I'm trying to figure out what is the best approach here. Basically I have a system where I receive external requests in order to set/get values in my model. The problem is that my model consists on C++ classes, which can be nested, whereas the requests are simple (key, value) pairs.
For example:
struct Foo {
void setX(int x);
int getX() const;
struct Boo {
void setY(float y);
float getY() const;
}:
};
If I receive a request that says set(y, 21)
for a given element e, then the actions I need to perform will be different depending on whether or not foo and boo already exist. Having to take care of the different possibilities for each property would end up in writing a lot of code.
Before reinventing the wheel, I was wondering if there is already a library or a well-known technique in C++ that allows mapping this flat actions into changes in C++ structures (which can be nested) in a generic way.
Thanks
Boost has Property Maps for this purpose.
- http://www.boost.org/doc/libs/1_57_0/libs/property_map/doc/property_map.html
The most elementary interface it exposes is
get(map, key)
put(pmap, key, val)
For lvalue/readable maps you can also get indexer style access
pmap[key];
pmap[key] = newval; // if not const/readonly
You can a existing property map adaptors:
identity_property_map
and typed_identity_property_map
function_property_map
iterator_property_map
shared_array_property_map
associative_property_map
const_associative_property_map
vector_property_map
ref_property_map
static_property_map
transform_value_property_map
compose_property_map
or write custom ones.
There is even a dynamic_property_map
that looks like this, in practice:
put("age",properties,fred,new_age);
put("gpa",properties,fred,new_gpa);
Note that age
and gpa
could be stored anywhere (even requiring a web-request, perhaps) but the difference in access is abstracted away by the properymap interface that sits in between.
Samples from my answers:
- Is it possible to have several edge weight property maps for one graph BOOST?
- Cut set of a graph, Boost Graph Library using a BiMap and a Boost Multi Index container to store the properties