Merge pull request rapidsai#3303 from rgsl888prabhu/2948_adding_copy_if

[REVIEW] Define and implement new stream compaction APIs `copy_if`, `drop_nulls`, `apply_boolean_mask`, `drop_duplicate` and `unique_count`.

CHANGELOG.md

@@ -133,6 +133,8 @@
 - PR #3380 Concatenate columns of strings
 - PR #3382 Add fill function for strings column
 - PR #3391 Move device_atomics_tests.cu files to legacy
+- PR #3303 Define and implement new stream compaction APIs `copy_if`, `drop_nulls`, 
+           `apply_boolean_mask`, `drop_duplicate` and `unique_count`.
 - PR #3387 Strings column gather function
 - PR #3389 Move quantiles.hpp + group_quantiles.hpp files to legacy
 - PR #3398 Move reshape.hpp files to legacy

conda/recipes/libcudf/meta.yaml

@@ -78,6 +78,7 @@ test:
     - test -f $PREFIX/include/cudf/legacy/replace.hpp
     - test -f $PREFIX/include/cudf/legacy/rolling.hpp
     - test -f $PREFIX/include/cudf/legacy/search.hpp
+    - test -f $PREFIX/include/cudf/stream_compaction.hpp
     - test -f $PREFIX/include/cudf/legacy/stream_compaction.hpp
     - test -f $PREFIX/include/cudf/transpose.hpp
     - test -f $PREFIX/include/cudf/legacy/transform.hpp

cpp/CMakeLists.txt

@@ -394,6 +394,9 @@ add_library(cudf
             src/transform/legacy/nans_to_nulls.cu
             src/transform/transform.cpp
             src/bitmask/legacy/bitmask_ops.cu
+            src/stream_compaction/apply_boolean_mask.cu
+            src/stream_compaction/drop_nulls.cu
+            src/stream_compaction/drop_duplicates.cu
             src/stream_compaction/legacy/apply_boolean_mask.cu
             src/stream_compaction/legacy/drop_nulls.cu
             src/stream_compaction/legacy/drop_duplicates.cu

cpp/include/cudf/copying.hpp

@@ -72,7 +72,7 @@ enum class  mask_allocation_policy {
  * @param[in] input Immutable view of input column to emulate
  * @return std::unique_ptr<column> An empty column of same type as `input`
  */
-std::unique_ptr<column> empty_like(column_view input);
+std::unique_ptr<column> empty_like(column_view const& input);
 
 /**
  * @brief Creates an uninitialized new column of the same size and type as the `input`.
@@ -83,7 +83,7 @@ std::unique_ptr<column> empty_like(column_view input);
  * @param[in] mr Optional, The resource to use for all allocations
  * @return std::unique_ptr<column> A column with sufficient uninitialized capacity to hold the same number of elements as `input` of the same type as `input.type()`
  */
-std::unique_ptr<column> allocate_like(column_view input,
+std::unique_ptr<column> allocate_like(column_view const& input,
                                       mask_allocation_policy mask_alloc = mask_allocation_policy::RETAIN,
                                       rmm::mr::device_memory_resource *mr =
                                           rmm::mr::get_default_resource());
@@ -98,7 +98,7 @@ std::unique_ptr<column> allocate_like(column_view input,
  * @param[in] mr Optional, The resource to use for all allocations
  * @return std::unique_ptr<column> A column with sufficient uninitialized capacity to hold the specified number of elements as `input` of the same type as `input.type()`
  */
-std::unique_ptr<column> allocate_like(column_view input, size_type size,
+std::unique_ptr<column> allocate_like(column_view const& input, size_type size,
                                       mask_allocation_policy mask_alloc = mask_allocation_policy::RETAIN,
                                       rmm::mr::device_memory_resource *mr =
                                           rmm::mr::get_default_resource());
@@ -112,7 +112,7 @@ std::unique_ptr<column> allocate_like(column_view input, size_type size,
  * @param[in] input_table Immutable view of input table to emulate
  * @return std::unique_ptr<table> A table of empty columns with the same types as the columns in `input_table`
  */
-std::unique_ptr<table> empty_like(table_view input_table);
+std::unique_ptr<table> empty_like(table_view const& input_table);
 
 /**
  * @brief Slices a `column_view` into a set of `column_view`s according to a set of indices.

cpp/include/cudf/detail/copy.hpp

@@ -70,7 +70,7 @@ ColumnView slice(ColumnView const& input,
  * @param[in] stream Optional CUDA stream on which to execute kernels
  * @return std::unique_ptr<column> An empty column of same type as `input`
  */
-std::unique_ptr<column> empty_like(column_view input, cudaStream_t stream = 0);
+std::unique_ptr<column> empty_like(column_view const& input, cudaStream_t stream = 0);
 
 /**
  * @brief Creates an uninitialized new column of the specified size and same type as the `input`.
@@ -83,7 +83,7 @@ std::unique_ptr<column> empty_like(column_view input, cudaStream_t stream = 0);
  * @param[in] stream Optional CUDA stream on which to execute kernels
  * @return std::unique_ptr<column> A column with sufficient uninitialized capacity to hold the specified number of elements as `input` of the same type as `input.type()`
  */
-std::unique_ptr<column> allocate_like(column_view input, size_type size,
+std::unique_ptr<column> allocate_like(column_view const& input, size_type size,
                                       mask_allocation_policy mask_alloc = 
                                           mask_allocation_policy::RETAIN,
                                       rmm::mr::device_memory_resource *mr =
@@ -100,7 +100,7 @@ std::unique_ptr<column> allocate_like(column_view input, size_type size,
  * @param[in] stream Optional CUDA stream on which to execute kernels
  * @return std::unique_ptr<table> A table of empty columns with the same types as the columns in `input_table`
  */
-std::unique_ptr<table> empty_like(table_view input_table, cudaStream_t stream = 0);
+std::unique_ptr<table> empty_like(table_view const& input_table, cudaStream_t stream = 0);
 
 /**
  * @brief   Returns a new column, where each element is selected from either @p lhs or