www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Transforming an argument list

reply Jean-Louis Leroy <jl leroy.nyc> writes:
Another meta-programming question ;-)

Inside a variadic function template that takes a list of 
arguments, I want to call another function and pass it the 
arguments run through another function. In C++ it would look like 
this:

template<typename T>
T double_int(T val) { return val; }

int double_int(int val) { return 2 * val; }

template<typename... T>
void double_ints(void F(T...), T... args) {
     F(double_int(args)...);
}

void test(const char* str, int val) {
     cout << str << " = " << val << "\n";
}

int main() {
     double_ints(test, "1 + 2", 3); // 1 + 2 = 6
     return 0;
}

I tried this in D:

void test(string name, int val)
{
   writefln("%s = %d", name, val);
}

template DoubleInt(A...)
{
   static if (A.length == 0)
     alias DoubleInt = AliasSeq!();
   else static if (is(typeof(A[0]) == int))
     alias DoubleInt = AliasSeq!(2 * A[0], DoubleInt!(A[1..$]));
   else
     alias DoubleInt = AliasSeq!(A[0], DoubleInt!(A[1..$]));
}

void double_ints(F, A...)(F f, A args)
{
   f(DoubleInt!args);
}

void main()
{
   double_ints(&test, "1 + 2", 3);
}

...but ldc2 complains that "variable __args_field_1 cannot be 
read at compile time". Yeah, I understand...

I also tried a solution similar to C++:

A DoubleInt(A)(A a)
   if (!is(typeof(A) == int))
   {
     return a;
   }

A DoubleInt(A)(A a)
   if (is(typeof(A) == int))
   {
     return 2 * a;
   }

void double_ints(F, A...)(F f, A args)
{
   f(DoubleInt!(args));
}

...but I get: "template instance callas.double_ints!(void 
function(string, int), string, int) error instantiating". The 
difference with C++ is that I don't get to control the expansion 
of the argument pack: "f(DoubleInt!(args))" is interpreted as 
"f(DoubleInt!(args...))", not "f(DoubleInt!(args))...".

Any ideas? Thanks...
Jun 20
parent ag0aep6g <anonymous example.com> writes:
On 06/20/2017 10:28 PM, Jean-Louis Leroy wrote:
 Any ideas? Thanks...
1) You can foreach over args: ---- void test(string name, int val) { import std.stdio; writefln("%s = %d", name, val); } T double_int(T)(T val) { return val; } int double_int(int val) { return 2 * val; } void double_ints(F, A...)(F f, A args) { foreach (ref arg; args) arg = double_int(arg); f(args); } void main() { double_ints(&test, "1 + 2", 3); } ---- 2) Or you could make a generic `tuple_map` function and use that: ---- /* ... test, double_int, and main as above ... */ auto tuple_map(alias fun, T ...)(T things) { import std.typecons: Tuple, tuple; static if (things.length == 0) return tuple(); else { auto head = fun(things[0]); auto tail = tuple_map!fun(things[1 .. $]); return tuple(head, tail.expand); } } void double_ints(F, A...)(F f, A args) { f(tuple_map!double_int(args).expand); } ----
Jun 20