Correct index construction

ydb/core/formats/arrow/scalar/serialization.cpp

@@ -0,0 +1,47 @@
+#include "serialization.h"
+#include <ydb/core/formats/arrow/switch/switch_type.h>
+#include <ydb/library/actors/core/log.h>
+
+namespace NKikimr::NArrow::NScalar {
+
+TConclusion<TString> TSerializer::SerializePayloadToString(const std::shared_ptr<arrow::Scalar>& scalar) {
+    TString resultString;
+    const bool resultFlag = NArrow::SwitchType(scalar->type->id(), [&](const auto& type) {
+        using TWrap = std::decay_t<decltype(type)>;
+        if constexpr (arrow::has_c_type<typename TWrap::T>()) {
+            using CType = typename TWrap::T::c_type;
+            using ScalarType = typename arrow::TypeTraits<typename TWrap::T>::ScalarType;
+            const ScalarType* scalarTyped = static_cast<const ScalarType*>(scalar.get());
+            resultString = TString(sizeof(CType), '\0');
+            memcpy(&resultString[0], scalarTyped->data(), sizeof(CType));
+            return true;
+        }
+        return false;
+    });
+    if (!resultFlag) {
+        return TConclusionStatus::Fail("incorrect scalar type for payload serialization: " + scalar->type->ToString());
+    }
+    return resultString;
+}
+
+TConclusion<std::shared_ptr<arrow::Scalar>> TSerializer::DeserializeFromStringWithPayload(const TString& data, const std::shared_ptr<arrow::DataType>& dataType) {
+    AFL_VERIFY(dataType);
+    std::shared_ptr<arrow::Scalar> result;
+    const bool resultFlag = NArrow::SwitchType(dataType->id(), [&](const auto& type) {
+        using TWrap = std::decay_t<decltype(type)>;
+        if constexpr (arrow::has_c_type<typename TWrap::T>()) {
+            using CType = typename TWrap::T::c_type;
+            AFL_VERIFY(data.size() == sizeof(CType));
+            using ScalarType = typename arrow::TypeTraits<typename TWrap::T>::ScalarType;
+            result = std::make_shared<ScalarType>(*(CType*)&data[0], dataType);
+            return true;
+        }
+        return false;
+    });
+    if (!resultFlag) {
+        return TConclusionStatus::Fail("incorrect scalar type for payload deserialization: " + dataType->ToString());
+    }
+    return result;
+}
+
+}

ydb/core/formats/arrow/scalar/serialization.h

@@ -0,0 +1,15 @@
+#pragma once
+#include <ydb/library/conclusion/result.h>
+
+#include <contrib/libs/apache/arrow/cpp/src/arrow/scalar.h>
+#include <contrib/libs/apache/arrow/cpp/src/arrow/type.h>
+
+#include <util/generic/string.h>
+
+namespace NKikimr::NArrow::NScalar {
+class TSerializer {
+public:
+    static TConclusion<TString> SerializePayloadToString(const std::shared_ptr<arrow::Scalar>& scalar);
+    static TConclusion<std::shared_ptr<arrow::Scalar>> DeserializeFromStringWithPayload(const TString& data, const std::shared_ptr<arrow::DataType>& dataType);
+};
+}

ydb/core/formats/arrow/scalar/ya.make

@@ -0,0 +1,14 @@
+LIBRARY()
+
+PEERDIR(
+    contrib/libs/apache/arrow
+    ydb/library/conclusion
+    ydb/core/formats/arrow/switch
+    ydb/library/actors/core
+)
+
+SRCS(
+    serialization.cpp
+)
+
+END()

ydb/core/formats/arrow/ya.make

@@ -12,6 +12,7 @@ PEERDIR(
     ydb/core/formats/arrow/dictionary
     ydb/core/formats/arrow/transformer
     ydb/core/formats/arrow/reader
+    ydb/core/formats/arrow/scalar
     ydb/core/formats/arrow/hash
     ydb/library/actors/core
     ydb/library/arrow_kernels

