GH-50859: [C++][Parquet] Move JsonWriter to simdjson utilities

cpp/cmake_modules/DefineOptions.cmake

@@ -599,8 +599,7 @@ takes precedence over ccache if a storage backend is configured" ON)
                 "Build support for encryption. Fail if OpenSSL is not found"
                 OFF
                 DEPENDS
-                ARROW_FILESYSTEM
-                ARROW_JSON)
+                ARROW_FILESYSTEM)
 
   #----------------------------------------------------------------------
   set_option_category("Gandiva")

cpp/cmake_modules/ThirdpartyToolchain.cmake

@@ -383,7 +383,6 @@ if(ARROW_WITH_OPENTELEMETRY)
 endif()
 
 if(ARROW_PARQUET)
-  set(ARROW_WITH_RAPIDJSON ON)
   set(ARROW_WITH_SIMDJSON ON)
   set(ARROW_WITH_THRIFT ON)
 endif()
@@ -411,11 +410,11 @@ if(ARROW_AZURE)
   set(ARROW_WITH_AZURE_SDK ON)
 endif()
 
-if(ARROW_JSON OR ARROW_FLIGHT_SQL_ODBC)
+if(ARROW_JSON)
   set(ARROW_WITH_RAPIDJSON ON)
 endif()
 
-if(ARROW_JSON)
+if(ARROW_JSON OR ARROW_FLIGHT_SQL_ODBC)
   set(ARROW_WITH_SIMDJSON ON)
 endif()
 

cpp/meson.build

@@ -104,6 +104,7 @@ needs_testing = (get_option('testing').enabled()
     or needs_integration
 )
 needs_json = get_option('json').enabled() or needs_testing
+needs_simdjson = needs_json or needs_parquet
 needs_brotli = get_option('brotli').enabled() or needs_fuzzing
 needs_bz2 = get_option('bz2').enabled()
 needs_lz4 = get_option('lz4').enabled()

cpp/src/arrow/CMakeLists.txt

@@ -644,6 +644,9 @@ endif()
 if(ARROW_WITH_OPENTELEMETRY)
   list(APPEND ARROW_UTIL_SRCS util/tracing_internal.cc)
 endif()
+if(ARROW_WITH_SIMDJSON)
+  list(APPEND ARROW_UTIL_SRCS util/simdjson_internal.cc)
+endif()
 if(ARROW_WITH_SNAPPY)
   list(APPEND ARROW_UTIL_SRCS util/compression_snappy.cc)
 endif()
@@ -664,6 +667,12 @@ foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
   target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE ${ARROW_XSIMD})
 endforeach()
 
+if(ARROW_WITH_SIMDJSON)
+  foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
+    target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE arrow::simdjson)
+  endforeach()
+endif()
+
 if(ARROW_USE_BOOST)
   foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
     target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE Boost::headers)
@@ -1088,8 +1097,6 @@ if(ARROW_JSON)
                            json/chunker.cc
                            json/converter.cc
                            json/from_string.cc
-                           json/json_writer_internal.cc
-                           json/object_parser.cc
                            json/parser.cc
                            json/reader.cc)
   foreach(ARROW_JSON_TARGET ${ARROW_JSON_TARGETS})

cpp/src/arrow/dataset/file_json_test.cc

@@ -20,11 +20,11 @@
 #include "arrow/dataset/plan.h"
 #include "arrow/dataset/test_util_internal.h"
 #include "arrow/filesystem/mockfs.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/json/parser.h"
 #include "arrow/testing/gtest_util.h"
 #include "arrow/testing/util.h"
 #include "arrow/util/logging_internal.h"
+#include "arrow/util/simdjson_internal.h"
 
 namespace arrow {
 
@@ -104,11 +104,11 @@ struct WriteVisitor {
     return Status::OK();
   }
 
-  json::JsonWriter& writer_;
+  ::arrow::internal::JsonWriter& writer_;
   const Scalar& scalar_;
 };
 
-Status WriteJson(const StructScalar& scalar, json::JsonWriter* writer) {
+Status WriteJson(const StructScalar& scalar, ::arrow::internal::JsonWriter* writer) {
   WriteVisitor visitor{*writer, scalar};
   return VisitWriteableTypeId(Type::STRUCT, &visitor);
 }
@@ -122,7 +122,7 @@ class JsonFormatHelper {
     std::stringstream ss;
 
     for (const auto& scalar : scalars) {
-      json::JsonWriter writer;
+      ::arrow::internal::JsonWriter writer;
       RETURN_NOT_OK(WriteJson(*scalar, &writer));
 
       ARROW_ASSIGN_OR_RAISE(auto json, writer.GetString());

cpp/src/arrow/extension/fixed_shape_tensor.cc

@@ -27,15 +27,14 @@
 
 #include "arrow/array/array_nested.h"
 #include "arrow/array/array_primitive.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/tensor.h"
 #include "arrow/util/logging_internal.h"
 #include "arrow/util/print_internal.h"
 #include "arrow/util/simdjson_internal.h"
 #include "arrow/util/sort_internal.h"
 #include "arrow/util/string.h"
 
-using ::arrow::json::JsonWriter;
+using ::arrow::internal::JsonWriter;
 
 namespace arrow::extension {
 

cpp/src/arrow/extension/opaque.cc

@@ -19,13 +19,12 @@
 
 #include <sstream>
 
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/util/logging_internal.h"
 #include "arrow/util/simdjson_internal.h"
 
 #include <simdjson.h>
 
-using ::arrow::json::JsonWriter;
+using ::arrow::internal::JsonWriter;
 
 namespace arrow::extension {
 

cpp/src/arrow/extension/variable_shape_tensor.cc

@@ -23,7 +23,6 @@
 #include "arrow/extension/variable_shape_tensor.h"
 
