Update parser to store Binsparse JSON metadata in attribute instead of

dataset

Author: BenBrock

examples/Makefile

@@ -11,8 +11,9 @@ CXX ?= clang++
 BINSPARSE_DIR ?= ../include
 
 # HDF5 library location
-HDF5_CXXFLAGS ?= -I/opt/homebrew/Cellar/hdf5/1.14.1/include
-HDF5_LIBRARY_FLAGS ?= -L/opt/homebrew/Cellar/hdf5/1.14.1/lib -lhdf5_hl_cpp -lhdf5_cpp -lhdf5_hl -lhdf5 
+HDF5_DIR ?= /opt/homebrew/Cellar/hdf5/1.14.2
+HDF5_CXXFLAGS ?= -I$(HDF5_DIR)/include
+HDF5_LIBRARY_FLAGS ?= -L$(HDF5_DIR)/lib -lhdf5_hl_cpp -lhdf5_cpp -lhdf5_hl -lhdf5 
 
 # = = = = = = = = = = = = = = = = = = = = = = = = = = = #
 

include/binsparse/binsparse.hpp

@@ -11,8 +11,6 @@
 #include <binsparse/c_bindings/allocator_wrapper.hpp>
 #include <binsparse/matrix_market/matrix_market.hpp>
 
-#include <iostream>
-
 namespace binsparse {
 
 // CSR Format
@@ -46,7 +44,7 @@ void write_csr_matrix(std::string fname,
     j[v.key()] = v.value();
   }
 
-  hdf5_tools::write_dataset(f, "metadata", j.dump(2));
+  hdf5_tools::set_attribute(f, "binsparse", j.dump(2));
 
   f.close();
 }
@@ -60,10 +58,12 @@ csr_matrix<T, I> read_csr_matrix(std::string fname, Allocator&& alloc) {
   using json = nlohmann::json;
   auto data = json::parse(metadata);
 
-  if (data["binsparse"]["format"] == "CSR") {
-    auto nrows = data["binsparse"]["shape"][0];
-    auto ncols = data["binsparse"]["shape"][1];
-    auto nnz = data["binsparse"]["nnz"];
+  auto binsparse_metadata = data["binsparse"];
+
+  if (binsparse_metadata["format"] == "CSR") {
+    auto nrows = binsparse_metadata["shape"][0];
+    auto ncols = binsparse_metadata["shape"][1];
+    auto nnz = binsparse_metadata["nnz"];
 
     typename std::allocator_traits<std::remove_cvref_t<Allocator>>
        :: template rebind_alloc<I> i_alloc(alloc);
@@ -114,7 +114,7 @@ void write_coo_matrix(std::string fname,
     j[v.key()] = v.value();
   }
 
-  hdf5_tools::write_dataset(f, "metadata", j.dump(2));
+  hdf5_tools::set_attribute(f, "binsparse", j.dump(2));
 
   f.close();
 }
@@ -128,10 +128,12 @@ coo_matrix<T, I> read_coo_matrix(std::string fname, Allocator&& alloc) {
   using json = nlohmann::json;
   auto data = json::parse(metadata);
 
-  if (data["binsparse"]["format"] == "COO") {
-    auto nrows = data["binsparse"]["shape"][0];
-    auto ncols = data["binsparse"]["shape"][1];
-    auto nnz = data["binsparse"]["nnz"];
+  auto binsparse_metadata = data["binsparse"];
+
+  if (binsparse_metadata["format"] == "COO") {
+    auto nrows = binsparse_metadata["shape"][0];
+    auto ncols = binsparse_metadata["shape"][1];
+    auto nnz = binsparse_metadata["nnz"];
 
     typename std::allocator_traits<std::remove_cvref_t<Allocator>>
        :: template rebind_alloc<I> i_alloc(alloc);
@@ -154,12 +156,14 @@ coo_matrix<T, I> read_coo_matrix(std::string fname) {
 inline auto inspect(std::string fname) {
   H5::H5File f(fname.c_str(), H5F_ACC_RDWR);
 
-  auto metadata = hdf5_tools::read_dataset<char>(f, "metadata");
+  auto metadata = hdf5_tools::get_attribute(f, "binsparse");
 
   using json = nlohmann::json;
   auto data = json::parse(metadata);
 
-  if (data["binsparse"]["version"] >= 0.1) {
+  auto binsparse_metadata = data["binsparse"];
+
+  if (binsparse_metadata["version"] >= 0.1) {
     return data;
   } else {
     assert(false);

include/binsparse/hdf5_tools.hpp

@@ -5,6 +5,8 @@
 #include <ranges>
 #include <H5Cpp.h>
 
+#include <iostream>
+
 namespace hdf5_tools {
 
 template <typename U>
@@ -119,8 +121,8 @@ inline H5::PredType get_type(H5::DataSet& dataset) {
   }
 }
 
-template <std::ranges::contiguous_range R>
-void write_dataset(H5::H5File& f, const std::string& label, R&& r) {
+template <typename H5GroupOrFile, std::ranges::contiguous_range R>
+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);
   H5::DataSpace dataspace(1, &size);
@@ -131,8 +133,32 @@ void write_dataset(H5::H5File& f, const std::string& label, R&& r) {
   dataspace.close();
 }
 
-template <typename T, typename Allocator>
-std::span<T> read_dataset(H5::H5File& f, const std::string& label, Allocator&& alloc) {
+inline std::string get_attribute(H5::H5Object& f, const std::string& key) {
+  auto attribute = f.openAttribute(key.c_str());
+
+  std::size_t size = attribute.getStorageSize();
+
+  auto type = attribute.getDataType();
+  std::string attribute_string(" ", size);
+
+  attribute.read(type, attribute_string);
+
+  return attribute_string;
+}
+
+inline void set_attribute(H5::H5Object& f, const std::string& key, const std::string& value) {
+  hsize_t size = value.size();
+  H5::DataSpace dataspace(1, &size);
+
+  f.createAttribute(key.c_str(), H5::PredType::NATIVE_CHAR, dataspace);
+
+  auto attribute = f.openAttribute(key.c_str());
+
+  attribute.write(H5::PredType::NATIVE_CHAR, value.data());
+}
+
+template <typename T, typename Allocator, typename H5GroupOrFile>
+std::span<T> read_dataset(H5GroupOrFile& f, const std::string& label, Allocator&& alloc) {
   H5::DataSet dataset = f.openDataSet(label.c_str());
 
   H5::DataSpace space = dataset.getSpace();
@@ -148,12 +174,13 @@ std::span<T> read_dataset(H5::H5File& f, const std::string& label, Allocator&& a
   return std::span<T>(data, dims);
 }
 
-template <typename T>
-std::span<T> read_dataset(H5::H5File& f, const std::string& label) {
+template <typename T, typename H5GroupOrFile>
+std::span<T> read_dataset(H5GroupOrFile& f, const std::string& label) {
   return read_dataset<T>(f, label, std::allocator<T>{});
 }
 
-inline H5::PredType dataset_type(H5::H5File& f, const std::string& label) {
+template <typename H5GroupOrFile>
+inline H5::PredType dataset_type(H5GroupOrFile& f, const std::string& label) {
   H5::DataSet dataset = f.openDataSet(label.c_str());
   auto type = get_type(dataset);
   dataset.close();