Improve summary output

examples/example_files.yml

@@ -1,10 +1,11 @@
 'data.root':
   type: data
   legend: 'Data'
+  pretty-name: 'Data'
 
 'MC_sample1.root':
   type: mc
-  sample-name: "TTJets_TuneCUETP8M1_amcatnloFXFX_25ns_15c5fa1_c106444"
+  pretty-name: 'MC sample 1'
   cross-section: 245.8
   generated-events: 2167
   fill-color: "#D95B43"
@@ -14,6 +15,7 @@
 
 'MC_sample2.root':
   type: mc
+  pretty-name: 'MC sample 2'
   cross-section: 666.3
   generated-events: 2404
   fill-color: '#53777A'

external/build-external.sh

@@ -2,10 +2,10 @@
 
 # YAML
 
-curl -O https://yaml-cpp.googlecode.com/files/yaml-cpp-0.5.1.tar.gz
-tar xf yaml-cpp-0.5.1.tar.gz
+curl -O -J -L https://github.com/jbeder/yaml-cpp/archive/release-0.5.3.tar.gz
+tar xf yaml-cpp-release-0.5.3.tar.gz
 
-cd yaml-cpp-0.5.1
+cd yaml-cpp-release-0.5.3
 mkdir build
 cd build
 
@@ -15,7 +15,7 @@ make -j4
 make install
 
 cd ../..
-rm yaml-cpp-0.5.1.tar.gz
+rm yaml-cpp-release-0.5.3.tar.gz
 
 # TCLAP
 curl -L "http://downloads.sourceforge.net/project/tclap/{tclap-1.2.1.tar.gz}?r=http%3A%2F%2Fsourceforge.net%2Fprojects%2Ftclap%2Ffiles%2F&ts=1431017326&use_mirror=freefr" -o "#1"

include/TH1Plotter.h

@@ -9,7 +9,7 @@ namespace plotIt {
         plotter(plotIt) {
         }
 
-      virtual bool plot(TCanvas& c, Plot& plot);
+      virtual boost::optional<Summary> plot(TCanvas& c, Plot& plot);
       virtual bool supports(TObject& object);
 
     private:

include/colors.h

@@ -0,0 +1,20 @@
+#pragma once
+
+#include <iostream>
+
+namespace Color {
+    enum Code {
+        RESET       = 0,
+        FG_RED      = 31,
+        FG_GREEN    = 32,
+        FG_YELLOW   = 33,
+        FG_BLUE     = 34,
+        FG_MAGENTA  = 35,
+        FG_CYAN     = 36,
+        FG_WHITE    = 37
+    };
+
+    std::ostream& operator<<(std::ostream& os, Code code) {
+        return os << "\033[" << static_cast<int>(code) << "m";
+    }
+}

include/plotter.h

@@ -1,6 +1,9 @@
 #pragma once
 
 #include <plotIt.h>
+#include <summary.h>
+
+#include <boost/optional.hpp>
 
 class TCanvas;
 class TObject;
@@ -15,7 +18,7 @@ namespace plotIt {
         }
 
 
-      virtual bool plot(TCanvas& c, Plot& plot) = 0;
+      virtual boost::optional<Summary> plot(TCanvas& c, Plot& plot) = 0;
       virtual bool supports(TObject& object) = 0;
 
     protected:

include/plotters.h

@@ -1,19 +1,22 @@
 #pragma once
 
 #include <TH1Plotter.h>
+#include <summary.h>
+
+#include <boost/optional.hpp>
 
 namespace plotIt {
   static std::vector<std::shared_ptr<plotter>> s_plotters;
   void createPlotters(plotIt& plotIt) {
     s_plotters.push_back(std::make_shared<TH1Plotter>(plotIt));
   }
 
-  bool plot(const File& file, TCanvas& c, Plot& plot) {
+  boost::optional<Summary> plot(const File& file, TCanvas& c, Plot& plot) {
     for (auto& plotter: s_plotters) {
       if (plotter->supports(*file.object))
         return plotter->plot(c, plot);
     }
 
