extract our arrow kernels into library

36 files changed, 514 insertions, 192 deletions

diff --git a/ydb/core/formats/CMakeLists.txt b/ydb/core/formats/CMakeLists.txt
index 4d75a9e95f2..625c3779f09 100644
--- a/ydb/core/formats/CMakeLists.txt
+++ b/ydb/core/formats/CMakeLists.txt
@@ -20,6 +20,7 @@ target_link_libraries(ydb-core-formats PUBLIC
   yutil
   libs-apache-arrow
   ydb-core-scheme
+  ydb-library-arrow_kernels
   ydb-library-binary_json
   ydb-library-dynumber
   ydb-library-arrow_clickhouse
@@ -29,7 +30,6 @@ target_sources(ydb-core-formats PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/arrow_helpers.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/clickhouse_block.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/custom_registry.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/func_cast.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/merging_sorted_input_stream.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/program.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ssa_program_optimizer.cpp
diff --git a/ydb/core/formats/arrow_helpers.cpp b/ydb/core/formats/arrow_helpers.cpp
index 0dc67b9e028..2137751d9cf 100644
--- a/ydb/core/formats/arrow_helpers.cpp
+++ b/ydb/core/formats/arrow_helpers.cpp
@@ -1241,24 +1241,6 @@ bool TArrowToYdbConverter::Process(const arrow::RecordBatch& batch, TString& err
     return true;
 }
 
-std::shared_ptr<arrow::Array> NumVecToArray(const std::shared_ptr<arrow::DataType>& type,
-                                            const std::vector<double>& vec) {
-    std::shared_ptr<arrow::Array> out;
-    SwitchType(type->id(), [&](const auto& type) {
-        using TWrap = std::decay_t<decltype(type)>;
-        if constexpr (arrow::is_number_type<typename TWrap::T>::value) {
-            typename arrow::TypeTraits<typename TWrap::T>::BuilderType builder;
-            for (const auto val : vec) {
-                Y_VERIFY(builder.Append(static_cast<typename TWrap::T::c_type>(val)).ok());
-            }
-            Y_VERIFY(builder.Finish(&out).ok());
-            return true;
-        }
-        return false;
-    });
-    return out;
-}
-
 std::shared_ptr<arrow::Array> BoolVecToArray(const std::vector<bool>& vec) {
     std::shared_ptr<arrow::Array> out;
     arrow::BooleanBuilder builder;
diff --git a/ydb/core/formats/arrow_helpers.h b/ydb/core/formats/arrow_helpers.h
index c9ba1b464a0..4ec2bae4cee 100644
--- a/ydb/core/formats/arrow_helpers.h
+++ b/ydb/core/formats/arrow_helpers.h
@@ -212,8 +212,6 @@ inline bool HasNulls(const std::shared_ptr<arrow::Array>& column) {
 }
 
 bool ArrayScalarsEqual(const std::shared_ptr<arrow::Array>& lhs, const std::shared_ptr<arrow::Array>& rhs);
-std::shared_ptr<arrow::Array> NumVecToArray(const std::shared_ptr<arrow::DataType>& type,
-                                            const std::vector<double>& vec);
 std::shared_ptr<arrow::Array> BoolVecToArray(const std::vector<bool>& vec);
 
 }
diff --git a/ydb/core/formats/custom_registry.cpp b/ydb/core/formats/custom_registry.cpp
index 40be5634f2d..e211a3670f3 100644
--- a/ydb/core/formats/custom_registry.cpp
+++ b/ydb/core/formats/custom_registry.cpp
@@ -1,7 +1,7 @@
 #include "custom_registry.h"
 
-#include "functions.h"
-#include "func_common.h"
+#include <ydb/library/arrow_kernels/functions.h>
+#include <ydb/library/arrow_kernels/func_common.h>
 #include "program.h"
 
 #include <util/system/yassert.h>
@@ -19,6 +19,8 @@ namespace cp = ::arrow::compute;
 
 namespace NKikimr::NArrow {
 
+using namespace NKernels;
+
 static void RegisterMath(cp::FunctionRegistry* registry) {
     Y_VERIFY(registry->AddFunction(MakeMathUnary<TAcosh>(TAcosh::Name)).ok());
     Y_VERIFY(registry->AddFunction(MakeMathUnary<TAtanh>(TAtanh::Name)).ok());
diff --git a/ydb/core/formats/program.h b/ydb/core/formats/program.h
index 7a94f4065a0..1c383b8b23d 100644
--- a/ydb/core/formats/program.h
+++ b/ydb/core/formats/program.h
@@ -4,87 +4,12 @@
 #include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
 #include <util/system/types.h>
 
