[ntuple] Add GetColumnIds methods in Inspector

GetColumnsByFieldId -> GetColumnIdsRecursive, GetColumnsByType -> GetColumnIds

Author: magnustymoteus

tree/ntupleutil/v7/inc/ROOT/RNTupleInspector.hxx

@@ -269,15 +269,15 @@ public:
    /// \param[in] colType The column type to collect, as defined by ROOT::ENTupleColumnType.
    ///
    /// \return A vector containing the physical IDs of columns of the provided type.
-   std::vector<ROOT::DescriptorId_t> GetColumnsByType(ROOT::ENTupleColumnType colType);
+   std::vector<ROOT::DescriptorId_t> GetColumnIds(ROOT::ENTupleColumnType colType);
 
    /////////////////////////////////////////////////////////////////////////////
    /// \brief Get the columns that make up the given field, including its subfields.
    ///
    /// \param [in] fieldId The ID of the field for which to collect the columns.
    ///
    /// \return A vector containing the IDs of all columns for the provided field ID.
-   std::vector<ROOT::DescriptorId_t> GetColumnsByFieldId(ROOT::DescriptorId_t fieldId) const;
+   std::vector<ROOT::DescriptorId_t> GetColumnIdsRecursive(ROOT::DescriptorId_t fieldId) const;
 
    /////////////////////////////////////////////////////////////////////////////
    /// \brief Get all column types present in the RNTuple being inspected.

tree/ntupleutil/v7/src/RNTupleInspector.cxx

@@ -127,7 +127,7 @@ ROOT::Experimental::RNTupleInspector::CollectFieldTreeInfo(ROOT::DescriptorId_t
 }
 
 std::vector<ROOT::DescriptorId_t>
-ROOT::Experimental::RNTupleInspector::GetColumnsByFieldId(ROOT::DescriptorId_t fieldId) const
+ROOT::Experimental::RNTupleInspector::GetColumnIdsRecursive(ROOT::DescriptorId_t fieldId) const
 {
    std::vector<ROOT::DescriptorId_t> colIds;
    std::deque<ROOT::DescriptorId_t> fieldIdQueue{fieldId};
@@ -203,8 +203,7 @@ size_t ROOT::Experimental::RNTupleInspector::GetColumnCountByType(ROOT::ENTupleC
    return typeCount;
 }
 
-std::vector<ROOT::DescriptorId_t>
-ROOT::Experimental::RNTupleInspector::GetColumnsByType(ROOT::ENTupleColumnType colType)
+std::vector<ROOT::DescriptorId_t> ROOT::Experimental::RNTupleInspector::GetColumnIds(ROOT::ENTupleColumnType colType)
 {
    std::vector<ROOT::DescriptorId_t> colIds;
 
@@ -419,7 +418,7 @@ ROOT::Experimental::RNTupleInspector::GetPageSizeDistribution(std::initializer_l
    }
 
    for (const auto colType : colTypeVec) {
-      auto colIds = GetColumnsByType(colType);
+      auto colIds = GetColumnIds(colType);
 
       if (colIds.empty())
          continue;

tree/ntupleutil/v7/test/ntuple_inspector.cxx

@@ -358,9 +358,9 @@ TEST(RNTupleInspector, ColumnTypeCount)
    EXPECT_EQ(3, inspector->GetColumnCountByType(ENTupleColumnType::kSplitInt32));
 }
 
-TEST(RNTupleInspector, ColumnsByType)
+TEST(RNTupleInspector, GetColumnIdsByType)
 {
-   FileRaii fileGuard("test_ntuple_inspector_columns_by_type.root");
+   FileRaii fileGuard("test_ntuple_inspector_column_ids.root");
    {
       auto model = RNTupleModel::Create();
       auto nFldInt1 = model->MakeField<std::int64_t>("int1");
@@ -374,24 +374,53 @@ TEST(RNTupleInspector, ColumnsByType)
    }
 
    auto inspector = RNTupleInspector::Create("ntuple", fileGuard.GetPath());
-
-   EXPECT_EQ(2U, inspector->GetColumnsByType(ENTupleColumnType::kSplitInt64).size());
-   for (const auto colId : inspector->GetColumnsByType(ENTupleColumnType::kSplitInt64)) {
+   EXPECT_EQ(2U, inspector->GetColumnIds(ENTupleColumnType::kSplitInt64).size());
+   for (const auto colId : inspector->GetColumnIds(ENTupleColumnType::kSplitInt64)) {
       EXPECT_EQ(ENTupleColumnType::kSplitInt64, inspector->GetColumnInspector(colId).GetType());
    }
 
-   EXPECT_EQ(2U, inspector->GetColumnsByType(ENTupleColumnType::kSplitReal32).size());
-   for (const auto colId : inspector->GetColumnsByType(ENTupleColumnType::kSplitReal32)) {
+   EXPECT_EQ(2U, inspector->GetColumnIds(ENTupleColumnType::kSplitReal32).size());
+   for (const auto colId : inspector->GetColumnIds(ENTupleColumnType::kSplitReal32)) {
       EXPECT_EQ(ENTupleColumnType::kSplitReal32, inspector->GetColumnInspector(colId).GetType());
    }
 