-    return false;
+    return boost::none;
   }
 }

include/summary.h

@@ -0,0 +1,46 @@
+#pragma once
+
+#include <types.h>
+
+#include <map>
+#include <string>
+#include <vector>
+
+namespace plotIt {
+
+    struct SummaryItem {
+        std::string name;
+
+        float events = 0;
+        float events_uncertainty = 0;
+
+        float efficiency = 0;
+        float efficiency_uncertainty = 0;
+    };
+
+    class Summary {
+        public:
+            void add(const Type& type, const SummaryItem& item);
+            void addSystematics(const Type& type, const SummaryItem& item);
+
+            std::vector<SummaryItem> get(const Type& type) const;
+            std::vector<SummaryItem> getSystematics(const Type& type) const;
+
+        private:
+            std::map<Type, std::vector<SummaryItem>> m_items;
+            std::map<Type, std::vector<SummaryItem>> m_systematics_items;
+    };
+
+    class SummaryPrinter {
+        public:
+            virtual void print(const Summary& summary) const = 0;
+    };
+
+    class ConsoleSummaryPrinter: public SummaryPrinter {
+        public:
+            virtual void print(const Summary& summary) const override;
+
+        private:
+            void printItems(const Type& type, const Summary& summary) const;
+    };
+}

include/types.h

@@ -9,6 +9,8 @@
 #include <defines.h>
 #include <uuid.h>
 
+#include <yaml-cpp/yaml.h>
+
 #include <TObject.h>
 #include <TFile.h>
 #include <TChain.h>
@@ -21,6 +23,21 @@ namespace plotIt {
     DATA
   };
 
+  inline std::string type_to_string(const Type& type) {
+      switch (type) {
+        case MC:
+            return "Simulation";
+
+        case SIGNAL:
+            return "Signal";
+
+        case DATA:
+            return "Data";
+      }
+
+      return "Unknown";
+  }
+
   enum ErrorsType {
       Normal = 0,
       Poisson = 1,
@@ -59,25 +76,11 @@ namespace plotIt {
       VERTICAL
   };
 
-  struct Summary {
-    float n_events = 0;
-    float n_events_error = 0;
-
-    float efficiency = 0;
-    float efficiency_error = 0;
-
-    void clear() {
-      n_events = n_events_error = efficiency = efficiency_error = 0;
-    }
-  };
-
   struct Systematic {
     std::string path;
     TObject* object = nullptr;
     std::map<std::string, TObject*> objects;
     std::shared_ptr<TFile> handle;
-
-    Summary summary;
   };
 
   struct LineStyle {
@@ -111,6 +114,7 @@ namespace plotIt {
 
   struct File {
     std::string path;
+    std::string pretty_name;
 
     // For MC and Signal
     float cross_section = 1.;
@@ -130,7 +134,6 @@ namespace plotIt {
     std::vector<Systematic> systematics;
 
     int16_t order = std::numeric_limits<int16_t>::min();
-    Summary summary;
 
     std::shared_ptr<TChain> chain;
     std::shared_ptr<TFile> handle;

include/utilities.h

@@ -164,4 +164,14 @@ namespace plotIt {
       globfree(&glob_result);
       return ret;
   }
+
+  inline std::string truncate(const std::string& str, size_t max_len) {
+      if (str.length() > max_len - 1) {
+          std::string ret = str;
+          ret.resize(max_len - 1);
+          return ret + u8"…";
+      } else {
+          return str;
+      }
+  }
 }

src/TH1Plotter.cc

@@ -1,6 +1,7 @@
 #include <TH1Plotter.h>
 
 #include <TCanvas.h>
+#include <TEfficiency.h>
 #include <TF1.h>
 #include <TFitResult.h>
 #include <TLatex.h>
@@ -61,9 +62,11 @@ namespace plotIt {
     return object.InheritsFrom("TH1");
   }
 