+#include <ydb/library/arrow_kernels/operations.h>
 #include <ydb/core/scheme_types/scheme_types_defs.h>
 
 namespace NKikimr::NArrow {
 
-enum class EOperation {
-    Unspecified = 0,
-    Constant,
-    //
-    CastBoolean,
-    CastInt8,
-    CastInt16,
-    CastInt32,
-    CastInt64,
-    CastUInt8,
-    CastUInt16,
-    CastUInt32,
-    CastUInt64,
-    CastFloat,
-    CastDouble,
-    CastBinary,
-    CastFixedSizeBinary,
-    CastString,
-    CastTimestamp,
-    //
-    IsValid,
-    IsNull,
-    //
-    Equal,
-    NotEqual,
-    Less,
-    LessEqual,
-    Greater,
-    GreaterEqual,
-    //
-    Invert,
-    And,
-    Or,
-    Xor,
-    //
-    Add,
-    Subtract,
-    Multiply,
-    Divide,
-    Abs,
-    Negate,
-    Gcd,
-    Lcm,
-    Modulo,
-    ModuloOrZero,
-    AddNotNull,
-    SubtractNotNull,
-    MultiplyNotNull,
-    DivideNotNull,
-    //
-    BinaryLength,
-    MatchSubstring,
-    // math
-    Acosh,
-    Atanh,
-    Cbrt,
-    Cosh,
-    E,
-    Erf,
-    Erfc,
-    Exp,
-    Exp2,
-    Exp10,
-    Hypot,
-    Lgamma,
-    Pi,
-    Sinh,
-    Sqrt,
-    Tgamma,
-    // round
-    Floor,
-    Ceil,
-    Trunc,
-    Round,
-    RoundBankers,
-    RoundToExp2
-};
+using EOperation = NKikimr::NKernels::EOperation;
 
 enum class EAggregate {
     Unspecified = 0,
diff --git a/ydb/core/formats/ut/CMakeLists.darwin.txt b/ydb/core/formats/ut/CMakeLists.darwin.txt
index d5029e328fe..fbe229af5b8 100644
--- a/ydb/core/formats/ut/CMakeLists.darwin.txt
+++ b/ydb/core/formats/ut/CMakeLists.darwin.txt
@@ -23,6 +23,7 @@ target_link_libraries(ydb-core-formats-ut PUBLIC
   cpp-testing-unittest_main
   ydb-core-formats
   libs-apache-arrow
+  ydb-library-arrow_kernels
   ydb-core-base
   udf-service-exception_policy
   yql-sql-pg_dummy
@@ -37,11 +38,7 @@ target_link_options(ydb-core-formats-ut PRIVATE
 )
 target_sources(ydb-core-formats-ut PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arrow.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arithmetic.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_math.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_round.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_program_step.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/custom_registry.cpp
 )
 add_test(
   NAME
diff --git a/ydb/core/formats/ut/CMakeLists.linux-aarch64.txt b/ydb/core/formats/ut/CMakeLists.linux-aarch64.txt
index 42a6b5a4c9f..e7c38545ffd 100644
--- a/ydb/core/formats/ut/CMakeLists.linux-aarch64.txt
+++ b/ydb/core/formats/ut/CMakeLists.linux-aarch64.txt
@@ -23,6 +23,7 @@ target_link_libraries(ydb-core-formats-ut PUBLIC
   cpp-testing-unittest_main
   ydb-core-formats
   libs-apache-arrow
+  ydb-library-arrow_kernels
   ydb-core-base
   udf-service-exception_policy
   yql-sql-pg_dummy
@@ -39,11 +40,7 @@ target_link_options(ydb-core-formats-ut PRIVATE
 )
 target_sources(ydb-core-formats-ut PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arrow.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arithmetic.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_math.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_round.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_program_step.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/custom_registry.cpp
 )
 add_test(
   NAME
diff --git a/ydb/core/formats/ut/CMakeLists.linux.txt b/ydb/core/formats/ut/CMakeLists.linux.txt
index 41b2721bf22..d86039b749a 100644
--- a/ydb/core/formats/ut/CMakeLists.linux.txt
+++ b/ydb/core/formats/ut/CMakeLists.linux.txt
@@ -25,6 +25,7 @@ target_link_libraries(ydb-core-formats-ut PUBLIC
   cpp-testing-unittest_main
   ydb-core-formats
   libs-apache-arrow
+  ydb-library-arrow_kernels
   ydb-core-base
   udf-service-exception_policy
   yql-sql-pg_dummy
@@ -41,11 +42,7 @@ target_link_options(ydb-core-formats-ut PRIVATE
 )
 target_sources(ydb-core-formats-ut PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arrow.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_arithmetic.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_math.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_round.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/formats/ut_program_step.cpp
-  ${CMAKE_SOURCE_DIR}/ydb/core/formats/custom_registry.cpp
 )
 add_test(
   NAME
diff --git a/ydb/core/formats/ut_program_step.cpp b/ydb/core/formats/ut_program_step.cpp
index a3a62ab9578..2bc31ecf473 100644
--- a/ydb/core/formats/ut_program_step.cpp
+++ b/ydb/core/formats/ut_program_step.cpp
@@ -6,12 +6,15 @@
 #include <contrib/libs/apache/arrow/cpp/src/arrow/compute/exec.h>
 #include <contrib/libs/apache/arrow/cpp/src/arrow/type_fwd.h>
 #include <library/cpp/testing/unittest/registar.h>
+#include <ydb/library/arrow_kernels/ut_common.h>
 #include "custom_registry.h"
 #include "program.h"
 #include "arrow_helpers.h"
 
