Apply primitive row operators into hash join

cpp/CMakeLists.txt

@@ -781,6 +781,7 @@ add_library(
   src/structs/structs_column_factories.cu
   src/structs/structs_column_view.cpp
   src/structs/utilities.cpp
+  src/table/primitive_row_operators.cu
   src/table/row_operators.cu
   src/table/table.cpp
   src/table/table_device_view.cu

cpp/include/cudf/table/primitive_row_operators.cuh

@@ -30,6 +30,18 @@
 #include <cuda/std/type_traits>
 
 namespace CUDF_EXPORT cudf {
+
+/**
+ * @brief Checks if a table is compatible with primitive row operations
+ *
+ * A table is compatible with primitive row operations if it contains exactly one column
+ * and that column contains only numeric data types.
+ *
+ * @param table The table to check for compatibility
+ * @return Boolean indicating if the table is compatible with primitive row operations
+ */
+bool is_primitive_row_op_compatible(cudf::table_view const& table);
+
 namespace row::primitive {
 
 /**
@@ -177,7 +189,7 @@ class element_hasher {
  *
  * @tparam Hash Hash functor to use for hashing elements.
  */
-template <template <typename> class Hash>
+template <template <typename> class Hash = cudf::hashing::detail::default_hash>
 class row_hasher {
  public:
   row_hasher() = delete;
@@ -189,7 +201,9 @@ class row_hasher {
    * @param t A table_device_view to hash
    * @param seed A seed value to use for hashing
    */
-  row_hasher(cudf::nullate::DYNAMIC const& has_nulls, table_device_view t, hash_value_type seed)
+  row_hasher(cudf::nullate::DYNAMIC const& has_nulls,
+             table_device_view t,
+             hash_value_type seed = DEFAULT_HASH_SEED)
     : _has_nulls{has_nulls}, _table{t}, _seed{seed}
   {
   }
@@ -203,7 +217,7 @@ class row_hasher {
    */
   row_hasher(cudf::nullate::DYNAMIC const& has_nulls,
              std::shared_ptr<cudf::experimental::row::equality::preprocessed_table> t,
-             hash_value_type seed)
+             hash_value_type seed = DEFAULT_HASH_SEED)
     : _has_nulls{has_nulls}, _table{*t}, _seed{seed}
   {
   }

cpp/src/join/distinct_hash_join.cu

@@ -53,7 +53,7 @@ bool constexpr has_nulls = true;  ///< Always has nulls
  */
 template <typename T>
 class primitive_keys_fn {
-  using hasher = cudf::row::primitive::row_hasher<cudf::hashing::detail::default_hash>;
+  using hasher = cudf::row::primitive::row_hasher<>;
 
  public:
   CUDF_HOST_DEVICE constexpr primitive_keys_fn(hasher const& hash) : _hash{hash} {}
@@ -207,9 +207,9 @@ distinct_hash_join::distinct_hash_join(cudf::table_view const& build,
     }
   };
 
-  if (_build.num_columns() == 1 and cudf::is_numeric(_build.column(0).type())) {
-    auto const d_hasher = cudf::row::primitive::row_hasher<cudf::hashing::detail::default_hash>{
-      nullate::DYNAMIC{has_nulls}, this->_preprocessed_build, DEFAULT_HASH_SEED};
+  if (cudf::is_primitive_row_op_compatible(_build)) {
+    auto const d_hasher =
+      cudf::row::primitive::row_hasher{nullate::DYNAMIC{has_nulls}, this->_preprocessed_build};
 
     auto const iter = cudf::detail::make_counting_transform_iterator(
       0, primitive_keys_fn<rhs_index_type>{d_hasher});
@@ -253,9 +253,9 @@ distinct_hash_join::inner_join(cudf::table_view const& probe,
 
   auto preprocessed_probe =
     cudf::experimental::row::equality::preprocessed_table::create(probe, stream);
-  if (_build.num_columns() == 1 and cudf::is_numeric(_build.column(0).type())) {
-    auto const d_hasher = cudf::row::primitive::row_hasher<cudf::hashing::detail::default_hash>{
-      nullate::DYNAMIC{has_nulls}, preprocessed_probe, DEFAULT_HASH_SEED};
+  if (cudf::is_primitive_row_op_compatible(_build)) {
+    auto const d_hasher =
+      cudf::row::primitive::row_hasher{nullate::DYNAMIC{has_nulls}, preprocessed_probe};
     auto const d_equal = cudf::row::primitive::row_equality_comparator{
       nullate::DYNAMIC{has_nulls}, preprocessed_probe, _preprocessed_build, _nulls_equal};
     auto const iter = cudf::detail::make_counting_transform_iterator(
@@ -349,9 +349,9 @@ std::unique_ptr<rmm::device_uvector<size_type>> distinct_hash_join::left_join(
   auto preprocessed_probe =
     cudf::experimental::row::equality::preprocessed_table::create(probe, stream);
 
-  if (_build.num_columns() == 1 and cudf::is_numeric(_build.column(0).type())) {
-    auto const d_hasher = cudf::row::primitive::row_hasher<cudf::hashing::detail::default_hash>{
-      nullate::DYNAMIC{has_nulls}, preprocessed_probe, DEFAULT_HASH_SEED};
+  if (cudf::is_primitive_row_op_compatible(_build)) {
+    auto const d_hasher =
+      cudf::row::primitive::row_hasher{nullate::DYNAMIC{has_nulls}, preprocessed_probe};
     auto const d_equal = cudf::row::primitive::row_equality_comparator{
       nullate::DYNAMIC{has_nulls}, preprocessed_probe, _preprocessed_build, _nulls_equal};