-   EXPECT_EQ(1U, inspector->GetColumnsByType(ENTupleColumnType::kSplitIndex64).size());
-   EXPECT_EQ(1U, inspector->GetColumnsByType(ENTupleColumnType::kSplitIndex64).size());
-   for (const auto colId : inspector->GetColumnsByType(ENTupleColumnType::kSplitIndex64)) {
+   EXPECT_EQ(1U, inspector->GetColumnIds(ENTupleColumnType::kSplitIndex64).size());
+   for (const auto colId : inspector->GetColumnIds(ENTupleColumnType::kSplitIndex64)) {
       EXPECT_EQ(ENTupleColumnType::kSplitIndex64, inspector->GetColumnInspector(colId).GetType());
    }
 
-   EXPECT_EQ(0U, inspector->GetColumnsByType(ENTupleColumnType::kSplitReal64).size());
+   EXPECT_EQ(0U, inspector->GetColumnIds(ENTupleColumnType::kSplitReal64).size());
+}
+
+TEST(RNTupleInspector, GetColumnIdsRecursive)
+{
+   FileRaii fileGuard("test_ntuple_inspector_column_ids_recursive.root");
+   {
+      auto model = RNTupleModel::Create();
+      auto nFldInt1 = model->MakeField<std::int64_t>("int1");
+      auto nFldInt2 = model->MakeField<std::int64_t>("int2");
+      auto nFldFloat = model->MakeField<float>("float");
+      auto nFldFloatVec = model->MakeField<std::vector<float>>("floatVec");
+
+      auto writeOptions = RNTupleWriteOptions();
+      writeOptions.SetCompression(505);
+      auto ntuple = RNTupleWriter::Recreate(std::move(model), "ntuple", fileGuard.GetPath(), writeOptions);
+   }
+
+   auto inspector = RNTupleInspector::Create("ntuple", fileGuard.GetPath());
+
+   EXPECT_EQ(1U, inspector->GetColumnIdsRecursive(0).size());
+   EXPECT_EQ(1U, inspector->GetColumnIdsRecursive(1).size());
+   for (const auto colId : inspector->GetColumnIdsRecursive(0)) {
+      EXPECT_EQ(ENTupleColumnType::kSplitInt64, inspector->GetColumnInspector(colId).GetType());
+   }
+
+   EXPECT_EQ(1U, inspector->GetColumnIdsRecursive(2).size());
+   for (const auto colId : inspector->GetColumnIdsRecursive(2)) {
+      EXPECT_EQ(ENTupleColumnType::kSplitReal32, inspector->GetColumnInspector(colId).GetType());
+   }
+
+   EXPECT_EQ(2U, inspector->GetColumnIdsRecursive(3).size());
 }
 
 TEST(RNTupleInspector, ColumnTypes)
@@ -566,7 +595,7 @@ TEST(RNTupleInspector, PageSizeDistribution)
 
    auto inspector = RNTupleInspector::Create("ntuple", fileGuard.GetPath());
 
-   int intColId = inspector->GetColumnsByType(ENTupleColumnType::kSplitInt64)[0];
+   int intColId = inspector->GetColumnIds(ENTupleColumnType::kSplitInt64)[0];
    auto intPageSizeHisto = inspector->GetPageSizeDistribution(intColId);
    EXPECT_STREQ("pageSizeHist", intPageSizeHisto->GetName());
    EXPECT_STREQ(Form("Page size distribution for column with ID %d", intColId), intPageSizeHisto->GetTitle());
@@ -586,7 +615,7 @@ TEST(RNTupleInspector, PageSizeDistribution)
    EXPECT_EQ(100, floatPageSizeHisto->GetNbinsX());
    // Make sure that all page sizes are included in the histogram
    int nFloatPages = 0;
-   for (const auto colId : inspector->GetColumnsByType(ENTupleColumnType::kSplitReal32)) {
+   for (const auto colId : inspector->GetColumnIds(ENTupleColumnType::kSplitReal32)) {
       nFloatPages += inspector->GetColumnInspector(colId).GetNPages();
    }
    EXPECT_EQ(nFloatPages, floatPageSizeHisto->Integral());
@@ -634,7 +663,7 @@ TEST(RNTupleInspector, PageSizeDistribution)
    EXPECT_EQ(0, emptyTypeHisto->Integral());
 
    // Requesting a histogram for a column  without pages in the given RNTuple should give an empty histogram
-   auto doubleColumns = inspector->GetColumnsByType(ENTupleColumnType::kSplitReal64);
+   auto doubleColumns = inspector->GetColumnIds(ENTupleColumnType::kSplitReal64);
    ASSERT_EQ(1, doubleColumns.size());
    auto emptyColumnHisto = inspector->GetPageSizeDistribution({doubleColumns[0]});
    EXPECT_EQ(0, emptyColumnHisto->Integral());