Expose stream parameter to get_json_object API (#14297)

Add stream parameter to public APIs `cudf::get_json_object()`
Also removed the API from the `strings` namespace since it does not fit with the other strings library functions.
This resulted in updating the source file locations as well.

Reference #13744

Authors:
  - David Wendt (https://github.com/davidwendt)

Approvers:
  - Karthikeyan (https://github.com/karthikeyann)
  - GALI PREM SAGAR (https://github.com/galipremsagar)
  - Bradley Dice (https://github.com/bdice)
  - Nghia Truong (https://github.com/ttnghia)

URL: #14297

cpp/CMakeLists.txt

@@ -440,6 +440,7 @@ add_library(
   src/join/mixed_join_size_kernel_nulls.cu
   src/join/mixed_join_size_kernels_semi.cu
   src/join/semi_join.cu
+  src/json/json_path.cu
   src/lists/contains.cu
   src/lists/combine/concatenate_list_elements.cu
   src/lists/combine/concatenate_rows.cu
@@ -571,7 +572,6 @@ add_library(
   src/strings/filter_chars.cu
   src/strings/like.cu
   src/strings/padding.cu
-  src/strings/json/json_path.cu
   src/strings/regex/regcomp.cpp
   src/strings/regex/regexec.cpp
   src/strings/regex/regex_program.cpp

cpp/benchmarks/CMakeLists.txt

@@ -320,7 +320,7 @@ ConfigureNVBench(
 
 # ##################################################################################################
 # * json benchmark -------------------------------------------------------------------
-ConfigureBench(JSON_BENCH string/json.cu)
+ConfigureBench(JSON_BENCH json/json.cu)
 ConfigureNVBench(FST_NVBENCH io/fst.cu)
 ConfigureNVBench(JSON_READER_NVBENCH io/json/nested_json.cpp io/json/json_reader_input.cpp)
 ConfigureNVBench(JSON_WRITER_NVBENCH io/json/json_writer.cpp)

cpp/benchmarks/string/json.cu → cpp/benchmarks/json/json.cu

@@ -21,9 +21,9 @@
 #include <cudf_test/column_wrapper.hpp>
 
 #include <cudf/column/column_factories.hpp>
+#include <cudf/json/json.hpp>
 #include <cudf/strings/detail/strings_children.cuh>
 #include <cudf/strings/detail/utilities.cuh>
-#include <cudf/strings/json.hpp>
 #include <cudf/strings/string_view.hpp>
 #include <cudf/strings/strings_column_view.hpp>
 #include <cudf/types.hpp>
@@ -196,7 +196,7 @@ void BM_case(benchmark::State& state, std::string query_arg)
 
   for (auto _ : state) {
     cuda_event_timer raii(state, true);
-    auto result = cudf::strings::get_json_object(scv, json_path);
+    auto result = cudf::get_json_object(scv, json_path);
     CUDF_CUDA_TRY(cudaStreamSynchronize(0));
   }
 

cpp/include/cudf/strings/json.hpp → cpp/include/cudf/json/json.hpp

@@ -16,16 +16,16 @@
 #pragma once
 
 #include <cudf/strings/strings_column_view.hpp>
+#include <cudf/utilities/default_stream.hpp>
 
 #include <rmm/mr/device/per_device_resource.hpp>
 
 #include <thrust/optional.h>
 
 namespace cudf {
-namespace strings {
 
 /**
- * @addtogroup strings_json
+ * @addtogroup json_object
  * @{
  * @file
  */
@@ -155,20 +155,21 @@ class get_json_object_options {
  * https://tools.ietf.org/id/draft-goessner-dispatch-jsonpath-00.html
  * Implements only the operators: $ . [] *
  *
+ * @throw std::invalid_argument if provided an invalid operator or an empty name
+ *
  * @param col The input strings column. Each row must contain a valid json string
  * @param json_path The JSONPath string to be applied to each row
  * @param options Options for controlling the behavior of the function
- * @param mr Resource for allocating device memory.
+ * @param stream CUDA stream used for device memory operations and kernel launches
+ * @param mr Resource for allocating device memory
  * @return New strings column containing the retrieved json object strings
- *
- * @throw std::invalid_argument if provided an invalid operator or an empty name
  */
 std::unique_ptr<cudf::column> get_json_object(
   cudf::strings_column_view const& col,
   cudf::string_scalar const& json_path,
   get_json_object_options options     = get_json_object_options{},
+  rmm::cuda_stream_view stream        = cudf::get_default_stream(),
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
 /** @} */  // end of doxygen group
-}  // namespace strings
 }  // namespace cudf

