Test json_exists

18 files changed, 146 insertions, 14 deletions

diff --git a/ydb/core/tx/columnshard/engines/ut/CMakeLists.darwin-x86_64.txt b/ydb/core/tx/columnshard/engines/ut/CMakeLists.darwin-x86_64.txt
index d59807c727..5810287df7 100644
--- a/ydb/core/tx/columnshard/engines/ut/CMakeLists.darwin-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/ut/CMakeLists.darwin-x86_64.txt
@@ -28,6 +28,7 @@ target_link_libraries(ydb-core-tx-columnshard-engines-ut PUBLIC
   yql-sql-pg_dummy
   core-arrow_kernels-request
   core-testlib-default
+  json2_udf
 )
 target_link_options(ydb-core-tx-columnshard-engines-ut PRIVATE
   -Wl,-platform_version,macos,11.0,11.0
diff --git a/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-aarch64.txt b/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-aarch64.txt
index 50f51e0ad2..3637b62e83 100644
--- a/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-aarch64.txt
+++ b/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-aarch64.txt
@@ -28,6 +28,7 @@ target_link_libraries(ydb-core-tx-columnshard-engines-ut PUBLIC
   yql-sql-pg_dummy
   core-arrow_kernels-request
   core-testlib-default
+  json2_udf
 )
 target_link_options(ydb-core-tx-columnshard-engines-ut PRIVATE
   -ldl
diff --git a/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-x86_64.txt b/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-x86_64.txt
index 47a14cdffd..62b2cbc742 100644
--- a/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/ut/CMakeLists.linux-x86_64.txt
@@ -29,6 +29,7 @@ target_link_libraries(ydb-core-tx-columnshard-engines-ut PUBLIC
   yql-sql-pg_dummy
   core-arrow_kernels-request
   core-testlib-default
+  json2_udf
 )
 target_link_options(ydb-core-tx-columnshard-engines-ut PRIVATE
   -ldl
diff --git a/ydb/core/tx/columnshard/engines/ut/CMakeLists.windows-x86_64.txt b/ydb/core/tx/columnshard/engines/ut/CMakeLists.windows-x86_64.txt
index c2cce5aea1..fc4d07a486 100644
--- a/ydb/core/tx/columnshard/engines/ut/CMakeLists.windows-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/ut/CMakeLists.windows-x86_64.txt
@@ -28,6 +28,7 @@ target_link_libraries(ydb-core-tx-columnshard-engines-ut PUBLIC
   yql-sql-pg_dummy
   core-arrow_kernels-request
   core-testlib-default
+  json2_udf
 )
 target_sources(ydb-core-tx-columnshard-engines-ut PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/ut_insert_table.cpp
diff --git a/ydb/core/tx/columnshard/engines/ut/ya.make b/ydb/core/tx/columnshard/engines/ut/ya.make
index 0d7ee35e64..04f31654b9 100644
--- a/ydb/core/tx/columnshard/engines/ut/ya.make
+++ b/ydb/core/tx/columnshard/engines/ut/ya.make
@@ -23,6 +23,8 @@ PEERDIR(
     ydb/library/yql/sql/pg_dummy
     ydb/library/yql/core/arrow_kernels/request
     ydb/core/testlib/default
+
+    ydb/library/yql/udfs/common/json2
 )
 
 YQL_LAST_ABI_VERSION()
diff --git a/ydb/core/tx/columnshard/engines/ut_program.cpp b/ydb/core/tx/columnshard/engines/ut_program.cpp
index f5f8561907..f6be9dcdc6 100644
--- a/ydb/core/tx/columnshard/engines/ut_program.cpp
+++ b/ydb/core/tx/columnshard/engines/ut_program.cpp
@@ -4,6 +4,7 @@
 #include <ydb/core/tx/columnshard/columnshard__index_scan.h>
 #include <ydb/core/tx/columnshard/columnshard_ut_common.h>
 #include <ydb/core/tx/program/program.h>
+#include <ydb/core/formats/arrow/converter.h>
 
