes)
{
return _f((get_ith<Args>(v, Is))...);
}
// Helper function that extracts a typed value from the variant.
template<typename T>
T get_ith(vector<VT> const& v, size_t i)
{
return boost::get<T>(v[i]);
}
// Wrapper that holds the function to be invoked.
function<R(Args...)> _f;
};
// Helper function that allows deducing the input function signature
template<typename VT, typename R, typename... Args>
function<R (vector<VT> const&)> get_dispatcher(R (*f)(Args...))
{
dispatcher<VT, R, Args...> d(f);
return d;
}
Finally, a short demonstration of how you could use this. Suppose we have two test functions such as the ones below:
#include <iostream>
bool test1(string s, double d)
{
cout << s << " " << d << endl;
return true;
}
bool test2(int i1, int i2, string s1, string s2)
{
cout << i1 << " " << i2 << " " << s1 << " " << s2 << endl;
return true;
}
What we want is to invoke them by building a vector of variants and have it dispatched to the desired function. Once again, I must stress the fact that we need to specify the list of all the types our variant can hold. Here, I will assume these types are string, double, and int, but your program might work with different ones.
Also, the solution is based on std::function<> for realizing the type erasure that allows you creating functors of different types and yet invoke them uniformly. Thus, a convenience type definition for this std::function<> (which in turn depends on the variant<> type we use) is provided as well:
int main()
{
// A helper type definition for the variant
typedef variant<int, double, string> vt;
// A helper type definition for the function wrapper
typedef function<bool (vector<vt>)> dispatcher_type;
// Get a caller for the first function
dispatcher_type f1 = get_dispatcher<vt>(test1);
// Prepare arguments for the first function
vector<vt> v = {"hello", 3.14};
// Invoke the first function
f1(v);
// Get a caller for the second function
dispatcher_type f2 = get_dispatcher<vt>(test2);
// Prepare arguments for the second function
v.assign({1, 42, "hello", "world"});
// Invoke the second function
f2(v);
}
Since all dispatchers have type dispatcher_type, you can easily put them into a container. However, you must be aware of the fact that attempts to invoke a function with the wrong number of arguments will be detected only at run-time (it is impossible to know at compile-time how many elements an std::vector<> contains). Thus, proper care must be taken.
As promised, I will now slightly modify this solution to use boost::any rather than boost::variant. The advantage is that since boost::any can hold any value, it is not necessary to specify the list of the possible types which can be used as function arguments.
While the helper machinery is unchanged, the core dispatcher class template must be modified as follows:
#include <vector>
#include <functional>
#include <boost/any.hpp>
using namespace std;
using boost::any;
//=============================================================================
// DISPATCHER IMPLEMENTATION
template<typename R, typename... Args>
struct dispatcher
{
template<typename F>
dispatcher(F f) : _f(f) { }
// The call operator which performs the dispatch
R operator () (vector<any> const& v)
{
if (v.size() != sizeof...(Args))
{
// Wrong number of arguments provided!
return false;
}
// Delegates to internal function call: needed for deducing
// a sequence of integers to be used for unpacking.
index_range<0, sizeof...(Args)> indexes;
return do_call(v, indexes);
}
private:
// The heart of the dispatching mechanism
template<size_t... Is>
R do_call(vector<any> const& v, index_list<Is...> indexes)
{
return _f((get_ith<Args>(v, Is))...);
}
// Helper function that extracts a typed value from the variant.
template<typename T>
T get_ith(vector<any> const& v, size_t i)
{
return boost::any_cast<T>(v[i]);
}
// Wrapper that holds the function to be invoked.
function<R(Args...)> _f;
};
// Helper function
template<typename R, typename... Args>
function<R (vector<any> const&)> get_dispatcher(R (*f)(Args...))
{
dispatcher<R, Args...> d(f);
return d;
}
As you see, the VT template argument has vanished. In particular, it is possible to call get_dispatcher without explicitly specifying any template argument. Using the same test functions we have defined for the variant-based solution, here is how you would adapt the main() routine:
int main()
{
// Helper type definition
typedef function<bool (vector<any>)> dispatcher_type;
// Get a caller for the first function
dispatcher_type f1 = get_dispatcher(test1);
// Get a caller for the second function
dispatcher_type f2 = get_dispatcher(test2);
// Prepare arguments for the first function
vector<any> v = {string("hello"), 3.14};
// Invoke the first function
f1(v);
// Prepare arguments for the second function
v.assign({1, 42, string("hello"), string("world")});
// Invoke the second function
f2(v);
}
The only disadvantage is that with boost::any you cannot assign string literals explicitly, because string literals are of type char [], and arrays cannot be used to initialize objects of type any:
any a = "hello"; // ERROR!
Thus, you have to either wrap them into string objects, or explicitly convert them to a pointer to char const*:
any a = string("hello"); // OK
any b = (char const*)"hello"; // OK
If this is not a huge problem for you, it's probably better to go for this second solution.