cpp/include/doxygen_groups.h

@@ -130,7 +130,6 @@
  *   @defgroup strings_replace Replacing
  *   @defgroup strings_split Splitting
  *   @defgroup strings_extract Extracting
- *   @defgroup strings_json JSON
  *   @defgroup strings_regex Regex
  * @}
  * @defgroup dictionary_apis Dictionary
@@ -146,6 +145,10 @@
  *   @defgroup io_datasources Data Sources
  *   @defgroup io_datasinks Data Sinks
  * @}
+ * @defgroup json_apis JSON
+ * @{
+ *   @defgroup json_object JSON Path
+ * @}
  * @defgroup lists_apis Lists
  * @{
  *   @defgroup lists_combine Combining

cpp/src/strings/json/json_path.cu → cpp/src/json/json_path.cu

@@ -20,9 +20,9 @@
 #include <cudf/detail/null_mask.hpp>
 #include <cudf/detail/utilities/cuda.cuh>
 #include <cudf/detail/utilities/vector_factories.hpp>
+#include <cudf/json/json.hpp>
 #include <cudf/scalar/scalar.hpp>
 #include <cudf/strings/detail/utilities.hpp>
-#include <cudf/strings/json.hpp>
 #include <cudf/strings/string_view.cuh>
 #include <cudf/strings/strings_column_view.hpp>
 #include <cudf/types.hpp>
@@ -41,7 +41,6 @@
 #include <thrust/tuple.h>
 
 namespace cudf {
-namespace strings {
 namespace detail {
 
 namespace {
@@ -224,7 +223,9 @@ enum json_element_type { NONE, OBJECT, ARRAY, VALUE };
 class json_state : private parser {
  public:
   __device__ json_state() : parser() {}
-  __device__ json_state(char const* _input, int64_t _input_len, get_json_object_options _options)
+  __device__ json_state(char const* _input,
+                        int64_t _input_len,
+                        cudf::get_json_object_options _options)
     : parser(_input, _input_len),
 
       options(_options)
@@ -956,9 +957,6 @@ __launch_bounds__(block_size) __global__
   }
 }
 
-/**
- * @copydoc cudf::strings::detail::get_json_object
- */
 std::unique_ptr<cudf::column> get_json_object(cudf::strings_column_view const& col,
                                               cudf::string_scalar const& json_path,
                                               get_json_object_options options,
@@ -1011,7 +1009,7 @@ std::unique_ptr<cudf::column> get_json_object(cudf::strings_column_view const& c
     cudf::detail::get_value<size_type>(offsets_view, col.size(), stream);
 
   // allocate output string column
-  auto chars = create_chars_child_column(output_size, stream, mr);
+  auto chars = cudf::strings::detail::create_chars_child_column(output_size, stream, mr);
 
   // potential optimization : if we know that all outputs are valid, we could skip creating
   // the validity mask altogether
@@ -1041,17 +1039,14 @@ std::unique_ptr<cudf::column> get_json_object(cudf::strings_column_view const& c
 }  // namespace
 }  // namespace detail
 
-/**
- * @copydoc cudf::strings::get_json_object
- */
 std::unique_ptr<cudf::column> get_json_object(cudf::strings_column_view const& col,
                                               cudf::string_scalar const& json_path,
                                               get_json_object_options options,
+                                              rmm::cuda_stream_view stream,
                                               rmm::mr::device_memory_resource* mr)
 {
   CUDF_FUNC_RANGE();
-  return detail::get_json_object(col, json_path, options, cudf::get_default_stream(), mr);
+  return detail::get_json_object(col, json_path, options, stream, mr);
 }
 
-}  // namespace strings
 }  // namespace cudf

cpp/tests/CMakeLists.txt

@@ -523,7 +523,6 @@ ConfigureTest(
   strings/format_lists_tests.cpp
   strings/integers_tests.cpp
   strings/ipv4_tests.cpp
-  strings/json_tests.cpp
   strings/like_tests.cpp
   strings/pad_tests.cpp
   strings/repeat_strings_tests.cpp
@@ -537,6 +536,10 @@ ConfigureTest(
   strings/urls_tests.cpp
 )
 
+# ##################################################################################################
+# * json path test --------------------------------------------------------------------------------
+ConfigureTest(JSON_PATH_TEST json/json_tests.cpp)
+
 # ##################################################################################################
 # * structs test ----------------------------------------------------------------------------------
 ConfigureTest(STRUCTS_TEST structs/structs_column_tests.cpp structs/utilities_tests.cpp)