Add patch to print RNTupleMetrics in RDF

Author: enirolf

tree/dataframe/inc/ROOT/RNTupleDS.hxx

@@ -21,6 +21,7 @@
 #include <ROOT/RDataSource.hxx>
 #include <ROOT/RNTupleUtil.hxx>
 #include <ROOT/RStringView.hxx>
+#include <ROOT/RNTupleMetrics.hxx>
 
 #include <cstdint>
 #include <memory>
@@ -55,6 +56,8 @@ class RNTupleDS final : public ROOT::RDF::RDataSource {
    std::vector<std::string> fColumnTypes;
    std::vector<size_t> fActiveColumns;
 
+   ROOT::Experimental::Detail::RNTupleMetrics fMetrics;
+
    unsigned fNSlots = 0;
    bool fHasSeenAllRanges = false;
 
@@ -66,9 +69,7 @@ class RNTupleDS final : public ROOT::RDF::RDataSource {
    ///    float eta;
    /// };
    /// AddField will recurse into Jet.pt and Jet.eta and provide the two inner fields as std::vector<float> each.
-   void AddField(const RNTupleDescriptor &desc,
-                 std::string_view colName,
-                 DescriptorId_t fieldId,
+   void AddField(const RNTupleDescriptor &desc, std::string_view colName, DescriptorId_t fieldId,
                  std::vector<DescriptorId_t> skeinIDs);
 
 public:
@@ -93,7 +94,7 @@ protected:
    Record_t GetColumnReadersImpl(std::string_view name, const std::type_info &) final;
 };
 
-} // ns Experimental
+} // namespace Experimental
 
 namespace RDF {
 namespace Experimental {
@@ -102,6 +103,6 @@ RDataFrame FromRNTuple(ROOT::Experimental::RNTuple *ntuple);
 } // namespace Experimental
 } // namespace RDF
 
-} // ns ROOT
+} // namespace ROOT
 
 #endif

tree/dataframe/src/RNTupleDS.cxx

@@ -66,7 +66,9 @@ class RRDFCardinalityField : public ROOT::Experimental::Detail::RFieldBase {
 public:
    static std::string TypeName() { return "std::size_t"; }
    RRDFCardinalityField()
-      : ROOT::Experimental::Detail::RFieldBase("", TypeName(), ENTupleStructure::kLeaf, false /* isSimple */) {}
+      : ROOT::Experimental::Detail::RFieldBase("", TypeName(), ENTupleStructure::kLeaf, false /* isSimple */)
+   {
+   }
    RRDFCardinalityField(RRDFCardinalityField &&other) = default;
    RRDFCardinalityField &operator=(RRDFCardinalityField &&other) = default;
    ~RRDFCardinalityField() = default;
@@ -151,7 +153,10 @@ class RNTupleColumnReader : public ROOT::Detail::RDF::RColumnReaderBase {
 
 } // namespace Internal
 
-RNTupleDS::~RNTupleDS() = default;
+RNTupleDS::~RNTupleDS()
+{
+   fMetrics.Print(std::cout);
+}
 
 void RNTupleDS::AddField(const RNTupleDescriptor &desc, std::string_view colName, DescriptorId_t fieldId,
                          std::vector<DescriptorId_t> skeinIDs)
@@ -202,8 +207,8 @@ void RNTupleDS::AddField(const RNTupleDescriptor &desc, std::string_view colName
          cardinalityField->SetOnDiskId(fieldId);
          fColumnNames.emplace_back("R_rdf_sizeof_" + std::string(colName));
          fColumnTypes.emplace_back(cardinalityField->GetType());
-         auto cardColReader = std::make_unique<ROOT::Experimental::Internal::RNTupleColumnReader>(
-            std::move(cardinalityField));
+         auto cardColReader =
+            std::make_unique<ROOT::Experimental::Internal::RNTupleColumnReader>(std::move(cardinalityField));
          fColumnReaderPrototypes.emplace_back(std::move(cardColReader));
 
          for (const auto &f : desc.GetFieldIterable(fieldDesc.GetId())) {
@@ -255,8 +260,8 @@ void RNTupleDS::AddField(const RNTupleDescriptor &desc, std::string_view colName
    if (cardinalityField) {
       fColumnNames.emplace_back("R_rdf_sizeof_" + std::string(colName));
       fColumnTypes.emplace_back(cardinalityField->GetType());
-      auto cardColReader = std::make_unique<ROOT::Experimental::Internal::RNTupleColumnReader>(
-         std::move(cardinalityField));
+      auto cardColReader =
+         std::make_unique<ROOT::Experimental::Internal::RNTupleColumnReader>(std::move(cardinalityField));
       fColumnReaderPrototypes.emplace_back(std::move(cardColReader));
    }
 
@@ -267,9 +272,11 @@ void RNTupleDS::AddField(const RNTupleDescriptor &desc, std::string_view colName
    fColumnReaderPrototypes.emplace_back(std::move(valColReader));
 }
 
-RNTupleDS::RNTupleDS(std::unique_ptr<Detail::RPageSource> pageSource)
+RNTupleDS::RNTupleDS(std::unique_ptr<Detail::RPageSource> pageSource) : fMetrics("RDF")
 {
    pageSource->Attach();
+   fMetrics.ObserveMetrics(pageSource->GetMetrics());
+   fMetrics.Enable();
    auto descriptorGuard = pageSource->GetSharedDescriptorGuard();
    fSources.emplace_back(std::move(pageSource));
 
@@ -355,6 +362,7 @@ void RNTupleDS::SetNSlots(unsigned int nSlots)
 } // namespace ROOT
 
 ROOT::RDataFrame ROOT::RDF::Experimental::FromRNTuple(std::string_view ntupleName, std::string_view fileName)
+
 {
    auto pageSource = ROOT::Experimental::Detail::RPageSource::Create(ntupleName, fileName);
    ROOT::RDataFrame rdf(std::make_unique<ROOT::Experimental::RNTupleDS>(std::move(pageSource)));