[cfe-dev] Using AST nodes with clang dynamic matchers

[David Come via cfe-dev]

Hello,

I found a way to to what I want usingParser::NamedValueMap. The 
following code works well

struct CALLBACK : public clang::ast_matchers::MatchFinder::MatchCallback {
//...
};
int main() {
   // V is a visitor and v.fct is a CXXMethodDecl*

   dynamic::Diagnostics Diag;
   Parser::NamedValueMap named_map;

   auto eq_node = equalsNode(v.fct);

   named_map["TEST"] = VariantMatcher::SingleMatcher(eq_node);

   Optional<DynTypedMatcher> unbound_matcher = 
Parser::parseMatcherExpression(
       llvm::StringRef("cxxMemberCallExpr(callee(TEST))"), nullptr, 
&named_map,
       &Diag);
   if (!unbound_matcher) {
     std::cout << "erreur creation unbind"
               << "\n";
   }

   auto final_matcher = unbound_matcher->tryBind("pipo");

   ast_matchers::MatchFinder Finder;
   CALLBACK cb;
   Finder.addDynamicMatcher(*final_matcher, &cb);
   Finder.matchAST(ast_context);
}

David
On 09/03/2017 10:20, Manuel Klimek wrote:
> +Sam for ideas
>
> On Wed, Mar 8, 2017 at 6:23 PM David Come via cfe-dev 
> <cfe-dev at lists.llvm.org <mailto:cfe-dev at lists.llvm.org>> wrote:
>
>     Hello,
>
>     I'm building a tool to find AST nodes according to an user
>     provided file
>     using a custom language. I'm also using clang dynamic matchers to find
>     interesting nodes within functions CFG and do a bunch of thing. I
>     would
>     like to re-use one of the node I found previously in one of the clang
>     matchers.
>
>
>     There is equalsNode for static matchers but it obviously does not work
>     with clang dynamic matchers.
>
>
>     My idea would be to circumvent the problem by introducing into the
>     matcher (once created from the string) already bounded nodes and use
>     equalsBoundNode. But I can't find a way to do it.
>
>
>     This could be handy to re-use in clang-query one match result into
>     another one (some kind of persistent matching)
>
>
>     Do you see another way to achieve what I wish to ?
>
>
>     My last resort will be to change clang source code (assuming I'm
>     successful) and let my tool rely on it but I would really like to
>     stick
>     to vanilla clang.
>
>
>     Thanks,
>
>     David.
>
>     _______________________________________________
>     cfe-dev mailing list
>     cfe-dev at lists.llvm.org <mailto:cfe-dev at lists.llvm.org>
>     http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20170314/686cbca6/attachment.html>