 namespace NKikimr::NArrow {
 
+using NKernels::NumVecToArray;
+
 size_t FilterTest(std::vector<std::shared_ptr<arrow::Array>> args, EOperation frst, EOperation scnd) {
     auto schema = std::make_shared<arrow::Schema>(std::vector{
                                                     std::make_shared<arrow::Field>("x", args.at(0)->type()),
diff --git a/ydb/core/tx/columnshard/engines/predicate.h b/ydb/core/tx/columnshard/engines/predicate.h
index 6b50e633e85..ee209421415 100644
--- a/ydb/core/tx/columnshard/engines/predicate.h
+++ b/ydb/core/tx/columnshard/engines/predicate.h
@@ -48,7 +48,7 @@ struct TPredicate {
     }
 
     friend IOutputStream& operator << (IOutputStream& out, const TPredicate& pred) {
-        out << GetFunctionName(pred.Operation);
+        out << NArrow::GetFunctionName(pred.Operation);
         if (pred.Inclusive) {
             out << "(incl) ";
         } else {
diff --git a/ydb/library/CMakeLists.txt b/ydb/library/CMakeLists.txt
index e6671974c27..5147904f2a2 100644
--- a/ydb/library/CMakeLists.txt
+++ b/ydb/library/CMakeLists.txt
@@ -9,6 +9,7 @@
 add_subdirectory(accessor)
 add_subdirectory(aclib)
 add_subdirectory(arrow_clickhouse)
+add_subdirectory(arrow_kernels)
 add_subdirectory(backup)
 add_subdirectory(binary_json)
 add_subdirectory(dynumber)
diff --git a/ydb/library/arrow_kernels/CMakeLists.txt b/ydb/library/arrow_kernels/CMakeLists.txt
new file mode 100644
index 00000000000..ac3feb64927
--- /dev/null
+++ b/ydb/library/arrow_kernels/CMakeLists.txt
@@ -0,0 +1,20 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+add_subdirectory(ut)
+
+add_library(ydb-library-arrow_kernels)
+target_link_libraries(ydb-library-arrow_kernels PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  libs-apache-arrow
+)
+target_sources(ydb-library-arrow_kernels PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/func_cast.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_common.cpp
+)
diff --git a/ydb/core/formats/bit_cast.h b/ydb/library/arrow_kernels/bit_cast.h
index 486645877d5..6875525e63e 100644
--- a/ydb/core/formats/bit_cast.h
+++ b/ydb/library/arrow_kernels/bit_cast.h
@@ -4,6 +4,7 @@
 #include <algorithm>
 #include <type_traits>
 
+namespace NKikimr::NKernels {
 
 /** \brief Returns value `from` converted to type `To` while retaining bit representation.
   *    `To` and `From` must satisfy `CopyConstructible`.
@@ -24,4 +25,6 @@ std::decay_t<To> safe_bit_cast(const From & from)
 {
     static_assert(sizeof(To) == sizeof(From), "bit cast on types of different width");
     return bit_cast<To, From>(from);
-}
\ No newline at end of file
+}
+
+}
diff --git a/ydb/core/formats/clickhouse_type_traits.h b/ydb/library/arrow_kernels/clickhouse_type_traits.h
index b694d2b91fb..15cf4956855 100644
--- a/ydb/core/formats/clickhouse_type_traits.h
+++ b/ydb/library/arrow_kernels/clickhouse_type_traits.h
@@ -3,7 +3,7 @@
 #include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
 #include <contrib/libs/apache/arrow/cpp/src/arrow/type.h>
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 constexpr size_t NextSize(size_t size) {
     if (size < 8) {
diff --git a/ydb/core/formats/execs.h b/ydb/library/arrow_kernels/execs.h
index 253bae48318..95311448450 100644
--- a/ydb/core/formats/execs.h
+++ b/ydb/library/arrow_kernels/execs.h
@@ -13,17 +13,17 @@
 
 #include <contrib/libs/apache/arrow/cpp/src/arrow/compute/function.h>
 #include <contrib/libs/apache/arrow/cpp/src/arrow/compute/cast.h>
+
 #include <util/datetime/base.h>
 #include <util/system/yassert.h>
 #include <cstdint>
 #include <memory>
 #include <type_traits>
 #include <vector>
-#include "switch_type.h"
 
 namespace cp = arrow::compute;
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 template <template <typename... Args> class KernelGenerator, typename Op>
 cp::ArrayKernelExec ArithmeticBinaryExec(cp::internal::detail::GetTypeId getId) {
diff --git a/ydb/core/formats/func_cast.cpp b/ydb/library/arrow_kernels/func_cast.cpp
index 7d480bdac07..9481584fad4 100644
--- a/ydb/core/formats/func_cast.cpp
+++ b/ydb/library/arrow_kernels/func_cast.cpp
@@ -13,7 +13,7 @@
 
 namespace cp = ::arrow::compute;
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 namespace {
 
@@ -118,7 +118,8 @@ YdbCastMetaFunction::YdbCastMetaFunction()
     return castFunc->Execute(args, options, ctx);
 }
 