 #include <ydb/library/yql/core/arrow_kernels/request/request.h>
 #include <ydb/library/yql/core/arrow_kernels/registry/registry.h>
@@ -37,12 +38,13 @@ Y_UNIT_TEST_SUITE(TestProgram) {
 
     class TKernelsWrapper {
         TIntrusivePtr<NMiniKQL::IFunctionRegistry> Reg;
-        NYql::TKernelRequestBuilder ReqBuilder;
+        std::unique_ptr<NYql::TKernelRequestBuilder> ReqBuilder;
         public:
-            TKernelsWrapper()
-                : Reg(CreateFunctionRegistry(NMiniKQL::CreateBuiltinRegistry()))
-                , ReqBuilder((*Reg)) {
-
+            TKernelsWrapper() {
+                auto reg = CreateFunctionRegistry(NMiniKQL::CreateBuiltinRegistry())->Clone();
+                NMiniKQL::FillStaticModules(*reg);
+                Reg.Reset(reg.Release());
+                ReqBuilder = std::make_unique<NYql::TKernelRequestBuilder>(*Reg);
             }
 
             ui32 Add(NYql::TKernelRequestBuilder::EBinaryOp operation) {
@@ -51,21 +53,21 @@ Y_UNIT_TEST_SUITE(TestProgram) {
                     {
                         NYql::TExprContext ctx;
                         auto blockInt32Type = ctx.template MakeType<NYql::TBlockExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Int32));
-                        return ReqBuilder.AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::Add, blockInt32Type, blockInt32Type, blockInt32Type);
+                        return ReqBuilder->AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::Add, blockInt32Type, blockInt32Type, blockInt32Type);
                     }
                     case NYql::TKernelRequestBuilder::EBinaryOp::StartsWith:
                     {
                         NYql::TExprContext ctx;
                         auto blockStringType = ctx.template MakeType<NYql::TBlockExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Utf8));
                         auto blockBoolType = ctx.template MakeType<NYql::TBlockExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Bool));
-                        return ReqBuilder.AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::StartsWith, blockStringType, blockStringType, blockBoolType);
+                        return ReqBuilder->AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::StartsWith, blockStringType, blockStringType, blockBoolType);
                     }
                     case NYql::TKernelRequestBuilder::EBinaryOp::StringContains:
                     {
                         NYql::TExprContext ctx;
                         auto blockStringType = ctx.template MakeType<NYql::TBlockExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::String));
                         auto blockBoolType = ctx.template MakeType<NYql::TBlockExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Bool));
-                        return ReqBuilder.AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::StringContains, blockStringType, blockStringType, blockBoolType);
+                        return ReqBuilder->AddBinaryOp(NYql::TKernelRequestBuilder::EBinaryOp::StringContains, blockStringType, blockStringType, blockBoolType);
                     }
                     default:
                         Y_FAIL("Not implemented");
@@ -73,8 +75,21 @@ Y_UNIT_TEST_SUITE(TestProgram) {
                 }
             } 
 
+            ui32 AddJsonExists(bool isBinaryType = true) {
+                NYql::TExprContext ctx;
+                auto blockOptJsonType = ctx.template MakeType<NYql::TBlockExprType>(
+                    ctx.template MakeType<NYql::TOptionalExprType>(
+                    ctx.template MakeType<NYql::TDataExprType>(isBinaryType ? NYql::EDataSlot::JsonDocument : NYql::EDataSlot::Json)));
+                auto scalarStringType = ctx.template MakeType<NYql::TScalarExprType>(ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Utf8));
+                auto blockBoolType = ctx.template MakeType<NYql::TBlockExprType>(
+                    ctx.template MakeType<NYql::TOptionalExprType>(
+                    ctx.template MakeType<NYql::TDataExprType>(NYql::EDataSlot::Bool)));
+
+                return ReqBuilder->JsonExists(blockOptJsonType, scalarStringType, blockBoolType);
+            }
+
             TString Serialize() {
-                return ReqBuilder.Serialize();
+                return ReqBuilder->Serialize();
             }            
     };
 