ydb/core/kqp/gateway/behaviour/tablestore/operations/alter_sharding.cpp

@@ -1,4 +1,5 @@
 #include "alter_sharding.h"
+#include <ydb/library/actors/core/log.h>
 #include <util/string/type.h>
 
 namespace NKikimr::NKqp {
@@ -26,4 +27,8 @@ void TAlterShardingOperation::DoSerializeScheme(NKikimrSchemeOp::TModifyScheme&
     scheme.MutableAlterColumnTable()->MutableReshardColumnTable()->SetIncrease(*Increase);
 }
 
+void TAlterShardingOperation::DoSerializeScheme(NKikimrSchemeOp::TAlterColumnTableSchema& /*scheme*/) const {
+    AFL_VERIFY(false);
+}
+
 }

ydb/core/kqp/gateway/behaviour/tablestore/operations/alter_sharding.h

@@ -1,5 +1,4 @@
 #include "abstract.h"
-#include <ydb/core/tx/columnshard/engines/scheme/statistics/abstract/constructor.h>
 
 namespace NKikimr::NKqp {
 
@@ -12,9 +11,7 @@ class TAlterShardingOperation: public ITableStoreOperation {
     static inline const auto Registrator = TFactory::TRegistrator<TAlterShardingOperation>(GetTypeName());
 private:
     std::optional<bool> Increase;
-    virtual void DoSerializeScheme(NKikimrSchemeOp::TAlterColumnTableSchema& /*scheme*/) const override {
-        AFL_VERIFY(false);
-    }
+    virtual void DoSerializeScheme(NKikimrSchemeOp::TAlterColumnTableSchema& /*scheme*/) const override;
     virtual void DoSerializeScheme(NKikimrSchemeOp::TModifyScheme& scheme, const bool isStandalone) const override;
 
 public:

ydb/core/kqp/gateway/behaviour/tablestore/operations/upsert_index.cpp

@@ -12,10 +12,6 @@ TConclusionStatus TUpsertIndexOperation::DoDeserialize(NYql::TObjectSettingsImpl
         }
         IndexName = *fValue;
     }
-    StorageId = features.Extract("STORAGE_ID");
-    if (StorageId && !*StorageId) {
-        return TConclusionStatus::Fail("STORAGE_ID cannot be empty string");
-    }
     TString indexType;
     {
         auto fValue = features.Extract("TYPE");
@@ -46,9 +42,6 @@ TConclusionStatus TUpsertIndexOperation::DoDeserialize(NYql::TObjectSettingsImpl
 
 void TUpsertIndexOperation::DoSerializeScheme(NKikimrSchemeOp::TAlterColumnTableSchema& schemaData) const {
     auto* indexProto = schemaData.AddUpsertIndexes();
-    if (StorageId) {
-        indexProto->SetStorageId(*StorageId);
-    }
     indexProto->SetName(IndexName);
     IndexMetaConstructor.SerializeToProto(*indexProto);
 }

ydb/core/kqp/gateway/behaviour/tablestore/operations/upsert_index.h

@@ -12,7 +12,6 @@ class TUpsertIndexOperation : public ITableStoreOperation {
     static inline auto Registrator = TFactory::TRegistrator<TUpsertIndexOperation>(GetTypeName());
 private:
     TString IndexName;
-    std::optional<TString> StorageId;
     NBackgroundTasks::TInterfaceProtoContainer<NOlap::NIndexes::IIndexMetaConstructor> IndexMetaConstructor;
 public:
     TConclusionStatus DoDeserialize(NYql::TObjectSettingsImpl::TFeaturesExtractor& features) override;

ydb/core/kqp/gateway/behaviour/tablestore/operations/ya.make

@@ -7,16 +7,13 @@ SRCS(
     GLOBAL drop_column.cpp
     GLOBAL upsert_index.cpp
     GLOBAL drop_index.cpp
-    GLOBAL upsert_stat.cpp
-    GLOBAL drop_stat.cpp
     GLOBAL upsert_opt.cpp
     GLOBAL alter_sharding.cpp
 )
 
