[cfe-dev] Attach additional information to AST nodes

[Whisperity via cfe-dev]

Hey Benedict,

Depending on where you want to actually "read" the additional
information, you might be better off creating the extra information in
the Traverse* functions (and don't forget to call
Base::TraverseWhatever(ptr); to do the traversal as the base class
would!), and read them in the "Visit" ones.

The hierarchy is that... because types can have subtypes, from what
I've seen, the TraverseWhatever methods are called for nodes that are
EXACTLY Whatever types. But for each node the full hierarchy of Visit*
methods are called... e.g. for a CXXConstructorDecl, VisitDecl,
VisitFunctionDecl, VisitCXXMemberDecl are called, in inheritance
order.

Also, use map::at() for reading from a map, gives better errors. ;)

Regards,
Whisperity.

2018-09-25 17:46 GMT+02:00 Benedikt Lukas Huber via cfe-dev
<cfe-dev at lists.llvm.org>:
> Hi Clang list,
>
> I am implementing a RecursiveASTVisitor so I can rely on a preorder
> traversal.
> In some of the Visit* callbacks I would like to pass some information
> downwards
> and do something like this:
>
>   bool VisitFunctionDecl(const FunctionDecl *d) {
>     ...
>     d->getBody()->attachToContext(myAdditionalData);
>     ...
>   }
>
> So that I can access it when the callback for the body is called:
>
>   bool VisitCompoundStmt(const CompoundStmt *n) {
>     ...
>     MyAdditionalData *myData = n->getFromContext();
>     ...
>   }
>
> Right now I am using a std::map with the pointer as key, i.e.
>
>   bool VisitFunctionDecl(const FunctionDecl *d) {
>     ...
>     myMap[d->getBody()] = myAdditionalData;
>     ...
>   }
>
> and then
>
>   bool VisitCompoundStmt(const CompoundStmt *n) {
>     ...
>     MyAdditionalData *myData = myMap[n];
>     ...
>   }
>
> That works, but it does not seem to be the right thing.
> Is there a way to attach additional information to AST nodes?
> Or should I approach that in a completely different way?
>
> Thank you and best regards,
> Benedikt
>
>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
>