-} // NKikimr::NArrow
+}
+
 
 namespace arrow::compute::internal {
 
@@ -129,17 +130,17 @@ struct CastFunctor<TimestampType, UInt32Type> {
             return ::arrow::Status::IndexError("Cast from uint32 to timestamp received empty batch.");
         }
         Y_VERIFY(batch[0].kind() == Datum::ARRAY, "Cast from uint32 to timestamp expected ARRAY as input.");
-    
+
         const auto& out_type = checked_cast<const ::arrow::TimestampType&>(*out->type());
         // get conversion MICROSECONDS -> unit
         auto conversion = ::arrow::util::GetTimestampConversion(::arrow::TimeUnit::MICRO, out_type.unit());
         Y_VERIFY(conversion.first == ::arrow::util::MULTIPLY, "Cast from uint32 to timestamp failed because timestamp unit is greater than seconds.");
-    
+
         auto input = batch[0].array();
         auto output = out->mutable_array();
         auto in_data = input->GetValues<uint32_t>(1);
         auto out_data = output->GetMutableValues<int64_t>(1);
-    
+
         for (int64_t i = 0; i < input->length; i++) {
             out_data[i] = static_cast<int64_t>(in_data[i] * conversion.second);
         }
@@ -147,4 +148,4 @@ struct CastFunctor<TimestampType, UInt32Type> {
     }
 };
 
-} // namespace arrow::compute::internal
\ No newline at end of file
+} // namespace arrow::compute::internal
diff --git a/ydb/core/formats/func_cast.h b/ydb/library/arrow_kernels/func_cast.h
index 532f3164d7c..5e2a20150c1 100644
--- a/ydb/core/formats/func_cast.h
+++ b/ydb/library/arrow_kernels/func_cast.h
@@ -4,7 +4,7 @@
 
 #include <type_traits>
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 // Metafunction for dispatching to appropriate CastFunction. This corresponds
 // to the standard SQL CAST(expr AS target_type)
@@ -19,4 +19,4 @@ class YdbCastMetaFunction : public ::arrow::compute::MetaFunction {
                             ::arrow::compute::ExecContext* ctx) const override;
 };
 
-} // NKikimr::NArrow
+}
diff --git a/ydb/core/formats/func_common.h b/ydb/library/arrow_kernels/func_common.h
index c697fcf03ed..62c74c06a5c 100644
--- a/ydb/core/formats/func_common.h
+++ b/ydb/library/arrow_kernels/func_common.h
@@ -10,14 +10,13 @@
 
 #include <type_traits>
 
-#include "switch_type.h"
 #include "execs.h"
 
 namespace cp = arrow::compute;
 using cp::internal::applicator::ScalarBinary;
 using cp::internal::applicator::ScalarUnary;
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 template <typename T>
 using IsUnsignedInteger =
diff --git a/ydb/core/formats/func_gcd.h b/ydb/library/arrow_kernels/func_gcd.h
index 0f6fe4bc8dc..bc2ad222067 100644
--- a/ydb/core/formats/func_gcd.h
+++ b/ydb/library/arrow_kernels/func_gcd.h
@@ -9,7 +9,7 @@
 #define INLINE __attribute__((always_inline))
 #endif
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 template<typename T>
 INLINE void FastIntSwap(T& lhs, T& rhs) {
diff --git a/ydb/core/formats/func_lcm.h b/ydb/library/arrow_kernels/func_lcm.h
index 1e31b0e516a..986b8193e26 100644
--- a/ydb/core/formats/func_lcm.h
+++ b/ydb/library/arrow_kernels/func_lcm.h
@@ -4,7 +4,7 @@
 #include "func_gcd.h"
 #include "clickhouse_type_traits.h"
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 struct TLeastCommonMultiple {
 
diff --git a/ydb/core/formats/func_math.h b/ydb/library/arrow_kernels/func_math.h
index 13474e0018e..85b38408e25 100644
--- a/ydb/core/formats/func_math.h
+++ b/ydb/library/arrow_kernels/func_math.h
@@ -2,7 +2,7 @@
 #include "func_common.h"
 #include <cmath>
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 struct TAcosh {
 
diff --git a/ydb/core/formats/func_modulo.h b/ydb/library/arrow_kernels/func_modulo.h
index 85bcc7b9a12..742bc083ce2 100644
--- a/ydb/core/formats/func_modulo.h
+++ b/ydb/library/arrow_kernels/func_modulo.h
@@ -2,7 +2,7 @@
 #include "func_common.h"
 #include "clickhouse_type_traits.h"
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 struct TModulo {
 
diff --git a/ydb/core/formats/func_modulo_or_zero.h b/ydb/library/arrow_kernels/func_modulo_or_zero.h
index a2a4a494a37..06e2d18901b 100644
--- a/ydb/core/formats/func_modulo_or_zero.h
+++ b/ydb/library/arrow_kernels/func_modulo_or_zero.h
@@ -2,7 +2,7 @@
 #include "func_common.h"
 #include "clickhouse_type_traits.h"
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 struct TModuloOrZero {
 
diff --git a/ydb/core/formats/func_mul.h b/ydb/library/arrow_kernels/func_mul.h
index d9be3198e9e..7cfb4d34676 100644
--- a/ydb/core/formats/func_mul.h
+++ b/ydb/library/arrow_kernels/func_mul.h
@@ -2,7 +2,7 @@
 
 namespace cp = arrow::compute;
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 template <typename T, typename TUnsigned = typename std::make_unsigned<T>::type>
 constexpr TUnsigned ToUnsigned(T sgnd) {
diff --git a/ydb/core/formats/func_round.h b/ydb/library/arrow_kernels/func_round.h
index 3b82591c45c..639d19af48a 100644
--- a/ydb/core/formats/func_round.h
+++ b/ydb/library/arrow_kernels/func_round.h
@@ -18,7 +18,7 @@
 #define CLZLL __builtin_clzll
 #endif
 
