Add struct utility functions. (#10776)

This PR adds some struct utility functions. This change is needed for the eventual support of structs in binary operations. See also: PR #9452.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - Vyas Ramasubramani (https://github.com/vyasr)
  - Jake Hemstad (https://github.com/jrhemstad)

URL: #10776

cpp/CMakeLists.txt

@@ -189,7 +189,6 @@ add_library(
   src/ast/expression_parser.cpp
   src/ast/expressions.cpp
   src/binaryop/binaryop.cpp
-  src/binaryop/compiled/binary_ops.cu
   src/binaryop/compiled/Add.cu
   src/binaryop/compiled/ATan2.cu
   src/binaryop/compiled/BitwiseAnd.cu
@@ -220,6 +219,7 @@ add_library(
   src/binaryop/compiled/ShiftRightUnsigned.cu
   src/binaryop/compiled/Sub.cu
   src/binaryop/compiled/TrueDiv.cu
+  src/binaryop/compiled/binary_ops.cu
   src/binaryop/compiled/util.cpp
   src/labeling/label_bins.cu
   src/bitmask/null_mask.cu

cpp/include/cudf/detail/structs/utilities.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2021, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2022, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -245,6 +245,20 @@ std::tuple<cudf::table_view, std::vector<rmm::device_buffer>> superimpose_parent
   table_view const& table,
   rmm::cuda_stream_view stream        = rmm::cuda_stream_default,
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+
+/**
+ * @brief Checks if a column or any of its children is a struct column with structs that are null.
+ *
+ * This function searches for structs that are null -- differentiating between structs that are null
+ * and structs containing null values. Null structs add a column to the result of the flatten column
+ * utility and necessitates column_nullability::FORCE when flattening the column for comparison
+ * operations.
+ *
+ * @param col Column to check for null structs
+ * @return A boolean indicating if the column is or contains a struct column that contains a null
+ * struct.
+ */
+bool contains_null_structs(column_view const& col);
 }  // namespace detail
 }  // namespace structs
 }  // namespace cudf

cpp/include/cudf/table/row_operators.cuh

@@ -74,7 +74,7 @@ __device__ weak_ordering compare_elements(Element lhs, Element rhs)
  * @brief A specialization for floating-point `Element` type relational comparison
  * to derive the order of the elements with respect to `lhs`.
  *
- * This Specialization handles `nan` in the following order:
+ * This specialization handles `nan` in the following order:
  * `[-Inf, -ve, 0, -0, +ve, +Inf, NaN, NaN, null] (for null_order::AFTER)`
  * `[null, -Inf, -ve, 0, -0, +ve, +Inf, NaN, NaN] (for null_order::BEFORE)`
  *

cpp/include/cudf/utilities/traits.hpp

@@ -699,6 +699,37 @@ constexpr inline bool is_nested(data_type type)
   return cudf::type_dispatcher(type, is_nested_impl{});
 }
 
+/**
+ * @brief Indicates whether `T` is a struct type.
+ *
+ * @param T The type to verify
+ * @return A boolean indicating if T is a struct type
+ */
+template <typename T>
+constexpr inline bool is_struct()
+{
+  return std::is_same_v<T, cudf::struct_view>;
+}
+
+struct is_struct_impl {
+  template <typename T>
+  constexpr bool operator()()
+  {
+    return is_struct<T>();
+  }
+};
+
+/**
+ * @brief Indicates whether `type` is a struct type.
+ *
+ * @param type The `data_type` to verify
+ * @return A boolean indicating if `type` is a struct type
+ */
+constexpr inline bool is_struct(data_type type)
+{
+  return cudf::type_dispatcher(type, is_struct_impl{});
+}
+
 template <typename FromType>
 struct is_bit_castable_to_impl {
   template <typename ToType, std::enable_if_t<is_compound<ToType>()>* = nullptr>

cpp/src/structs/utilities.cpp

@@ -441,6 +441,12 @@ std::tuple<cudf::table_view, std::vector<rmm::device_buffer>> superimpose_parent
   return {table_view{superimposed_columns}, std::move(superimposed_nullmasks)};
 }
 
+bool contains_null_structs(column_view const& col)
+{
+  return (is_struct(col) && col.has_nulls()) ||
+         std::any_of(col.child_begin(), col.child_end(), contains_null_structs);
+}
+
 }  // namespace detail
 }  // namespace structs
 }  // namespace cudf