Implement get_uncompressed_data_size_in_bytes()

include/cucascade/data/common.hpp

@@ -89,6 +89,17 @@ class idata_representation {
    */
   virtual std::size_t get_size_in_bytes() const = 0;
 
+  /**
+   * @brief Get the logical (uncompressed) data size in bytes
+   *
+   * For representations that store data in a compressed format (e.g. Parquet), this returns
+   * the uncompressed data size. For uncompressed representations, this returns the same value
+   * as get_size_in_bytes().
+   *
+   * @return std::size_t The logical data size in bytes
+   */
+  virtual std::size_t get_logical_data_size_in_bytes() const = 0;
+
   /**
    * @brief Create a deep copy of this data representation.
    *

include/cucascade/data/cpu_data_representation.hpp

@@ -56,6 +56,11 @@ class host_data_representation : public idata_representation {
    */
   std::size_t get_size_in_bytes() const override;
 
+  /**
+   * @copydoc idata_representation::get_logical_data_size_in_bytes
+   */
+  std::size_t get_logical_data_size_in_bytes() const override;
+
   /**
    * @brief Create a deep copy of this representation in the same memory space.
    *
@@ -101,6 +106,11 @@ class host_data_packed_representation : public idata_representation {
    */
   std::size_t get_size_in_bytes() const override;
 
+  /**
+   * @copydoc idata_representation::get_logical_data_size_in_bytes
+   */
+  std::size_t get_logical_data_size_in_bytes() const override;
+
   /**
    * @brief Create a deep copy of this host table representation.
    *

include/cucascade/data/gpu_data_representation.hpp

@@ -56,6 +56,11 @@ class gpu_table_representation : public idata_representation {
    */
   std::size_t get_size_in_bytes() const override;
 
+  /**
+   * @copydoc idata_representation::get_logical_data_size_in_bytes
+   */
+  std::size_t get_logical_data_size_in_bytes() const override;
+
   /**
    * @brief Create a deep copy of this GPU table representation.
    *

src/data/cpu_data_representation.cpp

@@ -34,6 +34,11 @@ host_data_representation::host_data_representation(
 
 std::size_t host_data_representation::get_size_in_bytes() const { return _host_table->data_size; }
 
+std::size_t host_data_representation::get_logical_data_size_in_bytes() const
+{
+  return get_size_in_bytes();
+}
+
 const std::unique_ptr<memory::host_table_allocation>& host_data_representation::get_host_table()
   const
 {
@@ -87,6 +92,11 @@ std::size_t host_data_packed_representation::get_size_in_bytes() const
   return _host_table->data_size;
 }
 
+std::size_t host_data_packed_representation::get_logical_data_size_in_bytes() const
+{
+  return get_size_in_bytes();
+}
+
 const std::unique_ptr<cucascade::memory::host_table_packed_allocation>&
 host_data_packed_representation::get_host_table() const
 {

src/data/gpu_data_representation.cpp

@@ -30,6 +30,11 @@ gpu_table_representation::gpu_table_representation(std::unique_ptr<cudf::table>
 
 std::size_t gpu_table_representation::get_size_in_bytes() const { return _table->alloc_size(); }
 
+std::size_t gpu_table_representation::get_logical_data_size_in_bytes() const
+{
+  return get_size_in_bytes();
+}
+
 const cudf::table& gpu_table_representation::get_table() const { return *_table; }
 
 std::unique_ptr<cudf::table> gpu_table_representation::release_table() { return std::move(_table); }

test/utils/mock_test_utils.hpp

@@ -114,6 +114,8 @@ class mock_data_representation : private mock_memory_space_holder, public idata_
 
   std::size_t get_size_in_bytes() const override { return _size; }
 
+  std::size_t get_logical_data_size_in_bytes() const override { return _size; }
+
   std::unique_ptr<idata_representation> clone(
     [[maybe_unused]] rmm::cuda_stream_view stream) override
   {