Change how strings are stored, use UTF-8 encoding

Author: BenBrock

examples/text2hdf5.cpp

@@ -0,0 +1,35 @@
+#include <binsparse/binsparse.hpp>
+#include <complex>
+#include <concepts>
+#include <iostream>
+#include <sstream>
+
+int main(int argc, char** argv) {
+
+  if (argc < 4) {
+    std::cout << "usage: ./write_text_file [input_file.txt] "
+                 "[output_file.hdf5] [HDF5 Dataset Name]\n";
+    return 1;
+  }
+
+  std::string input_file(argv[1]);
+  std::string output_file(argv[2]);
+  std::string dataset_name(argv[3]);
+
+  std::cout << "Read in text file " << input_file << " write to dataset "
+            << dataset_name << " in file " << output_file << std::endl;
+
+  std::ifstream f(input_file);
+  std::stringstream ss;
+  ss << f.rdbuf();
+
+  std::string file_contents = ss.str();
+
+  {
+    H5::H5File f(output_file.c_str(), H5F_ACC_RDWR);
+
+    hdf5_tools::write_dataset(f, dataset_name, file_contents);
+  }
+
+  return 0;
+}

include/binsparse/hdf5_tools.hpp

@@ -3,6 +3,7 @@
 #include <H5Cpp.h>
 #include <cassert>
 #include <ranges>
+#include <type_traits>
 #include <vector>
 
 #include <iostream>
@@ -120,6 +121,7 @@ inline H5::PredType get_type(H5::DataSet& dataset) {
 }
 
 template <typename H5GroupOrFile, std::ranges::contiguous_range R>
+  requires(!std::is_same_v<std::remove_cvref_t<R>, std::string>)
 void write_dataset(H5GroupOrFile& f, const std::string& label, R&& r) {
   using T = std::ranges::range_value_t<R>;
   hsize_t size = std::ranges::size(r);
@@ -132,6 +134,21 @@ void write_dataset(H5GroupOrFile& f, const std::string& label, R&& r) {
   dataspace.close();
 }
 
+template <typename H5GroupOrFile, std::ranges::contiguous_range R>
+  requires(std::is_same_v<std::remove_cvref_t<R>, std::string>)
+void write_dataset(H5GroupOrFile& f, const std::string& label, R&& r) {
+  H5::StrType string_type(H5::PredType::C_S1, r.size());
+  H5Tset_cset(string_type, H5T_CSET_UTF8);
+  hsize_t size = r.size();
+  H5::DataSpace dataspace(1, &size);
+
+  auto dataset = f.createDataSet(label.c_str(), string_type, H5S_SCALAR);
+
+  dataset.write(r.c_str(), string_type);
+
+  dataset.close();
+}
+
 inline std::string get_attribute(H5::H5Object& f, const std::string& key) {
   auto attribute = f.openAttribute(key.c_str());
 
@@ -151,6 +168,7 @@ inline std::string get_attribute(H5::H5Object& f, const std::string& key) {
 inline void set_attribute(H5::H5Object& f, const std::string& key,
                           const std::string& value) {
   H5::StrType string_type(H5::PredType::C_S1, value.size());
+  H5Tset_cset(string_type, H5T_CSET_UTF8);
   hsize_t size = value.size();
   H5::DataSpace dataspace(1, &size);
 

include/binsparse/type_info.hpp

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <cassert>
+#include <functional>
 #include <type_traits>
 
 namespace binsparse {