Read fixed size strings without padding characters

src/h5cpp/attribute/attribute.cpp

@@ -128,6 +128,52 @@ void Attribute::close()
 }
 
 
+// special overload for reading attributes into a std::string
+void Attribute::read(std::string &data, bool trim) const
+{
+  auto file_type = datatype();
+  if (file_type.get_class() == datatype::Class::String)
+  {
+    read(data, file_type, trim);
+  }
+  else
+  {
+    hdf5::datatype::DatatypeHolder mem_type_holder;
+    read(data, mem_type_holder.get<std::string>(), file_type, trim);
+  }
+}
+
+
+void Attribute::read(std::string &data, const datatype::Datatype &mem_type, bool trim) const
+{
+  datatype::Datatype file_type = datatype();
+  read(data, mem_type, file_type, trim);
+}
+
+
+void Attribute::read(std::string &data, const datatype::Datatype &mem_type, const datatype::Datatype &file_type, bool trim) const
+{
+  check_size(dataspace::create(data), dataspace(), "read");
+
+  if (file_type.get_class() == datatype::Class::String)
+  {
+    datatype::String string_type(file_type);
+
+    if (string_type.is_variable_length())
+    {
+      read_variable_length_string(data, mem_type);
+    }
+    else
+    {
+      read_fixed_length_string(data, mem_type, trim);
+    }
+  }
+  else
+  {
+    read_contiguous_data(data, mem_type);
+  }
+}
+
 
 } // namespace attribute
 } // namespace hdf5

src/h5cpp/attribute/attribute.hpp

@@ -202,13 +202,19 @@ class DLL_EXPORT Attribute
     template<typename T>
     void read(T &data,const datatype::Datatype &mem_type) const;
 
+    void read(std::string &data, bool trim = false) const;
+
+    void read(std::string &data, const datatype::Datatype &mem_type, bool trim = false) const;
+
   private:
     node::Link   parent_link_;
     ObjectHandle handle_;
 
     template<typename T>
     void read(T &data,const datatype::Datatype &mem_type, const datatype::Datatype &file_type) const;
 
+  void read(std::string &data, const datatype::Datatype &mem_type, const datatype::Datatype &file_type, bool trim = false) const;
+
     template<typename T>
     void write_fixed_length_string(const T &data,
                                    const datatype::Datatype &mem_type) const
@@ -273,7 +279,32 @@ class DLL_EXPORT Attribute
 
     }
 
-    template<typename T>
+    void read_fixed_length_string(std::string &data,
+                                  const datatype::Datatype &mem_type, bool trim = false) const
+    {
+      using Trait = FixedLengthStringTrait<std::string>;
+      using SpaceTrait = hdf5::dataspace::TypeTrait<std::string>;
+
+      auto buffer = Trait::BufferType::create(mem_type, SpaceTrait::create(data));
+
+      if (H5Aread(static_cast<hid_t>(handle_),
+                  static_cast<hid_t>(mem_type),
+                  buffer.data()) < 0)
+      {
+        error::Singleton::instance().throw_with_stack("Failure to read data from attribute!");
+      }
+
+      if (trim)
+      {
+        data = Trait::from_buffer_trimmed(buffer, mem_type, SpaceTrait::create(data));
+      }
+      else
+      {
+        data = Trait::from_buffer(buffer, mem_type, SpaceTrait::create(data));
+      }
+    }
+
+    template <typename T>
     void read_variable_length_string(T &data,
                                      const datatype::Datatype &mem_type) const
     {

src/h5cpp/core/fixed_length_string.hpp

@@ -25,6 +25,7 @@
 //
 #pragma once
 
+#include <algorithm>
 #include <vector>
 #include <string>
 #include <h5cpp/dataspace/dataspace.hpp>
@@ -120,7 +121,42 @@ struct FixedLengthStringTrait<std::string>
                                 const datatype::String &,
                                 const dataspace::Dataspace &)
     {
-      return DataType(buffer.begin(),buffer.end());
+      return DataType(buffer.begin(), buffer.end());
+    }
+
+    //!
+    //! @brief store data from buffer in target memory and trim any tailing null-terminating characters
+    //!
+    //! @param buffer reference to the IO buffer
+    //! @param memory_type the datatype describing how the string is stored in the file
+    //! @return new data instance
+    static DataType from_buffer_trimmed(const BufferType &buffer,
+                                        const datatype::String &memory_type,
+                                        const dataspace::Dataspace &)
+    {
+      const auto padding = memory_type.padding();
+      auto end_it = buffer.end();
+      switch (padding)
+      {
+      case datatype::StringPad::NullTerm:
+        end_it = std::find(buffer.begin(), buffer.end(), '\0');
+        break;
+      case datatype::StringPad::NullPad:
+        // get iterator to last padding \0 character
+        end_it = std::find_if_not(buffer.rbegin(), buffer.rend(), [](const BufferType::value_type &c)
+                                  { return c == '\0'; })
+                     .base();
+        break;
+      case datatype::StringPad::SpacePad:
+        // get iterator to last padding space character
+        end_it = std::find_if_not(buffer.rbegin(), buffer.rend(), [](const BufferType::value_type &c)
+                                  { return c == ' '; })
+                     .base();
+        break;
+      default:
+        break;
+      }
+      return DataType(buffer.begin(), end_it);
     }
 };
 

test/CMakeLists.txt

@@ -19,6 +19,7 @@ include_directories(${PROJECT_SOURCE_DIR}/src
 
 configure_file(h5py_test_data.h5 h5py_test_data.h5 COPYONLY)
 configure_file(h5py_test_boolattr.h5 h5py_test_boolattr.h5 COPYONLY)
+configure_file(h5py_test_string_attributes.h5 h5py_test_string_attributes.h5 COPYONLY)
 configure_file(pniio_test_boolattr.h5 pniio_test_boolattr.h5 COPYONLY)
 
 include_directories(${CMAKE_CURRENT_SOURCE_DIR})

test/core/CMakeLists.txt

@@ -1,5 +1,5 @@
 
-set(test_sources 
+set(test_sources
     ObjectHandleDefault.cpp
     object_handle_test.cpp
     file_object_handle_test.cpp
@@ -10,6 +10,7 @@ set(test_sources
     attribute_object_handle_test.cpp
     property_objects_handle_test.cpp
     error_objects_handle_test.cpp
+    fixed_length_string_test.cpp
     test_environment.cpp
     iteration_index_test.cpp
     iteration_order_test.cpp