primary keys naming unification

8 files changed, 26 insertions, 47 deletions

diff --git a/ydb/core/tx/columnshard/columnshard__scan.cpp b/ydb/core/tx/columnshard/columnshard__scan.cpp
index d3e7138b5a..c261b70dfa 100644
--- a/ydb/core/tx/columnshard/columnshard__scan.cpp
+++ b/ydb/core/tx/columnshard/columnshard__scan.cpp
@@ -749,7 +749,7 @@ bool TTxScan::Execute(TTransactionContext& txc, const TActorContext& /*ctx*/) {
 
     auto ydbKey = isIndexStats ?
         NOlap::GetColumns(PrimaryIndexStatsSchema, PrimaryIndexStatsSchema.KeyColumns) :
-        indexInfo->GetPrimaryKey();
+        indexInfo->GetPrimaryKeyColumns();
 
     for (auto& range: record.GetRanges()) {
         if (!FillPredicatesFromRange(read, range, ydbKey, Self->TabletID(), isIndexStats ? nullptr : indexInfo, ErrorDescription)) {
diff --git a/ydb/core/tx/columnshard/engines/column_engine.h b/ydb/core/tx/columnshard/engines/column_engine.h
index 7697e28fb0..d823af9629 100644
--- a/ydb/core/tx/columnshard/engines/column_engine.h
+++ b/ydb/core/tx/columnshard/engines/column_engine.h
@@ -288,7 +288,7 @@ public:
 
 class TVersionedIndex {
     std::map<TSnapshot, ISnapshotSchema::TPtr> Snapshots;
-    std::shared_ptr<arrow::Schema> IndexKey;
+    std::shared_ptr<arrow::Schema> PrimaryKey;
     std::map<ui64, ISnapshotSchema::TPtr> SnapshotByVersion;
     ui64 LastSchemaVersion = 0;
 public:
@@ -327,15 +327,15 @@ public:
         return Snapshots.rbegin()->second;
     }
 
-    const std::shared_ptr<arrow::Schema>& GetIndexKey() const noexcept {
-        return IndexKey;
+    const std::shared_ptr<arrow::Schema>& GetPrimaryKey() const noexcept {
+        return PrimaryKey;
     }
 
     void AddIndex(const TSnapshot& snapshot, TIndexInfo&& indexInfo) {
         if (Snapshots.empty()) {
-            IndexKey = indexInfo.GetIndexKey();
+            PrimaryKey = indexInfo.GetPrimaryKey();
         } else {
-            Y_ABORT_UNLESS(IndexKey->Equals(indexInfo.GetIndexKey()));
+            Y_ABORT_UNLESS(PrimaryKey->Equals(indexInfo.GetPrimaryKey()));
         }
 
         auto newVersion = indexInfo.GetVersion();
@@ -358,8 +358,6 @@ public:
 
     virtual const TVersionedIndex& GetVersionedIndex() const = 0;
     virtual const std::shared_ptr<arrow::Schema>& GetReplaceKey() const { return GetVersionedIndex().GetLastSchema()->GetIndexInfo().GetReplaceKey(); }
-    virtual const std::shared_ptr<arrow::Schema>& GetSortingKey() const { return GetVersionedIndex().GetLastSchema()->GetIndexInfo().GetSortingKey(); }
-    virtual const std::shared_ptr<arrow::Schema>& GetIndexKey() const { return GetVersionedIndex().GetLastSchema()->GetIndexInfo().GetIndexKey(); }
 
     virtual bool HasDataInPathId(const ui64 pathId) const = 0;
     virtual bool Load(IDbWrapper& db) = 0;
diff --git a/ydb/core/tx/columnshard/engines/predicate/range.cpp b/ydb/core/tx/columnshard/engines/predicate/range.cpp
index adebf3860b..53e10174b8 100644
--- a/ydb/core/tx/columnshard/engines/predicate/range.cpp
+++ b/ydb/core/tx/columnshard/engines/predicate/range.cpp
@@ -40,7 +40,7 @@ NKikimr::NArrow::TColumnFilter TPKRangeFilter::BuildFilter(const arrow::Datum& d
 }
 
 bool TPKRangeFilter::IsPortionInUsage(const TPortionInfo& info, const TIndexInfo& indexInfo) const {
-    if (auto from = PredicateFrom.ExtractKey(indexInfo.GetIndexKey())) {
+    if (auto from = PredicateFrom.ExtractKey(indexInfo.GetPrimaryKey())) {
         const auto& portionEnd = info.IndexKeyEnd();
         const int commonSize = std::min(from->Size(), portionEnd.Size());
         if (std::is_gt(from->ComparePartNotNull(portionEnd, commonSize))) {
@@ -48,7 +48,7 @@ bool TPKRangeFilter::IsPortionInUsage(const TPortionInfo& info, const TIndexInfo
         }
     }
 
