Add test for std::string

... with all possible index column types.

Closes #2

Author: hahnjo

types/README.md

@@ -1,3 +1,4 @@
 # Types
 
  * [`fundamental`](fundamental): fundamental column types
+ * [`string`](string): `std::string` with all `[Split]Index{32,64}` column types

types/fundamental/README.md

@@ -5,7 +5,7 @@
  * [`real`](real): `Real{16,32,64}`, `SplitReal{32,64}`
 
 __Covered under a different category:__
- * `[Split]Index{32,64}`
+ * `[Split]Index{32,64}`: with [`std::string`](../string)
  * `Switch`
 
 __Missing:__

types/string/README.md

@@ -0,0 +1,15 @@
+# `std::string`
+
+## Fields
+
+ * `[Split]Index{32,64}`
+
+with the corresponding column type for the first (principal) column.
+
+## Entries
+
+1. One character strings, with ascending values
+2. Empty strings
+3. Unicode characters as UTF-8
+4. Increasing length of strings:
+   one character in the first field, two characters in the second field, etc.

types/string/read.C

@@ -0,0 +1,51 @@
+#include <ROOT/REntry.hxx>
+#include <ROOT/RNTupleReader.hxx>
+
+using ROOT::Experimental::REntry;
+using ROOT::Experimental::RNTupleReader;
+
+#include <fstream>
+#include <ostream>
+#include <string>
+#include <string_view>
+
+static void PrintStringValue(const REntry &entry, std::string_view name,
+                             std::ostream &os, bool last = false) {
+  std::string &value = *entry.GetPtr<std::string>(name);
+  os << "    \"" << name << "\": \"" << value << "\"";
+  if (!last) {
+    os << ",";
+  }
+  os << "\n";
+}
+
+void read(std::string_view input = "types.string.root",
+          std::string_view output = "types.string.json") {
+  std::ofstream os(std::string{output});
+  os << "[\n";
+
+  auto reader = RNTupleReader::Open("ntpl", input);
+  auto &entry = reader->GetModel().GetDefaultEntry();
+  bool first = true;
+  for (auto index : *reader) {
+    reader->LoadEntry(index);
+
+    if (first) {
+      first = false;
+    } else {
+      os << ",\n";
+    }
+    os << "  {\n";
+
+    PrintStringValue(entry, "Index32", os);
+    PrintStringValue(entry, "Index64", os);
+    PrintStringValue(entry, "SplitIndex32", os);
+    PrintStringValue(entry, "SplitIndex64", os, /*last=*/true);
+
+    os << "  }";
+    // Newline is intentionally missing, may need to print a comma before the
+    // next entry.
+  }
+  os << "\n";
+  os << "]\n";
+}

types/string/write.C

@@ -0,0 +1,71 @@
+#include <ROOT/RField.hxx>
+#include <ROOT/RNTupleModel.hxx>
+#include <ROOT/RNTupleUtil.hxx>
+#include <ROOT/RNTupleWriteOptions.hxx>
+#include <ROOT/RNTupleWriter.hxx>
+
+using ROOT::Experimental::EColumnType;
+using ROOT::Experimental::RField;
+using ROOT::Experimental::RNTupleModel;
+using ROOT::Experimental::RNTupleWriteOptions;
+using ROOT::Experimental::RNTupleWriter;
+
+#include <memory>
+#include <string>
+#include <string_view>
+
+static std::shared_ptr<std::string> MakeStringField(RNTupleModel &model,
+                                                    std::string_view name,
+                                                    EColumnType indexType) {
+  auto field = std::make_unique<RField<std::string>>(name);
+  field->SetColumnRepresentatives({{indexType, EColumnType::kChar}});
+  model.AddField(std::move(field));
+  return model.GetDefaultEntry().GetPtr<std::string>(name);
+}
+
+void write(std::string_view filename = "types.string.root") {
+  auto model = RNTupleModel::Create();
+
+  // Non-split index encoding
+  auto Index32 = MakeStringField(*model, "Index32", EColumnType::kIndex32);
+  auto Index64 = MakeStringField(*model, "Index64", EColumnType::kIndex64);
+
+  // Split index encoding
+  auto SplitIndex32 =
+      MakeStringField(*model, "SplitIndex32", EColumnType::kSplitIndex32);
+  auto SplitIndex64 =
+      MakeStringField(*model, "SplitIndex64", EColumnType::kSplitIndex64);
+
+  RNTupleWriteOptions options;
+  options.SetCompression(0);
+  auto writer =
+      RNTupleWriter::Recreate(std::move(model), "ntpl", filename, options);
+
+  // First entry: one character strings, with ascending values
+  *Index32 = "a";
+  *Index64 = "b";
+  *SplitIndex32 = "c";
+  *SplitIndex64 = "d";
+  writer->Fill();
+
+  // Second entry: empty strings
+  *Index32 = "";
+  *Index64 = "";
+  *SplitIndex32 = "";
+  *SplitIndex64 = "";
+  writer->Fill();
+
+  // Third entry: Unicode characters as UTF-8
+  *Index32 = "🐣";
+  *Index64 = "🐣";
+  *SplitIndex32 = "🐣";
+  *SplitIndex64 = "🐣";
+  writer->Fill();
+
+  // Fourth entry: increasing length of strings
+  *Index32 = "a";
+  *Index64 = "bc";
+  *SplitIndex32 = "def";
+  *SplitIndex64 = "ghij";
+  writer->Fill();
+}