@@ -216,6 +231,75 @@ Y_UNIT_TEST_SUITE(TestProgram) {
         }
     }
 
+    void JsonExistsImpl(bool isBinaryType) {
+        TIndexInfo indexInfo = BuildTableInfo(testColumns, testKey);
+        TIndexColumnResolver columnResolver(indexInfo);
+
+        NKikimrSSA::TProgram programProto;
+        {
+            auto* command = programProto.AddCommand();
+            auto* constantProto = command->MutableAssign()->MutableConstant();
+            constantProto->SetText("$.key");
+            command->MutableAssign()->MutableColumn()->SetName("json_path");
+        }
+        {
+            auto* command = programProto.AddCommand();
+            auto* functionProto = command->MutableAssign()->MutableFunction();
+            functionProto->SetFunctionType(NKikimrSSA::TProgram::EFunctionType::TProgram_EFunctionType_YQL_KERNEL);
+            functionProto->SetKernelIdx(0);
+            functionProto->AddArguments()->SetName("json_data");
+            functionProto->AddArguments()->SetName("json_path");
+            functionProto->SetId(NKikimrSSA::TProgram::TAssignment::EFunction::TProgram_TAssignment_EFunction_FUNC_STR_LENGTH);
+        }
+        {
+            auto* command = programProto.AddCommand();
+            auto* prjectionProto = command->MutableProjection();
+            auto* column = prjectionProto->AddColumns();
+            column->SetName("0");
+        }
+
+        TKernelsWrapper kernels;
+        kernels.AddJsonExists(isBinaryType);
+        const auto programSerialized = SerializeProgram(programProto, kernels.Serialize());
+       
+        TProgramContainer program;
+        TString errors;
+        UNIT_ASSERT_C(program.Init(columnResolver, NKikimrSchemeOp::EOlapProgramType::OLAP_PROGRAM_SSA_PROGRAM_WITH_PARAMETERS, programSerialized, errors), errors);
+
+        TTableUpdatesBuilder updates(NArrow::MakeArrowSchema({{"json_data", TTypeInfo(isBinaryType ? NTypeIds::JsonDocument : NTypeIds::Utf8) }}));
+        NJson::TJsonValue testJson;
+        testJson["key"] = "value";
+        updates.AddRow().Add<std::string>(testJson.GetStringRobust());
+        updates.AddRow().Add<std::string>(NJson::TJsonValue(NJson::JSON_ARRAY).GetStringRobust());
+
+        auto batch = updates.BuildArrow();
+        Cerr << batch->ToString() << Endl;
+
+        if (isBinaryType) {
+            THashMap<TString, NScheme::TTypeInfo> cc;
+            cc["json_data"] = TTypeInfo(NTypeIds::JsonDocument);
+            batch = NArrow::ConvertColumns(batch, cc);
+            Cerr << batch->ToString() << Endl;
+        }
+        auto res = program.ApplyProgram(batch);
+        UNIT_ASSERT_C(res.ok(), res.ToString());
+
+        TTableUpdatesBuilder result(NArrow::MakeArrowSchema( { std::make_pair("0", TTypeInfo(NTypeIds::Uint8)) }));
+        result.AddRow().Add<ui8>(1);
+        result.AddRow().Add<ui8>(0);
+
+        auto expected = result.BuildArrow();
+        UNIT_ASSERT_VALUES_EQUAL(batch->ToString(), expected->ToString());
+    }
+
+    Y_UNIT_TEST(JsonExists) {
+        JsonExistsImpl(false);
+    }
+
+    Y_UNIT_TEST(JsonExistsBinary) {
+        JsonExistsImpl(true);
+    }
+
     Y_UNIT_TEST(SimpleFunction) {
         TIndexInfo indexInfo = BuildTableInfo(testColumns, testKey);;
         TIndexColumnResolver columnResolver(indexInfo);
diff --git a/ydb/core/tx/program/registry.cpp b/ydb/core/tx/program/registry.cpp
index d7e31ff8c4..159dbc456f 100644
--- a/ydb/core/tx/program/registry.cpp
+++ b/ydb/core/tx/program/registry.cpp
@@ -7,7 +7,8 @@
 namespace NKikimr::NOlap {
 