 PEERDIR(
     ydb/services/metadata/manager
     ydb/core/formats/arrow/serializer
-    ydb/core/tx/columnshard/engines/scheme/statistics/abstract
     ydb/core/tx/columnshard/engines/storage/optimizer/abstract
     ydb/core/kqp/gateway/utils
     ydb/core/protos

ydb/core/kqp/ut/olap/helpers/typed_local.cpp

@@ -143,16 +143,16 @@ void TTypedLocalHelper::FillPKOnly(const double pkKff /*= 0*/, const ui32 numRow
     TBase::SendDataViaActorSystem(TablePath, batch);
 }
 
-void TTypedLocalHelper::GetStats(std::vector<NKikimrColumnShardStatisticsProto::TPortionStorage>& stats, const bool verbose /*= false*/) {
+void TTypedLocalHelper::GetStats(std::vector<NJson::TJsonValue>& stats, const bool verbose /*= false*/) {
     TString selectQuery = "SELECT * FROM `" + TablePath + "/.sys/primary_index_portion_stats` WHERE Activity = true";
     auto tableClient = KikimrRunner.GetTableClient();
     auto rows = ExecuteScanQuery(tableClient, selectQuery, verbose);
     for (auto&& r : rows) {
         for (auto&& c : r) {
             if (c.first == "Stats") {
-                NKikimrColumnShardStatisticsProto::TPortionStorage store;
-                AFL_VERIFY(google::protobuf::TextFormat::ParseFromString(GetUtf8(c.second), &store));
-                stats.emplace_back(store);
+                NJson::TJsonValue jsonStore;
+                AFL_VERIFY(NJson::ReadJsonFastTree(GetUtf8(c.second), &jsonStore));
+                stats.emplace_back(jsonStore);
             }
         }
     }

ydb/core/kqp/ut/olap/helpers/typed_local.h

@@ -1,11 +1,14 @@
 #pragma once
 #include <ydb/core/testlib/cs_helper.h>
 #include <ydb/core/kqp/ut/common/kqp_ut_common.h>
-#include <ydb/public/sdk/cpp/client/ydb_types/status_codes.h>
 #include <ydb/core/formats/arrow/simple_builder/array.h>
 #include <ydb/core/formats/arrow/simple_builder/batch.h>
 #include <ydb/core/formats/arrow/simple_builder/filler.h>
 
+#include <ydb/public/sdk/cpp/client/ydb_types/status_codes.h>
+
+#include <library/cpp/json/writer/json_value.h>
+
 namespace NKikimr::NKqp {
 
 class TTypedLocalHelper: public Tests::NCS::THelper {
@@ -19,7 +22,7 @@ class TTypedLocalHelper: public Tests::NCS::THelper {
 protected:
     virtual TString GetTestTableSchema() const override;
     virtual std::vector<TString> GetShardingColumns() const override {
-        return {"pk_int"};
+        return { "pk_int" };
     }
 public:
     TTypedLocalHelper(const TString& typeName, TKikimrRunner& kikimrRunner, const TString& tableName = "olapTable", const TString& storeName = "olapStore")
@@ -66,7 +69,7 @@ class TTypedLocalHelper: public Tests::NCS::THelper {
 
     void GetVolumes(ui64& rawBytes, ui64& bytes, const bool verbose = false, const std::vector<TString> columnNames = {});
 
-    void GetStats(std::vector<NKikimrColumnShardStatisticsProto::TPortionStorage>& stats, const bool verbose = false);
+    void GetStats(std::vector<NJson::TJsonValue>& stats, const bool verbose = false);
 
     void GetCount(ui64& count);