%%cpp -d

auto Select = [](ROOT::RDataFrame &dataFrame) {
   using namespace ROOT;

   auto ret = dataFrame.Filter("TMath::Abs(md0_d - 1.8646) < 0.04")
                 .Filter("ptds_d > 2.5")
                 .Filter("TMath::Abs(etads_d) < 1.5")
                 .Filter([](int ik, int ipi, RVecI& nhitrp) { return nhitrp[ik - 1] * nhitrp[ipi - 1] > 1; },
                         {"ik", "ipi", "nhitrp"})
                 .Filter([](int ik, RVecF& rstart, RVecF& rend) { return rend[ik - 1] - rstart[ik - 1] > 22; },
                         {"ik", "rstart", "rend"})
                 .Filter([](int ipi, RVecF& rstart, RVecF& rend) { return rend[ipi - 1] - rstart[ipi - 1] > 22; },
                         {"ipi", "rstart", "rend"})
                 .Filter([](int ik, RVecF& nlhk) { return nlhk[ik - 1] > 0.1; }, {"ik", "nlhk"})
                 .Filter([](int ipi, RVecF& nlhpi) { return nlhpi[ipi - 1] > 0.1; }, {"ipi", "nlhpi"})
                 .Filter([](int ipis, RVecF& nlhpi) { return nlhpi[ipis - 1] > 0.1; }, {"ipis", "nlhpi"})
                 .Filter("njets >= 1");

   return ret;
};

const Double_t dxbin = (0.17 - 0.13) / 40; // Bin-width

%%cpp -d
Double_t fdm5(Double_t *xx, Double_t *par)
{
   Double_t x = xx[0];
   if (x <= 0.13957)
      return 0;
   Double_t xp3 = (x - par[3]) * (x - par[3]);
   Double_t res =
      dxbin * (par[0] * pow(x - 0.13957, par[1]) + par[2] / 2.5066 / par[4] * exp(-xp3 / 2 / par[4] / par[4]));
   return res;
}

%%cpp -d
Double_t fdm2(Double_t *xx, Double_t *par)
{
   static const Double_t sigma = 0.0012;
   Double_t x = xx[0];
   if (x <= 0.13957)
      return 0;
   Double_t xp3 = (x - 0.1454) * (x - 0.1454);
   Double_t res = dxbin * (par[0] * pow(x - 0.13957, 0.25) + par[1] / 2.5066 / sigma * exp(-xp3 / 2 / sigma / sigma));
   return res;
}

%%cpp -d
void FitAndPlotHdmd(TH1 &hdmd)
{
   // create the canvas for the h1analysis fit
   gStyle->SetOptFit();
   auto c1 = new TCanvas("c1", "h1analysis analysis", 10, 10, 800, 600);
   hdmd.GetXaxis()->SetTitleOffset(1.4);

   // fit histogram hdmd with function f5 using the loglikelihood option
   auto f5 = new TF1("f5", fdm5, 0.139, 0.17, 5);
   f5->SetParameters(1000000, .25, 2000, .1454, .001);
   hdmd.Fit("f5", "lr");

   hdmd.DrawClone();
}

%%cpp -d
void FitAndPlotH2(TH2 &h2)
{
   // create the canvas for tau d0
   auto c2 = new TCanvas("c2", "tauD0", 100, 100, 800, 600);

   c2->SetGrid();
   c2->SetBottomMargin(0.15);

   // Project slices of 2-d histogram h2 along X , then fit each slice
   // with function f2 and make a histogram for each fit parameter
   // Note that the generated histograms are added to the list of objects
   // in the current directory.
   auto f2 = new TF1("f2", fdm2, 0.139, 0.17, 2);
   f2->SetParameters(10000, 10);
   h2.FitSlicesX(f2, 0, -1, 1, "qln");

   // See TH2::FitSlicesX documentation
   auto h2_1 = (TH1D *)gDirectory->Get("h2_1");
   h2_1->GetXaxis()->SetTitle("#tau [ps]");
   h2_1->SetMarkerStyle(21);
   h2_1->DrawClone();
   c2->Update();

   auto line = new TLine(0, 0, 0, c2->GetUymax());
   line->Draw();
}

TChain chain("h42");
chain.Add("root://eospublic.cern.ch//eos/root-eos/h1/dstarmb.root");
chain.Add("root://eospublic.cern.ch//eos/root-eos/h1/dstarp1a.root");
chain.Add("root://eospublic.cern.ch//eos/root-eos/h1/dstarp1b.root");
chain.Add("root://eospublic.cern.ch//eos/root-eos/h1/dstarp2.root");

ROOT::EnableImplicitMT(4);

ROOT::RDataFrame dataFrame(chain);
auto selected = Select(dataFrame);

auto hdmdARP = selected.Histo1D({"hdmd", "Dm_d;m_{K#pi#pi} - m_{K#pi}[GeV/c^{2}]", 40, 0.13, 0.17}, "dm_d");
auto selectedAddedBranch = selected.Define("h2_y", "rpd0_t / 0.029979f * 1.8646f / ptd0_d");
auto h2ARP = selectedAddedBranch.Histo2D({"h2", "ptD0 vs Dm_d", 30, 0.135, 0.165, 30, -3, 6}, "dm_d", "h2_y");

FitAndPlotHdmd(*hdmdARP);
FitAndPlotH2(*h2ARP);

gROOT->GetListOfCanvases()->Draw()