 bool TKernelsRegistry::Parse(const TString& serialized) {
-    auto functionRegistry = NMiniKQL::CreateFunctionRegistry(NMiniKQL::CreateBuiltinRegistry());
+    auto functionRegistry = NMiniKQL::CreateFunctionRegistry(NMiniKQL::CreateBuiltinRegistry())->Clone();
+    NMiniKQL::FillStaticModules(*functionRegistry);
     auto nodeFactory = NMiniKQL::GetBuiltinFactory();
     auto kernels =  NYql::LoadKernels(serialized, *functionRegistry, nodeFactory);
     Kernels.swap(kernels);
diff --git a/ydb/library/yql/core/arrow_kernels/registry/ut/registry_ut.cpp b/ydb/library/yql/core/arrow_kernels/registry/ut/registry_ut.cpp
index 026878ecb9..797343859a 100644
--- a/ydb/library/yql/core/arrow_kernels/registry/ut/registry_ut.cpp
+++ b/ydb/library/yql/core/arrow_kernels/registry/ut/registry_ut.cpp
@@ -139,5 +139,28 @@ Y_UNIT_TEST_SUITE(TKernelRegistryTest) {
                 ctx.template MakeType<TDataExprType>(EDataSlot::Bool)));
             return b.JsonExists(blockJsonType, scalarUtf8Type, blockOptBoolType);
         });
+
+        TestOne([](auto& b,auto& ctx) {
+            auto blockOptJsonType = ctx.template MakeType<TBlockExprType>(
+                ctx.template MakeType<TOptionalExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::JsonDocument)));
+            auto scalarUtf8Type = ctx.template MakeType<TScalarExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::Utf8));
+            auto blockOptBoolType = ctx.template MakeType<TBlockExprType>(
+                ctx.template MakeType<TOptionalExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::Bool)));
+            return b.JsonExists(blockOptJsonType, scalarUtf8Type, blockOptBoolType);
+        });
+
+        TestOne([](auto& b,auto& ctx) {
+            auto blockJsonType = ctx.template MakeType<TBlockExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::JsonDocument));
+            auto scalarUtf8Type = ctx.template MakeType<TScalarExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::Utf8));
+            auto blockOptBoolType = ctx.template MakeType<TBlockExprType>(
+                ctx.template MakeType<TOptionalExprType>(
+                ctx.template MakeType<TDataExprType>(EDataSlot::Bool)));
+            return b.JsonExists(blockJsonType, scalarUtf8Type, blockOptBoolType);
+        });
     }
 }
diff --git a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.darwin-x86_64.txt b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.darwin-x86_64.txt
index 5877718c2a..ec3f3c6f6f 100644
--- a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.darwin-x86_64.txt
+++ b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.darwin-x86_64.txt
@@ -17,6 +17,8 @@ target_link_libraries(core-arrow_kernels-request PUBLIC
   library-yql-ast
   library-yql-minikql
   providers-common-mkql
+  library-yql-core
+  library-yql-sql
 )
 target_sources(core-arrow_kernels-request PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/library/yql/core/arrow_kernels/request/request.cpp
diff --git a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-aarch64.txt b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-aarch64.txt
index a38846dc53..9627c10463 100644
--- a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-aarch64.txt
+++ b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-aarch64.txt
@@ -18,6 +18,8 @@ target_link_libraries(core-arrow_kernels-request PUBLIC
   library-yql-ast
   library-yql-minikql
   providers-common-mkql
+  library-yql-core
+  library-yql-sql
 )
 target_sources(core-arrow_kernels-request PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/library/yql/core/arrow_kernels/request/request.cpp
diff --git a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-x86_64.txt b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-x86_64.txt
index a38846dc53..9627c10463 100644
--- a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-x86_64.txt
+++ b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.linux-x86_64.txt
@@ -18,6 +18,8 @@ target_link_libraries(core-arrow_kernels-request PUBLIC
   library-yql-ast
   library-yql-minikql
   providers-common-mkql
+  library-yql-core
+  library-yql-sql
 )
 target_sources(core-arrow_kernels-request PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/library/yql/core/arrow_kernels/request/request.cpp
diff --git a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.windows-x86_64.txt b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.windows-x86_64.txt
index 5877718c2a..ec3f3c6f6f 100644
--- a/ydb/library/yql/core/arrow_kernels/request/CMakeLists.windows-x86_64.txt
+++ b/ydb/library/yql/core/arrow_kernels/request/CMakeLists.windows-x86_64.txt
@@ -17,6 +17,8 @@ target_link_libraries(core-arrow_kernels-request PUBLIC
   library-yql-ast
   library-yql-minikql
   providers-common-mkql
+  library-yql-core
+  library-yql-sql
 )
 target_sources(core-arrow_kernels-request PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/library/yql/core/arrow_kernels/request/request.cpp
diff --git a/ydb/library/yql/core/arrow_kernels/request/request.cpp b/ydb/library/yql/core/arrow_kernels/request/request.cpp
index eb5c1c5c4e..8cd544e80f 100644
--- a/ydb/library/yql/core/arrow_kernels/request/request.cpp
+++ b/ydb/library/yql/core/arrow_kernels/request/request.cpp
@@ -2,6 +2,7 @@
 #include <ydb/library/yql/providers/common/mkql/yql_type_mkql.h>
 #include <ydb/library/yql/minikql/mkql_node_cast.h>
 #include <ydb/library/yql/minikql/mkql_node_serialization.h>
+#include <ydb/library/yql/core/yql_opt_utils.h>
 