-namespace NKikimr::NArrow {
+namespace NKikimr::NKernels {
 
 struct TRound {
 
diff --git a/ydb/core/formats/functions.h b/ydb/library/arrow_kernels/functions.h
index 2f4523a4fe2..2f4523a4fe2 100644
--- a/ydb/core/formats/functions.h
+++ b/ydb/library/arrow_kernels/functions.h
diff --git a/ydb/library/arrow_kernels/operations.h b/ydb/library/arrow_kernels/operations.h
new file mode 100644
index 00000000000..02f04ce75f9
--- /dev/null
+++ b/ydb/library/arrow_kernels/operations.h
@@ -0,0 +1,81 @@
+namespace NKikimr::NKernels {
+
+enum class EOperation {
+    Unspecified = 0,
+    Constant,
+    //
+    CastBoolean,
+    CastInt8,
+    CastInt16,
+    CastInt32,
+    CastInt64,
+    CastUInt8,
+    CastUInt16,
+    CastUInt32,
+    CastUInt64,
+    CastFloat,
+    CastDouble,
+    CastBinary,
+    CastFixedSizeBinary,
+    CastString,
+    CastTimestamp,
+    //
+    IsValid,
+    IsNull,
+    //
+    Equal,
+    NotEqual,
+    Less,
+    LessEqual,
+    Greater,
+    GreaterEqual,
+    //
+    Invert,
+    And,
+    Or,
+    Xor,
+    //
+    Add,
+    Subtract,
+    Multiply,
+    Divide,
+    Abs,
+    Negate,
+    Gcd,
+    Lcm,
+    Modulo,
+    ModuloOrZero,
+    AddNotNull,
+    SubtractNotNull,
+    MultiplyNotNull,
+    DivideNotNull,
+    //
+    BinaryLength,
+    MatchSubstring,
+    // math
+    Acosh,
+    Atanh,
+    Cbrt,
+    Cosh,
+    E,
+    Erf,
+    Erfc,
+    Exp,
+    Exp2,
+    Exp10,
+    Hypot,
+    Lgamma,
+    Pi,
+    Sinh,
+    Sqrt,
+    Tgamma,
+    // round
+    Floor,
+    Ceil,
+    Trunc,
+    Round,
+    RoundBankers,
+    RoundToExp2
+};
+
+}
diff --git a/ydb/library/arrow_kernels/ut/CMakeLists.darwin.txt b/ydb/library/arrow_kernels/ut/CMakeLists.darwin.txt
new file mode 100644
index 00000000000..e57be38692d
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut/CMakeLists.darwin.txt
@@ -0,0 +1,49 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_executable(ydb-library-arrow_kernels-ut)
+target_compile_options(ydb-library-arrow_kernels-ut PRIVATE
+  -Wno-unused-parameter
+)
+target_include_directories(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_clickhouse
+)
+target_link_libraries(ydb-library-arrow_kernels-ut PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-cpuid_check
+  cpp-testing-unittest_main
+  ydb-library-arrow_kernels
+  libs-apache-arrow
+)
+target_link_options(ydb-library-arrow_kernels-ut PRIVATE
+  -Wl,-no_deduplicate
+  -Wl,-sdk_version,10.15
+  -fPIC
+  -fPIC
+)
+target_sources(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_common.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_arithmetic.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_math.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_round.cpp
+)
+add_test(
+  NAME
+  ydb-library-arrow_kernels-ut
+  COMMAND
+  ydb-library-arrow_kernels-ut
+  --print-before-suite
+  --print-before-test
+  --fork-tests
+  --print-times
+  --show-fails
+)
+vcs_info(ydb-library-arrow_kernels-ut)
diff --git a/ydb/library/arrow_kernels/ut/CMakeLists.linux-aarch64.txt b/ydb/library/arrow_kernels/ut/CMakeLists.linux-aarch64.txt
new file mode 100644
index 00000000000..65aef6876b7
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut/CMakeLists.linux-aarch64.txt
@@ -0,0 +1,53 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_executable(ydb-library-arrow_kernels-ut)
+target_compile_options(ydb-library-arrow_kernels-ut PRIVATE
+  -Wno-unused-parameter
+)
+target_include_directories(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_clickhouse
+)
+target_link_libraries(ydb-library-arrow_kernels-ut PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-lfalloc
+  cpp-testing-unittest_main
+  ydb-library-arrow_kernels
+  libs-apache-arrow
+)
+target_link_options(ydb-library-arrow_kernels-ut PRIVATE
+  -ldl
+  -lrt
+  -Wl,--no-as-needed
+  -fPIC
+  -fPIC
+  -lpthread
+  -lrt
+  -ldl
+)
+target_sources(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_common.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_arithmetic.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_math.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_round.cpp
+)
+add_test(
+  NAME
+  ydb-library-arrow_kernels-ut
+  COMMAND
+  ydb-library-arrow_kernels-ut
+  --print-before-suite
+  --print-before-test
+  --fork-tests
+  --print-times
+  --show-fails
+)
+vcs_info(ydb-library-arrow_kernels-ut)
diff --git a/ydb/library/arrow_kernels/ut/CMakeLists.linux.txt b/ydb/library/arrow_kernels/ut/CMakeLists.linux.txt
new file mode 100644
index 00000000000..f3e17ee71dd
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut/CMakeLists.linux.txt
@@ -0,0 +1,55 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_executable(ydb-library-arrow_kernels-ut)
+target_compile_options(ydb-library-arrow_kernels-ut PRIVATE
+  -Wno-unused-parameter
+)
+target_include_directories(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_clickhouse
+)
+target_link_libraries(ydb-library-arrow_kernels-ut PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  cpp-malloc-tcmalloc
+  libs-tcmalloc-no_percpu_cache
+  library-cpp-cpuid_check
+  cpp-testing-unittest_main
+  ydb-library-arrow_kernels
+  libs-apache-arrow
+)
+target_link_options(ydb-library-arrow_kernels-ut PRIVATE
+  -ldl
+  -lrt
+  -Wl,--no-as-needed
+  -fPIC
+  -fPIC
+  -lpthread
+  -lrt
+  -ldl
+)
+target_sources(ydb-library-arrow_kernels-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_common.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_arithmetic.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_math.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/arrow_kernels/ut_round.cpp
+)
+add_test(
+  NAME
+  ydb-library-arrow_kernels-ut
+  COMMAND
+  ydb-library-arrow_kernels-ut
+  --print-before-suite
+  --print-before-test
+  --fork-tests
+  --print-times
+  --show-fails
+)
+vcs_info(ydb-library-arrow_kernels-ut)
diff --git a/ydb/library/arrow_kernels/ut/CMakeLists.txt b/ydb/library/arrow_kernels/ut/CMakeLists.txt
new file mode 100644
index 00000000000..3e0811fb22e
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut/CMakeLists.txt
@@ -0,0 +1,15 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+if (CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64" AND UNIX AND NOT APPLE AND NOT ANDROID)
+  include(CMakeLists.linux-aarch64.txt)
+elseif (APPLE)
+  include(CMakeLists.darwin.txt)
+elseif (CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64" AND UNIX AND NOT APPLE AND NOT ANDROID)
+  include(CMakeLists.linux.txt)
+endif()
diff --git a/ydb/core/formats/ut_arithmetic.cpp b/ydb/library/arrow_kernels/ut_arithmetic.cpp
index ffffad99cc8..8523dc0f748 100644
--- a/ydb/core/formats/ut_arithmetic.cpp
+++ b/ydb/library/arrow_kernels/ut_arithmetic.cpp
@@ -1,25 +1,12 @@
-#include <cmath>
-#include <cstdint>
-#include <iterator>
-#include <library/cpp/testing/unittest/registar.h>
-#include <ctime>
-#include <vector>
-#include <algorithm>
-
-#include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
-#include <contrib/libs/apache/arrow/cpp/src/arrow/compute/api.h>
-
-#include "func_common.h"
-#include "functions.h"
-#include "custom_registry.h"
-#include "arrow_helpers.h"
-
-namespace NKikimr::NArrow {
+#include "ut_common.h"
 
