[ntuple] ADD PrintFieldsTreeAsDot in Inspector

Author: magnustymoteus

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

@@ -40,8 +40,16 @@ class RPageSource;
 
 namespace Experimental {
 
-enum class ENTupleInspectorPrintFormat { kTable, kCSV };
-enum class ENTupleInspectorHist { kCount, kNElems, kCompressedSize, kUncompressedSize };
+enum class ENTupleInspectorPrintFormat {
+   kTable,
+   kCSV
+};
+enum class ENTupleInspectorHist {
+   kCount,
+   kNElems,
+   kCompressedSize,
+   kUncompressedSize
+};
 
 // clang-format off
 /**
@@ -93,15 +101,16 @@ public:
          : fColumnDescriptor(colDesc),
            fCompressedPageSizes(compressedPageSizes),
            fElementSize(elemSize),
-           fNElements(nElems) {};
+           fNElements(nElems){};
       ~RColumnInspector() = default;
 
       const ROOT::RColumnDescriptor &GetDescriptor() const { return fColumnDescriptor; }
       const std::vector<std::uint64_t> &GetCompressedPageSizes() const { return fCompressedPageSizes; }
       std::uint64_t GetNPages() const { return fCompressedPageSizes.size(); }
       std::uint64_t GetCompressedSize() const
       {
-         return std::accumulate(fCompressedPageSizes.begin(), fCompressedPageSizes.end(), static_cast<std::uint64_t>(0));
+         return std::accumulate(fCompressedPageSizes.begin(), fCompressedPageSizes.end(),
+                                static_cast<std::uint64_t>(0));
       }
       std::uint64_t GetUncompressedSize() const { return fElementSize * fNElements; }
       std::uint64_t GetElementSize() const { return fElementSize; }
@@ -123,7 +132,7 @@ public:
 
    public:
       RFieldTreeInspector(const ROOT::RFieldDescriptor &fieldDesc, std::uint64_t onDiskSize, std::uint64_t inMemSize)
-         : fRootFieldDescriptor(fieldDesc), fCompressedSize(onDiskSize), fUncompressedSize(inMemSize) {};
+         : fRootFieldDescriptor(fieldDesc), fCompressedSize(onDiskSize), fUncompressedSize(inMemSize){};
       ~RFieldTreeInspector() = default;
 
       const ROOT::RFieldDescriptor &GetDescriptor() const { return fRootFieldDescriptor; }
@@ -469,6 +478,23 @@ public:
    {
       return GetFieldsByName(std::regex{std::string(fieldNamePattern)}, searchInSubfields);
    }
+   /////////////////////////////////////////////////////////////////////////////
+   /// \brief Print a .dot string that represents the tree of the (sub)fields of an RNTuple
+   ///
+   /// \param[in] fieldDescriptor The descriptor of the root field (this method works recursively)
+   ///
+
+   void PrintFieldsTreeAsDot(const ROOT::RFieldDescriptor &fieldDescriptor, std::ostream &output = std::cout) const;
+
+   /////////////////////////////////////////////////////////////////////////////
+   /// \brief Print the tree of all the (sub)fields of an RNTuple
+   /// \param[in] output
+   ///
+   /// \see PrintFieldsTreeAsDot(const ROOT::RFieldDescriptor &fieldDescriptor, std::ostream &output=std::cout) const
+   void PrintFieldsTreeAsDot(std::ostream &output = std::cout) const
+   {
+      PrintFieldsTreeAsDot(GetDescriptor().GetFieldZero(), output);
+   }
 };
 } // namespace Experimental
 } // namespace ROOT

tree/ntupleutil/v7/src/RNTupleInspector.cxx

@@ -516,3 +516,38 @@ ROOT::Experimental::RNTupleInspector::GetFieldsByName(const std::regex &fieldNam
 
    return fieldIds;
 }
+
+void ROOT::Experimental::RNTupleInspector::PrintFieldsTreeAsDot(const ROOT::RFieldDescriptor &fieldDescriptor,
+                                                                std::ostream &output) const
+{
+   const auto &tupleDescriptor = GetDescriptor();
+   const bool isZeroField = fieldDescriptor.GetParentId() == ROOT::kInvalidDescriptorId;
+   if (isZeroField) {
+      output << "digraph D {\n";
+      output << "node[shape=box]\n";
+   }
+   const std::string &nodeId = (isZeroField) ? "0" : std::to_string(fieldDescriptor.GetId() + 1);
+   const std::string &fieldName = (isZeroField) ? "RFieldZero" : fieldDescriptor.GetFieldName();
+   const std::string &description = fieldDescriptor.GetFieldDescription();
+   const std::uint32_t &version = fieldDescriptor.GetFieldVersion();
+
+   output << nodeId << "[label=<";
+   if (!isZeroField) {
+      output << "<b>Name: </b>" << fieldName << "<br></br>";
+      output << "<b>Type: </b>" << fieldDescriptor.GetTypeName() << "<br></br>";
+      output << "<b>ID: </b>" << std::to_string(fieldDescriptor.GetId()) << "<br></br>";
+      if (description != "")
+         output << "<b>Description: </b>" << description << "<br></br>";
+      if (version != 0)
+         output << "<b>Version: </b>" << version << "<br></br>";
+   } else
+      output << "<b>" << fieldName << "</b>";
+   output << ">]\n";
+   for (const auto &childFieldId : fieldDescriptor.GetLinkIds()) {
+      const auto &childFieldDescriptor = tupleDescriptor.GetFieldDescriptor(childFieldId);
+      output << nodeId + "->" + std::to_string(childFieldDescriptor.GetId() + 1) + "\n";
+      PrintFieldsTreeAsDot(childFieldDescriptor, output);
+   }
+   if (isZeroField)
+      output << "}";
+}

tree/ntupleutil/v7/test/ntuple_inspector.cxx

@@ -813,3 +813,23 @@ TEST(RNTupleInspector, MultiColumnRepresentations)
    EXPECT_EQ(ENTupleColumnType::kReal16, px1Inspector.GetType());
    EXPECT_EQ(1u, px1Inspector.GetNElements());
 }
+
+TEST(RNTupleInspector, FieldsTreeAsDot)
+{
+   FileRaii fileGuard("test_ntuple_inspector_fields_tree_as_dot.root");
+   {
+      auto model = RNTupleModel::Create();
+      auto fldFloat1 = model->MakeField<float>("float1");
+      auto fldInt = model->MakeField<std::int32_t>("int");
+      auto writer = RNTupleWriter::Recreate(std::move(model), "ntuple", fileGuard.GetPath());
+   }
+   auto inspector = RNTupleInspector::Create("ntuple", fileGuard.GetPath());
+   std::ostringstream dotStream;
+   inspector->PrintFieldsTreeAsDot(dotStream);
+   const std::string dot = dotStream.str();
+
+   EXPECT_NE(dot.find("digraph D {"), std::string::npos);
+   EXPECT_NE(dot.find("0->1"), std::string::npos);
+   EXPECT_NE(dot.find("0->2"), std::string::npos);
+   EXPECT_TRUE(dot.size() >= 2 && dot.substr(dot.size() - 1) == "}");
+}