 namespace NYql {
 
@@ -91,7 +92,10 @@ ui32 TKernelRequestBuilder::Udf(const TString& name, bool isPolymorphic, const s
 
 ui32 TKernelRequestBuilder::JsonExists(const TTypeAnnotationNode* arg1Type, const TTypeAnnotationNode* arg2Type, const TTypeAnnotationNode* retType) {
     TGuard<NKikimr::NMiniKQL::TScopedAlloc> allocGuard(Alloc_);
-    auto exists = Pb_.Udf("Json2.SqlExists");
+
+    bool isBinaryJson = (RemoveOptionalType(arg1Type->Cast<TBlockExprType>()->GetItemType())->Cast<TDataExprType>()->GetSlot() == EDataSlot::JsonDocument);
+
+    auto exists = Pb_.Udf(isBinaryJson ? "Json2.JsonDocumentSqlExists" : "Json2.SqlExists");
     auto parse = Pb_.Udf("Json2.Parse");
     auto compilePath = Pb_.Udf("Json2.CompilePath");
     auto outType = MakeType(retType);
@@ -100,7 +104,7 @@ ui32 TKernelRequestBuilder::JsonExists(const TTypeAnnotationNode* arg1Type, cons
     auto scalarApply = Pb_.ScalarApply({arg1, arg2}, [&](const auto& args) {
         auto json = args[0];
         auto processJson = [&](auto unpacked) {
-            auto input = Pb_.NewOptional(Pb_.Apply(parse, { unpacked }));
+            auto input = Pb_.NewOptional(isBinaryJson ? unpacked : Pb_.Apply(parse, { unpacked }));
             auto path = Pb_.Apply(compilePath, { args[1] });
             auto dictType = Pb_.NewDictType(Pb_.NewDataType(NUdf::EDataSlot::Utf8), Pb_.NewResourceType("JsonNode"), false);
             return Pb_.Apply(exists, { input, path, Pb_.NewDict(dictType, {}), Pb_.NewOptional(Pb_.NewDataLiteral(false)) });
diff --git a/ydb/library/yql/core/arrow_kernels/request/ya.make b/ydb/library/yql/core/arrow_kernels/request/ya.make
index 03d179915a..45b85ecf06 100644
--- a/ydb/library/yql/core/arrow_kernels/request/ya.make
+++ b/ydb/library/yql/core/arrow_kernels/request/ya.make
@@ -8,6 +8,8 @@ PEERDIR(
     ydb/library/yql/ast
     ydb/library/yql/minikql
     ydb/library/yql/providers/common/mkql
+    ydb/library/yql/core
+    ydb/library/yql/sql
 )
 
