include/Dataset.h

#ifndef DATASET_H_
#define DATASET_H_

#include <cstdint>
#include <filesystem>
#include <string>
#include <string_view>
#include <vector>

#include <yaml-cpp/yaml.h>

#include <TChain.h>
#include <TTreeReader.h>

#include <Options.h>


class DatasetInfo {
 public:
  DatasetInfo(std::filesystem::path const &path, Options const &options);

  std::filesystem::path const &DefinitionFile() const {
    return definitionFile_;
  }

  double CrossSection() const {
    return crossSection_;
  }

  std::vector<std::string> const &Files() const {
    return files_;
  }

  bool IsSimulation() const {
    return isSimulation_;
  };

  double MeanWeight() const {
    return meanWeight_;
  }
  
  std::string const &Name() const {
    return name_;
  }

  int64_t NumEventsTotal() const {
    return numEventsTotal_;
  }

  YAML::Node const &Parameters() const {
    return parameters_;
  }

 private:
  YAML::Node const FindStem(std::string_view name) const;

  YAML::Node const GetNode(YAML::Node const root, std::string const &key) const;

  void ReadYaml(std::filesystem::path const &path);

  void SpliceYaml(YAML::Node info) const;

  std::vector<std::filesystem::path> stemsFiles_;

  std::filesystem::path definitionFile_;

  std::vector<std::string> files_;

  YAML::Node parameters_;

  std::string name_;

  bool isSimulation_;

  double crossSection_;

  int64_t numEventsTotal_;

  double meanWeight_;
};


class Dataset {
 public:
  Dataset(DatasetInfo info, int skipFiles = 0, int maxFiles = -1);

  DatasetInfo const &Info() const {
    return info_;
  }

  bool NextEntry() {
    return reader_.Next();
  }

  int64_t NumEntries() {
    return reader_.GetEntries(true);
  }

  TTreeReader &Reader() {
    return reader_;
  }

  std::vector<std::string> const &SelectedFiles() const {
    return selectedFiles_;
  }

  void SetEntry(int64_t index) {
    reader_.SetEntry(index);
  }

 private:
  DatasetInfo info_;

  std::vector<std::string> selectedFiles_;

  TChain chain_;

  TTreeReader reader_;
};

#endif  // DATASET_H_

---