[DF] Wrong regex substitution when generating code to jit #11002
Description
A reproducer:
#include <ROOT/RDataFrame.hxx>
int main() {
{
auto df = ROOT::RDataFrame(10).Define("x", [] { return 42; });
df.Snapshot("t", "f.root");
df.Snapshot("fr", "fr.root");
}
TFile f("f.root");
auto *t = f.Get<TTree>("t");
TFile frf("fr.root");
auto *fr = frf.Get<TTree>("fr");
t->AddFriend(fr);
ROOT::RDataFrame df(*t);
df.Filter("x > 0 && fr.x > 0").Count().GetValue();
}errors out with:
input_line_32:2:67: error: use of undeclared identifier 'fr'
auto func0(const Int_t var0, const Int_t var1){return var0 > 0 && fr.var0 > 0
^
The reason is that in this case we substitute column names with var0, var1 placeholder names starting with "x", resulting in the broken expression with fr.var0.
I think a possible fix is to perform these substitutions from the longest to the shortest column names.
First reported at https://root-forum.cern.ch/t/rdataframe-string-filter-question/50872 .