 YQL_LAST_ABI_VERSION()
diff --git a/ydb/library/yql/minikql/comp_nodes/mkql_block_impl.cpp b/ydb/library/yql/minikql/comp_nodes/mkql_block_impl.cpp
index 70ba311261..ed7637addf 100644
--- a/ydb/library/yql/minikql/comp_nodes/mkql_block_impl.cpp
+++ b/ydb/library/yql/minikql/comp_nodes/mkql_block_impl.cpp
@@ -69,11 +69,12 @@ arrow::Datum ConvertScalar(TType* type, const NUdf::TUnboxedValuePod& value, arr
         case NUdf::EDataSlot::String:
         case NUdf::EDataSlot::Utf8:
         case NUdf::EDataSlot::Yson:
-        case NUdf::EDataSlot::Json: {
+        case NUdf::EDataSlot::Json:
+        case NUdf::EDataSlot::JsonDocument: {
             const auto& str = value.AsStringRef();
             std::shared_ptr<arrow::Buffer> buffer(ARROW_RESULT(arrow::AllocateBuffer(str.Size(), &pool)));
             std::memcpy(buffer->mutable_data(), str.Data(), str.Size());
-            auto type = (slot == NUdf::EDataSlot::String || slot == NUdf::EDataSlot::Yson) ? arrow::binary() : arrow::utf8();
+            auto type = (slot == NUdf::EDataSlot::String || slot == NUdf::EDataSlot::Yson || slot == NUdf::EDataSlot::JsonDocument) ? arrow::binary() : arrow::utf8();
             std::shared_ptr<arrow::Scalar> scalar = std::make_shared<arrow::BinaryScalar>(buffer, type);
             return arrow::Datum(scalar);
         }
diff --git a/ydb/library/yql/minikql/mkql_type_builder.cpp b/ydb/library/yql/minikql/mkql_type_builder.cpp
index 45e13342cc..c6dd847f91 100644
--- a/ydb/library/yql/minikql/mkql_type_builder.cpp
+++ b/ydb/library/yql/minikql/mkql_type_builder.cpp
@@ -1478,6 +1478,7 @@ bool ConvertArrowType(NUdf::EDataSlot slot, std::shared_ptr<arrow::DataType>& ty
         return true;
     case NUdf::EDataSlot::String:
     case NUdf::EDataSlot::Yson:
+    case NUdf::EDataSlot::JsonDocument:
         type = arrow::binary();
         return true;
     case NUdf::EDataSlot::Utf8:
diff --git a/ydb/library/yql/public/udf/arrow/block_builder.h b/ydb/library/yql/public/udf/arrow/block_builder.h
index bf6e54bc89..a9d06df0c3 100644
--- a/ydb/library/yql/public/udf/arrow/block_builder.h
+++ b/ydb/library/yql/public/udf/arrow/block_builder.h
@@ -1121,6 +1121,7 @@ inline std::unique_ptr<TArrayBuilderBase> MakeArrayBuilderImpl(
             return std::make_unique<TFixedSizeArrayBuilder<double, Nullable>>(typeInfoHelper, type, pool, maxLen);
         case NUdf::EDataSlot::String:
         case NUdf::EDataSlot::Yson:
+        case NUdf::EDataSlot::JsonDocument:
             return std::make_unique<TStringArrayBuilder<arrow::BinaryType, Nullable>>(typeInfoHelper, type, pool, maxLen);
         case NUdf::EDataSlot::Utf8:
         case NUdf::EDataSlot::Json:
diff --git a/ydb/library/yql/public/udf/arrow/block_reader.h b/ydb/library/yql/public/udf/arrow/block_reader.h
index 5d0d1fa78e..a451486854 100644
--- a/ydb/library/yql/public/udf/arrow/block_reader.h
+++ b/ydb/library/yql/public/udf/arrow/block_reader.h
@@ -385,6 +385,7 @@ std::unique_ptr<typename TTraits::TResult> MakeBlockReaderImpl(const ITypeInfoHe
             return MakeFixedSizeBlockReaderImpl<TTraits, double>(isOptional);
         case NUdf::EDataSlot::String:
         case NUdf::EDataSlot::Yson:
+        case NUdf::EDataSlot::JsonDocument:
             return MakeStringBlockReaderImpl<TTraits, arrow::BinaryType>(isOptional);
         case NUdf::EDataSlot::Utf8:
         case NUdf::EDataSlot::Json: