rDataFrame declaration in if block

[fstocker]

Hi,
not sure if this question is obsolete.

I’ve been working with rDataFrame to analyse data for a while now.
For some MC analyses sometimes I’d like study a filtered event sample and sometimes I’d like to study the full sample. In order not to rewrite the full macro I use a bool to enable the filtering or not.

However if I create a filtered set of the RDF within the if block I can’t use it outside the block. How can I allocate a variable before the if blocks that then can hold the filtered entities? Just doing RDataFrame d_filter; at the beginning doesn’t work.

I’ve tried to illustrated my intention with a code example.

Thank you
Francesca

int my_analysis(bool doFilter){

RDataFrame d("bla","bla.root");

if(doFilter) {
auto d_filter = d.Filter("variable < 10");
}

else {
auto d_filter = d.Filter("other_variable <10");
}

auto doStuff = d_filter.Define(...);


return 0;
}

[eguiraud]

Hi Francesca,
two ways I can think of:

#include <ROOT/RDataFrame.hxx>

int main() {
   ROOT::RDataFrame df(10);

   // with a helper lambda (could also be a helper function) that you can also call on the spot
   ROOT::RDF::RNode filtered_df2 = [&](bool p) -> ROOT::RDF::RNode { return p ? df.Filter("true") : df.Filter("false"); }(true);

   // with a unique_ptr<RNode>
   std::unique_ptr<ROOT::RDF::RNode> filtered_df;
   if (true) {
      filtered_df = std::make_unique<ROOT::RDF::RNode>(df.Filter("true"));
   } else {
      filtered_df = std::make_unique<ROOT::RDF::RNode>(df.Filter("false"));
   }

   return 0;
}

Cheers,
Enrico

[eguiraud]

Also, with your permission, can I move this thread out of the newbie section? I think it could be useful to reference it in the future and posts in this section self-destruct after some time.

[fstocker]

thank you, this is very helpful. I didn’t know this was possible.

you’re welcome to move the thread wherever most useful.

Francesca