-    if (auto to = PredicateTo.ExtractKey(indexInfo.GetIndexKey())) {
+    if (auto to = PredicateTo.ExtractKey(indexInfo.GetPrimaryKey())) {
         const auto& portionStart = info.IndexKeyStart();
         const int commonSize = std::min(to->Size(), portionStart.Size());
         if (std::is_lt(to->ComparePartNotNull(portionStart, commonSize))) {
diff --git a/ydb/core/tx/columnshard/engines/reader/read_metadata.cpp b/ydb/core/tx/columnshard/engines/reader/read_metadata.cpp
index 83d0ded5f5..729a50ae7a 100644
--- a/ydb/core/tx/columnshard/engines/reader/read_metadata.cpp
+++ b/ydb/core/tx/columnshard/engines/reader/read_metadata.cpp
@@ -69,7 +69,7 @@ std::set<ui32> TReadMetadata::GetEarlyFilterColumnIds() const {
 std::set<ui32> TReadMetadata::GetPKColumnIds() const {
     std::set<ui32> result;
     auto& indexInfo = ResultIndexSchema->GetIndexInfo();
-    for (auto&& i : indexInfo.GetPrimaryKey()) {
+    for (auto&& i : indexInfo.GetPrimaryKeyColumns()) {
         Y_ABORT_UNLESS(result.emplace(indexInfo.GetColumnId(i.first)).second);
     }
     return result;
diff --git a/ydb/core/tx/columnshard/engines/reader/read_metadata.h b/ydb/core/tx/columnshard/engines/reader/read_metadata.h
index 01ca8575e4..1d26b3399e 100644
--- a/ydb/core/tx/columnshard/engines/reader/read_metadata.h
+++ b/ydb/core/tx/columnshard/engines/reader/read_metadata.h
@@ -214,16 +214,12 @@ public:
         return SelectInfo->PortionsOrderedPK.empty() && CommittedBlobs.empty();
     }
 
-    std::shared_ptr<arrow::Schema> GetSortingKey() const {
-        return ResultIndexSchema->GetIndexInfo().GetSortingKey();
-    }
-
     std::shared_ptr<arrow::Schema> GetReplaceKey() const {
         return ResultIndexSchema->GetIndexInfo().GetReplaceKey();
     }
 
     std::vector<TNameTypeInfo> GetKeyYqlSchema() const override {
-        return ResultIndexSchema->GetIndexInfo().GetPrimaryKey();
+        return ResultIndexSchema->GetIndexInfo().GetPrimaryKeyColumns();
     }
 
     size_t NumIndexedChunks() const {
diff --git a/ydb/core/tx/columnshard/engines/scheme/abstract_scheme.cpp b/ydb/core/tx/columnshard/engines/scheme/abstract_scheme.cpp
index 7bcfff8a55..f138fc6b5f 100644
--- a/ydb/core/tx/columnshard/engines/scheme/abstract_scheme.cpp
+++ b/ydb/core/tx/columnshard/engines/scheme/abstract_scheme.cpp
@@ -78,7 +78,7 @@ std::shared_ptr<arrow::RecordBatch> ISnapshotSchema::PrepareForInsert(const TStr
         return nullptr;
     }
 
-    const auto& sortingKey = GetIndexInfo().GetSortingKey();
+    const auto& sortingKey = GetIndexInfo().GetPrimaryKey();
     Y_ABORT_UNLESS(sortingKey);
 
     // Check PK is NOT NULL
diff --git a/ydb/core/tx/columnshard/engines/scheme/index_info.cpp b/ydb/core/tx/columnshard/engines/scheme/index_info.cpp
index b438621663..82eeb52686 100644
--- a/ydb/core/tx/columnshard/engines/scheme/index_info.cpp
+++ b/ydb/core/tx/columnshard/engines/scheme/index_info.cpp
@@ -29,10 +29,7 @@ TIndexInfo::TIndexInfo(const TString& name, ui32 id)
 {}
 
 bool TIndexInfo::CheckCompatible(const TIndexInfo& other) const {
-    if (!other.GetReplaceKey()->Equals(GetReplaceKey())) {
-        return false;
-    }
-    if (!other.GetIndexKey()->Equals(GetIndexKey())) {
+    if (!other.GetPrimaryKey()->Equals(GetPrimaryKey())) {
         return false;
     }
     return true;
@@ -240,19 +237,14 @@ void TIndexInfo::SetAllKeys() {
     /// @note Setting replace and sorting key to PK we are able to:
     /// * apply REPLACE by MergeSort
     /// * apply PK predicate before REPLACE
-    const auto& primaryKeyNames = NamesOnly(GetPrimaryKey());
+    const auto& primaryKeyNames = NamesOnly(GetPrimaryKeyColumns());
     // Update set of required columns with names from primary key.
     for (const auto& name: primaryKeyNames) {
         RequiredColumns.insert(name);
     }
-
-    std::vector<std::shared_ptr<arrow::Field>> fields;
-    if (primaryKeyNames.size()) {
-        SortingKey = ArrowSchema(primaryKeyNames);
-        ReplaceKey = SortingKey;
-        fields = ReplaceKey->fields();
-        IndexKey = ReplaceKey;
-    }
+    AFL_VERIFY(primaryKeyNames.size());
+    PrimaryKey = ArrowSchema(primaryKeyNames);
+    std::vector<std::shared_ptr<arrow::Field>> fields = PrimaryKey->fields();
 
