ROOT/RDF/RSample.hxx

0001 // Author: Ivan Kabadzhov CERN  11/2022
0002
0003 /*************************************************************************
0004  * Copyright (C) 1995-2022, Rene Brun and Fons Rademakers.               *
0005  * All rights reserved.                                                  *
0006  *                                                                       *
0007  * For the licensing terms see $ROOTSYS/LICENSE.                         *
0008  * For the list of contributors see $ROOTSYS/README/CREDITS.             *
0009  *************************************************************************/
0010
0011 #ifndef ROOT_RDF_RSAMPLE
0012 #define ROOT_RDF_RSAMPLE
0013
0014 #include <ROOT/RDF/RMetaData.hxx>
0015
0016 #include <string>
0017 #include <vector>
0018
0019 namespace ROOT {
0020 namespace RDF {
0021 namespace Experimental {
0022
0023 /**
0024 \ingroup dataframe
0025 \brief Class representing a sample which is a grouping of trees and their fileglobs, and, optionally, the sample's
0026 metadata information via the RMetaData object.
0027
0028  The class is passed to an RDatasetSpec object in order to build an RDataFrame.
0029
0030  For example, an RSample object can be built as follows:
0031  ~~~{.cpp}
0032  // First, create the RMetaData object (to, optionally, add to the sample)
0033  ROOT::RDF::Experimental::RMetaData meta;
0034  meta.Add("sample_name", "name"");
0035  // Create an RSample with metadata information
0036  ROOT::RDF::Experimental::RSample mySample("mySampleName", "outputTree1", "outputFile.root", meta);
0037  ~~~
0038 */
0039 class RSample {
0040    /// Name of the sample.
0041    std::string fSampleName;
0042    /**
0043     * A list of names of trees.
0044     * This list should go in lockstep with fFileNameGlobs, only in case this dataset is a TChain where each file
0045     * contains its own tree with a different name from the global name of the dataset.
0046     * Otherwise, fTreeNames contains 1 treename, that is common for all file globs.
0047     */
0048    std::vector<std::string> fTreeNames;
0049    /**
0050     * A list of file names.
0051     * They can contain the globbing characters supported by TChain. See TChain::Add for more information.
0052     */
0053    std::vector<std::string> fFileNameGlobs;
0054    /// An instance of the RMetaData class.
0055    RMetaData fMetaData;
0056
0057    /// Global sample index, set inside of the RDatasetSpec.
0058    unsigned int fSampleId{0};
0059
0060 public:
0061    RSample(RSample &&) = default;
0062    RSample &operator=(RSample &&) = default;
0063    RSample(const RSample &) = default;
0064    RSample &operator=(const RSample &) = default;
0065    RSample() = delete;
0066
0067    RSample(const std::string &sampleName, const std::string &treeName, const std::string &fileNameGlob,
0068            const RMetaData &metaData = {});
0069
0070    RSample(const std::string &sampleName, const std::string &treeName, const std::vector<std::string> &fileNameGlobs,
0071            const RMetaData &metaData = {});
0072
0073    RSample(const std::string &sampleName, const std::vector<std::pair<std::string, std::string>> &treeAndFileNameGlobs,
0074            const RMetaData &metaData = {});
0075
0076    RSample(const std::string &sampleName, const std::vector<std::string> &treeNames,
0077            const std::vector<std::string> &fileNameGlobs, const RMetaData &metaData = {});
0078
0079    const std::string &GetSampleName() const;
0080    const std::vector<std::string> &GetTreeNames() const;
0081    const std::vector<std::string> &GetFileNameGlobs() const;
0082    const RMetaData &GetMetaData() const;
0083
0084    /// \cond HIDDEN_SYMBOLS
0085    unsigned int GetSampleId() const; // intended to be used only after the RDataSpec is build, otherwise is 0
0086    void SetSampleId(unsigned int id);
0087    /// \endcond
0088 };
0089
0090 } // namespace Experimental
0091 } // namespace RDF
0092 } // namespace ROOT
0093
0094 #endif // ROOT_RDF_RSAMPLE