-  bool TH1Plotter::plot(TCanvas& c, Plot& plot) {
+  boost::optional<Summary> TH1Plotter::plot(TCanvas& c, Plot& plot) {
     c.cd();
 
+    Summary global_summary;
+
     // Rescale and style histograms
     for (File& file: m_plotIt.getFiles()) {
       setHistogramStyle(file);
@@ -77,20 +80,35 @@ namespace plotIt {
           factor *= m_plotIt.getConfiguration().scale * file.scale;
         }
 
-        float n_entries = h->Integral();
-        file.summary.efficiency = n_entries / file.generated_events;
-        file.summary.efficiency_error = sqrt( (file.summary.efficiency * (1 - file.summary.efficiency)) / file.generated_events );
+        SummaryItem summary;
+        summary.name = file.pretty_name;
 
-        file.summary.n_events = n_entries * factor;
-        file.summary.n_events_error = m_plotIt.getConfiguration().luminosity * file.cross_section * file.branching_ratio * file.summary.efficiency_error;
-        if (! m_plotIt.getConfiguration().ignore_scales) {
-          file.summary.n_events_error *= m_plotIt.getConfiguration().scale * file.scale;
-        }
+        double integral_error = 0;
+        double integral = h->IntegralAndError(h->GetXaxis()->GetFirst(), h->GetXaxis()->GetLast(), integral_error);
+
+        summary.events = integral * factor;
+        summary.events_uncertainty = integral_error * factor;
+
+        // FIXME: Probably invalid in case of weights...
+
+        // Bayesian efficiency
+        // Taken from https://root.cern.ch/doc/master/TEfficiency_8cxx_source.html#l02428
+
+        // Use a flat prior (equivalent to Beta(1, 1))
+        float alpha = 1.;
+        float beta = 1.;
+
+        summary.efficiency = TEfficiency::BetaMean(integral + alpha, file.generated_events - integral + beta);
+        summary.efficiency_uncertainty = TEfficiency::Bayesian(file.generated_events, integral, 0.68, alpha, beta, true) - summary.efficiency;
+
+        global_summary.add(file.type, summary);
 
         h->Scale(factor);
       } else {
-        file.summary.n_events = h->Integral();
-        file.summary.n_events_error = 0;
+        SummaryItem summary;
+        summary.name = file.pretty_name;
+        summary.events = h->Integral();
+        global_summary.add(file.type, summary);
       }
 
       h->Rebin(plot.rebin);
@@ -225,8 +243,8 @@ namespace plotIt {
         for (Systematic& syst: file.systematics) {
           // This histogram should contains syst errors
           // in percent
-          float total_syst_error = 0;
           TH1* h = dynamic_cast<TH1*>(syst.object);
+          float total_syst_error = 0;
           for (uint32_t i = 1; i <= (uint32_t) mc_histo_syst_only->GetNbinsX(); i++) {
             float total_error = mc_histo_syst_only->GetBinError(i);
             float syst_error_percent = h->GetBinError(i);
@@ -237,9 +255,13 @@ namespace plotIt {
             mc_histo_syst_only->SetBinError(i, std::sqrt(total_error * total_error + syst_error * syst_error));
           }
 
-          syst.summary.n_events = file.summary.n_events;
-          syst.summary.n_events_error = std::sqrt(total_syst_error);
+          SummaryItem summary;
+          summary.name = fs::path(syst.path).stem().native();
+          summary.events_uncertainty = std::sqrt(total_syst_error);
+
+          global_summary.addSystematics(file.type, summary);
         }
+
       }
 
       // Propagate syst errors to the stat + syst histogram
@@ -265,7 +287,7 @@ namespace plotIt {
 
     if (!toDraw.size()) {
       std::cerr << "Error: nothing to draw." << std::endl;
-      return false;
+      return boost::none;
     };
 
     // Sort object by minimum
@@ -610,7 +632,7 @@ namespace plotIt {
     if (hi_pad.get())
       hi_pad->cd();
 
-    return true;
+    return global_summary;
   }
 
   void TH1Plotter::setHistogramStyle(const File& file) {