     fields.push_back(arrow::field(SPEC_COL_PLAN_STEP, arrow::uint64()));
     fields.push_back(arrow::field(SPEC_COL_TX_ID, arrow::uint64()));
@@ -271,8 +263,8 @@ void TIndexInfo::SetAllKeys() {
 }
 
 std::shared_ptr<NArrow::TSortDescription> TIndexInfo::SortDescription() const {
-    if (GetSortingKey()) {
-        auto key = GetExtendedKey(); // Sort with extended key, greater snapshot first
+    if (GetPrimaryKey()) {
+        auto key = ExtendedKey; // Sort with extended key, greater snapshot first
         Y_ABORT_UNLESS(key && key->num_fields() > 2);
         auto description = std::make_shared<NArrow::TSortDescription>(key);
         description->Directions[key->num_fields() - 1] = -1;
@@ -284,10 +276,10 @@ std::shared_ptr<NArrow::TSortDescription> TIndexInfo::SortDescription() const {
 }
 
 std::shared_ptr<NArrow::TSortDescription> TIndexInfo::SortReplaceDescription() const {
-    if (GetSortingKey()) {
-        auto key = GetExtendedKey(); // Sort with extended key, greater snapshot first
+    if (GetPrimaryKey()) {
+        auto key = ExtendedKey; // Sort with extended key, greater snapshot first
         Y_ABORT_UNLESS(key && key->num_fields() > 2);
-        auto description = std::make_shared<NArrow::TSortDescription>(key, GetReplaceKey());
+        auto description = std::make_shared<NArrow::TSortDescription>(key, GetPrimaryKey());
         description->Directions[key->num_fields() - 1] = -1;
         description->Directions[key->num_fields() - 2] = -1;
         description->NotNull = true; // TODO
diff --git a/ydb/core/tx/columnshard/engines/scheme/index_info.h b/ydb/core/tx/columnshard/engines/scheme/index_info.h
index 4d27e29ebe..b73bf480c1 100644
--- a/ydb/core/tx/columnshard/engines/scheme/index_info.h
+++ b/ydb/core/tx/columnshard/engines/scheme/index_info.h
@@ -130,7 +130,7 @@ public:
     std::vector<TNameTypeInfo> GetColumns(const std::vector<ui32>& ids) const;
 
     /// Traditional Primary Key (includes uniqueness, search and sorting logic)
-    std::vector<TNameTypeInfo> GetPrimaryKey() const {
+    std::vector<TNameTypeInfo> GetPrimaryKeyColumns() const {
         return GetColumns(KeyColumns);
     }
 
@@ -143,10 +143,8 @@ public:
     // Sorting key: could be less or greater then traditional PK
     // It could be empty for append-only tables. It could be greater then PK for better columns compression.
     // If sorting key includes uniqueness key as a prefix we are able to use MergeSort for REPLACE.
-    const std::shared_ptr<arrow::Schema>& GetSortingKey() const { return SortingKey; }
-    const std::shared_ptr<arrow::Schema>& GetReplaceKey() const { return ReplaceKey; }
-    const std::shared_ptr<arrow::Schema>& GetExtendedKey() const { return ExtendedKey; }
-    const std::shared_ptr<arrow::Schema>& GetIndexKey() const { return IndexKey; }
+    const std::shared_ptr<arrow::Schema>& GetReplaceKey() const { return PrimaryKey; }
+    const std::shared_ptr<arrow::Schema>& GetPrimaryKey() const { return PrimaryKey; }
 
     /// Initializes sorting, replace, index and extended keys.
     void SetAllKeys();
@@ -180,12 +178,9 @@ public:
     bool AllowTtlOverColumn(const TString& name) const;
 
     /// Returns whether the sorting keys defined.
-    bool IsSorted() const { return SortingKey.get(); }
+    bool IsSorted() const { return true; }
     bool IsSortedColumn(const ui32 columnId) const { return GetPKFirstColumnId() == columnId; }
 
-     /// Returns whether the replace keys defined.
-    bool IsReplacing() const { return ReplaceKey.get(); }
-
     std::shared_ptr<NArrow::TSortDescription> SortDescription() const;
     std::shared_ptr<NArrow::TSortDescription> SortReplaceDescription() const;
 
@@ -216,10 +211,8 @@ private:
     TString Name;
     std::shared_ptr<arrow::Schema> Schema;
     std::shared_ptr<arrow::Schema> SchemaWithSpecials;
-    std::shared_ptr<arrow::Schema> SortingKey;
-    std::shared_ptr<arrow::Schema> ReplaceKey;
+    std::shared_ptr<arrow::Schema> PrimaryKey;
     std::shared_ptr<arrow::Schema> ExtendedKey; // Extend PK with snapshot columns to allow old shapshot reads
-    std::shared_ptr<arrow::Schema> IndexKey;
     THashSet<TString> RequiredColumns;
     THashSet<ui32> MinMaxIdxColumnsIds;
     std::optional<NArrow::TCompression> DefaultCompression;