 namespace cp = ::arrow::compute;
 using cp::internal::applicator::ScalarBinary;
 using cp::internal::applicator::ScalarUnary;
 
+
+namespace NKikimr::NKernels {
+
 const std::vector<std::shared_ptr<arrow::DataType>> nonPrimitiveTypes = {
     arrow::list(arrow::utf8()),
     arrow::list(arrow::int64()),
@@ -343,7 +330,6 @@ Y_UNIT_TEST_SUITE(ModuloTest) {
             UNIT_ASSERT(!res.ok());
         }
     }
-
 }
 
 Y_UNIT_TEST_SUITE(ModuloOrZeroTest) {
@@ -366,7 +352,6 @@ Y_UNIT_TEST_SUITE(ModuloOrZeroTest) {
             UNIT_ASSERT(res->Equals(expected));
         }
     }
-
 }
 
 }
diff --git a/ydb/library/arrow_kernels/ut_common.cpp b/ydb/library/arrow_kernels/ut_common.cpp
new file mode 100644
index 00000000000..e2ff267b7e8
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut_common.cpp
@@ -0,0 +1,168 @@
+#include "ut_common.h"
+#include "operations.h"
+
+#include <util/system/yassert.h>
+
+namespace cp = ::arrow::compute;
+
+
+namespace NKikimr::NKernels {
+
+template <typename TType>
+struct TTypeWrapper
+{
+    using T = TType;
+};
+
+template <typename TFunc, bool EnableNull = false>
+bool SwitchType(arrow::Type::type typeId, TFunc&& f) {
+    switch (typeId) {
+        case arrow::Type::NA: {
+            if constexpr (EnableNull) {
+                return f(TTypeWrapper<arrow::NullType>());
+            }
+            break;
+        }
+        case arrow::Type::BOOL:
+            return f(TTypeWrapper<arrow::BooleanType>());
+        case arrow::Type::UINT8:
+            return f(TTypeWrapper<arrow::UInt8Type>());
+        case arrow::Type::INT8:
+            return f(TTypeWrapper<arrow::Int8Type>());
+        case arrow::Type::UINT16:
+            return f(TTypeWrapper<arrow::UInt16Type>());
+        case arrow::Type::INT16:
+            return f(TTypeWrapper<arrow::Int16Type>());
+        case arrow::Type::UINT32:
+            return f(TTypeWrapper<arrow::UInt32Type>());
+        case arrow::Type::INT32:
+            return f(TTypeWrapper<arrow::Int32Type>());
+        case arrow::Type::UINT64:
+            return f(TTypeWrapper<arrow::UInt64Type>());
+        case arrow::Type::INT64:
+            return f(TTypeWrapper<arrow::Int64Type>());
+        case arrow::Type::HALF_FLOAT:
+            return f(TTypeWrapper<arrow::HalfFloatType>());
+        case arrow::Type::FLOAT:
+            return f(TTypeWrapper<arrow::FloatType>());
+        case arrow::Type::DOUBLE:
+            return f(TTypeWrapper<arrow::DoubleType>());
+        case arrow::Type::STRING:
+            return f(TTypeWrapper<arrow::StringType>());
+        case arrow::Type::BINARY:
+            return f(TTypeWrapper<arrow::BinaryType>());
+        case arrow::Type::FIXED_SIZE_BINARY:
+            return f(TTypeWrapper<arrow::FixedSizeBinaryType>());
+        case arrow::Type::DATE32:
+            return f(TTypeWrapper<arrow::Date32Type>());
+        case arrow::Type::DATE64:
+            return f(TTypeWrapper<arrow::Date64Type>());
+        case arrow::Type::TIMESTAMP:
+            return f(TTypeWrapper<arrow::TimestampType>());
+        case arrow::Type::TIME32:
+            return f(TTypeWrapper<arrow::Time32Type>());
+        case arrow::Type::TIME64:
+            return f(TTypeWrapper<arrow::Time64Type>());
+        case arrow::Type::INTERVAL_MONTHS:
+            return f(TTypeWrapper<arrow::MonthIntervalType>());
+        case arrow::Type::DECIMAL:
+            return f(TTypeWrapper<arrow::Decimal128Type>());
+        case arrow::Type::DURATION:
+            return f(TTypeWrapper<arrow::DurationType>());
+        case arrow::Type::LARGE_STRING:
+            return f(TTypeWrapper<arrow::LargeStringType>());
+        case arrow::Type::LARGE_BINARY:
+            return f(TTypeWrapper<arrow::LargeBinaryType>());
+        case arrow::Type::DECIMAL256:
+        case arrow::Type::DENSE_UNION:
+        case arrow::Type::DICTIONARY:
+        case arrow::Type::EXTENSION:
+        case arrow::Type::FIXED_SIZE_LIST:
+        case arrow::Type::INTERVAL_DAY_TIME:
+        case arrow::Type::LARGE_LIST:
+        case arrow::Type::LIST:
+        case arrow::Type::MAP:
+        case arrow::Type::MAX_ID:
+        case arrow::Type::SPARSE_UNION:
+        case arrow::Type::STRUCT:
+            break;
+    }
+
+    return false;
+}
+
+std::shared_ptr<arrow::Array> NumVecToArray(const std::shared_ptr<arrow::DataType>& type,
+                                            const std::vector<double>& vec) {
+    std::shared_ptr<arrow::Array> out;
+    SwitchType(type->id(), [&](const auto& type) {
+        using TWrap = std::decay_t<decltype(type)>;
+        if constexpr (arrow::is_number_type<typename TWrap::T>::value) {
+            typename arrow::TypeTraits<typename TWrap::T>::BuilderType builder;
+            for (const auto val : vec) {
