Re: multi block macro
On May 30, 8:17 pm, gel...@gmail.com wrote:
Is it any possibility to define a macro that produces several blocks.
For example the input is :
BEFIN
ITEM(one)
ITEM(two)
ITEM(three)
END
or, alternatively,
ITEMS(one, two, three)
After preprocessing I need:
enum {one, two, three};
void on_one();
void on_two();
void on_three();
int map(char* id)
{
if (strcmp(id, "one") == 0) return one;
if (strcmp(id, "two") == 0) return two;
if (strcmp(id, "three") == 0) return three;
}
You can reformulate this so that you're doing it in a different way.
This is the way that I've done this sort of thing in the past.
Use a global for the next id number:
int g_nextID = 1;
Use a map to store the function to run given an id or a name (use
wstring because JavaScript is UTF-16):
std::map< int, boost::function< void () > > g_byID;
std::map< std::wstring, boost::function< void () > > g_byName;
A superclass can now handle the registration:
class Function {
public:
Function( const std::wstring &name )
: m_id( g_nextID++ ), m_name( name ) {
g_byID[ m_id ] = boost::bind( &Function::execute, this );
g_byName[ m_name ] = g_byID[ m_id ];
}
virtual void execute() const = 0;
};
Now to add a "read" function:
class Read : public Function {
public:
Read() : Function( L"read" ) {}
void execute() const {
// Code to execute
}
} g_doRead;
To execute then by name or by id you simply execute the thing in the
map:
g_byID[ theID ]();
or
g_byName[ theName ]();
You can use boost::lambda to bind extra parameters and then pass the
missing ones in at invocation.
It looks different, but works just the same. If the ID numbers must be
g'teed the same all the time then issue the ID numbers yourself and
pass them from the constructor like the name. Or consider generating
the ID as a hash of the name.
A big advantage of this is that you can late load libraries (using
something like LoadLibrary in Windows) so the range of functions
supported can be extended much more neatly than otherwise possible as
you don't have an enum to edit.
K