 #include "arrow/array/array_primitive.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/scalar.h"
 #include "arrow/tensor.h"
 #include "arrow/util/logging_internal.h"
@@ -32,7 +31,7 @@
 #include "arrow/util/sort_internal.h"
 #include "arrow/util/string.h"
 
-using ::arrow::json::JsonWriter;
+using ::arrow::internal::JsonWriter;
 
 namespace arrow::extension {
 

cpp/src/arrow/flight/sql/odbc/odbc_impl/json_converter.cc

@@ -20,8 +20,8 @@
 #include <boost/beast/core/detail/base64.hpp>
 #include "arrow/builder.h"
 #include "arrow/flight/sql/odbc/odbc_impl/util.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/scalar.h"
+#include "arrow/util/simdjson_internal.h"
 #include "arrow/visitor.h"
 
 using boost::beast::detail::base64::encode;
@@ -30,7 +30,7 @@ namespace base64 = boost::beast::detail::base64;
 
 namespace arrow::flight::sql::odbc {
 
-using ::arrow::json::JsonWriter;
+using ::arrow::internal::JsonWriter;
 using util::ThrowIfNotOK;
 
 namespace {

cpp/src/arrow/integration/json_integration.cc

@@ -29,7 +29,6 @@
 #include "arrow/integration/json_internal.h"
 #include "arrow/io/file.h"
 #include "arrow/ipc/dictionary.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/record_batch.h"
 #include "arrow/result.h"
 #include "arrow/status.h"
@@ -40,7 +39,7 @@
 using arrow::ipc::DictionaryFieldMapper;
 using arrow::ipc::DictionaryMemo;
 
-using JsonWriter = arrow::json::JsonWriter;
+using JsonWriter = arrow::internal::JsonWriter;
 
 namespace arrow::internal::integration {
 

cpp/src/arrow/integration/json_integration_test.cc

@@ -38,7 +38,6 @@
 #include "arrow/ipc/reader.h"
 #include "arrow/ipc/test_common.h"
 #include "arrow/ipc/writer.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/pretty_print.h"
 #include "arrow/status.h"
 #include "arrow/testing/builder.h"
@@ -725,7 +724,7 @@ static const char* json_example6 = R"example(
 )example";
 
 void TestSchemaRoundTrip(const std::shared_ptr<Schema>& schema) {
-  arrow::json::JsonWriter writer;
+  arrow::internal::JsonWriter writer;
 
   DictionaryFieldMapper mapper(*schema);
 
@@ -749,7 +748,7 @@ void TestSchemaRoundTrip(const std::shared_ptr<Schema>& schema) {
 void TestArrayRoundTrip(const Array& array) {
   static std::string name = "dummy";
 
-  arrow::json::JsonWriter writer;
+  arrow::internal::JsonWriter writer;
 
   ASSERT_OK(json::WriteArray(name, array, &writer));
 

cpp/src/arrow/integration/json_internal.cc

@@ -36,7 +36,6 @@
 #include "arrow/array/builder_time.h"
 #include "arrow/extension_type.h"
 #include "arrow/ipc/dictionary.h"
-#include "arrow/json/json_writer_internal.h"
 #include "arrow/record_batch.h"
 #include "arrow/result.h"
 #include "arrow/scalar.h"
@@ -66,7 +65,7 @@ using arrow::ipc::DictionaryFieldMapper;
 using arrow::ipc::DictionaryMemo;
 using arrow::ipc::internal::FieldPosition;
 
-using JsonWriter = arrow::json::JsonWriter;
+using JsonWriter = arrow::internal::JsonWriter;
 
 namespace arrow::internal::integration::json {
 

cpp/src/arrow/integration/json_internal.h

@@ -32,26 +32,25 @@ using JsonValue = simdjson::dom::element;
 using JsonObject = simdjson::dom::object;
 using JsonArray = simdjson::dom::array;
 
-namespace arrow::json {
+namespace arrow::internal {
 class JsonWriter;
-}  // namespace arrow::json
+}  // namespace arrow::internal
 
 namespace arrow::internal::integration::json {
 
 /// \brief Append integration test Schema format to JSON writer
 ARROW_EXPORT
 Status WriteSchema(const Schema& schema, const ipc::DictionaryFieldMapper& mapper,
-                   arrow::json::JsonWriter*);
+                   JsonWriter*);
 
 ARROW_EXPORT
-Status WriteDictionary(int64_t id, const std::shared_ptr<Array>& dictionary,
-                       arrow::json::JsonWriter*);
+Status WriteDictionary(int64_t id, const std::shared_ptr<Array>& dictionary, JsonWriter*);
 
 ARROW_EXPORT
-Status WriteRecordBatch(const RecordBatch& batch, arrow::json::JsonWriter*);
+Status WriteRecordBatch(const RecordBatch& batch, JsonWriter*);
 
 ARROW_EXPORT
-Status WriteArray(const std::string& name, const Array& array, arrow::json::JsonWriter*);
+Status WriteArray(const std::string& name, const Array& array, JsonWriter*);
 
 ARROW_EXPORT
 Result<std::shared_ptr<Schema>> ReadSchema(const JsonValue& json_obj, MemoryPool* pool,

cpp/src/arrow/json/CMakeLists.txt

@@ -21,8 +21,6 @@ add_arrow_test(test
                chunker_test.cc
                converter_test.cc
                from_string_test.cc
-               json_writer_internal_test.cc
-               object_parser_test.cc
                parser_test.cc
                reader_test.cc
                PREFIX