+                Y_VERIFY(builder.Append(static_cast<typename TWrap::T::c_type>(val)).ok());
+            }
+            Y_VERIFY(builder.Finish(&out).ok());
+            return true;
+        }
+        return false;
+    });
+    return out;
+}
+
+static void RegisterMath(cp::FunctionRegistry* registry) {
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TAcosh>(TAcosh::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TAtanh>(TAtanh::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TCbrt>(TCbrt::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TCosh>(TCosh::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeConstNullary<TE>(TE::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TErf>(TErf::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TErfc>(TErfc::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TExp>(TExp::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TExp2>(TExp2::Name)).ok());
+    // Temporarily disabled because of compilation error on Windows.
+#if 0
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TExp10>(TExp10::Name)).ok());
+#endif
+    Y_VERIFY(registry->AddFunction(MakeMathBinary<THypot>(THypot::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TLgamma>(TLgamma::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeConstNullary<TPi>(TPi::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TSinh>(TSinh::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TSqrt>(TSqrt::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeMathUnary<TTgamma>(TTgamma::Name)).ok());
+}
+
+static void RegisterRound(cp::FunctionRegistry* registry) {
+    Y_VERIFY(registry->AddFunction(MakeArithmeticUnary<TRound>(TRound::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeArithmeticUnary<TRoundBankers>(TRoundBankers::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeArithmeticUnary<TRoundToExp2>(TRoundToExp2::Name)).ok());
+}
+
+static void RegisterArithmetic(cp::FunctionRegistry* registry) {
+    Y_VERIFY(registry->AddFunction(MakeArithmeticIntBinary<TGreatestCommonDivisor>(TGreatestCommonDivisor::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeArithmeticIntBinary<TLeastCommonMultiple>(TLeastCommonMultiple::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeArithmeticBinary<TModulo>(TModulo::Name)).ok());
+    Y_VERIFY(registry->AddFunction(MakeArithmeticBinary<TModuloOrZero>(TModuloOrZero::Name)).ok());
+}
+
+static std::unique_ptr<cp::FunctionRegistry> CreateCustomRegistry() {
+    auto registry = cp::FunctionRegistry::Make();
+    RegisterMath(registry.get());
+    RegisterRound(registry.get());
+    RegisterArithmetic(registry.get());
+    cp::internal::RegisterScalarCast(registry.get());
+    return registry;
+}
+
+// Creates singleton custom registry
+cp::FunctionRegistry* GetCustomFunctionRegistry() {
+    static auto g_registry = CreateCustomRegistry();
+    return g_registry.get();
+}
+
+// We want to have ExecContext per thread. All these context use one custom registry.
+cp::ExecContext* GetCustomExecContext() {
+    static thread_local cp::ExecContext context(arrow::default_memory_pool(), nullptr, GetCustomFunctionRegistry());
+    return &context;
+}
+
+}
diff --git a/ydb/library/arrow_kernels/ut_common.h b/ydb/library/arrow_kernels/ut_common.h
new file mode 100644
index 00000000000..b1095c3def6
--- /dev/null
+++ b/ydb/library/arrow_kernels/ut_common.h
@@ -0,0 +1,24 @@
+#include <cmath>
+#include <cstdint>
+#include <iterator>
+#include <ctime>
+#include <vector>
+#include <algorithm>
+
+#include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
+#include <contrib/libs/apache/arrow/cpp/src/arrow/compute/api.h>
+#include <contrib/libs/apache/arrow/cpp/src/arrow/compute/registry_internal.h>
+
+#include <library/cpp/testing/unittest/registar.h>
+
+#include "func_common.h"
+#include "functions.h"
+
+namespace NKikimr::NKernels {
+
+std::shared_ptr<arrow::Array> NumVecToArray(const std::shared_ptr<arrow::DataType>& type,
+                                            const std::vector<double>& vec);
+
+arrow::compute::ExecContext* GetCustomExecContext();
+
+}
diff --git a/ydb/core/formats/ut_math.cpp b/ydb/library/arrow_kernels/ut_math.cpp
index c9945c4e096..44507bbd19d 100644
--- a/ydb/core/formats/ut_math.cpp
+++ b/ydb/library/arrow_kernels/ut_math.cpp
@@ -1,26 +1,10 @@
-#include <cmath>
-#include <cstdint>
-#include <iterator>
-#include <library/cpp/testing/unittest/registar.h>
-#include <ctime>
-#include <vector>
-#include <algorithm>
-
-#include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
-#include <contrib/libs/apache/arrow/cpp/src/arrow/compute/api.h>
-
-
-#include "func_common.h"
-#include "functions.h"
-#include "custom_registry.h"
-#include "arrow_helpers.h"
-
-
-namespace NKikimr::NArrow {
+#include "ut_common.h"
 
 namespace cp = ::arrow::compute;
 
 
+namespace NKikimr::NKernels {
+
 Y_UNIT_TEST_SUITE(MathTest) {
     Y_UNIT_TEST(E) {
         auto res = arrow::compute::CallFunction(TE::Name, {}, GetCustomExecContext());
@@ -66,7 +50,6 @@ Y_UNIT_TEST_SUITE(MathTest) {
             UNIT_ASSERT(res->Equals(expRes));
         }
     }
-
 }
 
 }
diff --git a/ydb/core/formats/ut_round.cpp b/ydb/library/arrow_kernels/ut_round.cpp
index 7b639b0888d..39cb983615f 100644
--- a/ydb/core/formats/ut_round.cpp
+++ b/ydb/library/arrow_kernels/ut_round.cpp
@@ -1,25 +1,10 @@
-#include <cmath>
-#include <cstdint>
-#include <iterator>
-#include <library/cpp/testing/unittest/registar.h>
-#include <ctime>
-#include <vector>
-#include <algorithm>
-
-#include <contrib/libs/apache/arrow/cpp/src/arrow/api.h>
-#include <contrib/libs/apache/arrow/cpp/src/arrow/compute/api.h>
-
-#include "func_common.h"
-#include "functions.h"
-#include "custom_registry.h"
-#include "arrow_helpers.h"
-
-
-namespace NKikimr::NArrow {
+#include "ut_common.h"
 
 namespace cp = ::arrow::compute;
 
 
+namespace NKikimr::NKernels {
+
 Y_UNIT_TEST_SUITE(RoundsTest) {
     Y_UNIT_TEST(RoundTest) {
         for (auto ty : cp::internal::FloatingPointTypes()) {
@@ -47,7 +32,6 @@ Y_UNIT_TEST_SUITE(RoundsTest) {
             UNIT_ASSERT(res->Equals(expRes));
         }
     }
-
 }
 
 }