fix column name index

Author: swalrus1

ydb/core/tx/columnshard/engines/scheme/index_info.cpp

@@ -37,7 +37,8 @@ std::optional<ui32> TIndexInfo::GetColumnIdOptional(const std::string& name) con
 
 std::optional<ui32> TIndexInfo::GetColumnIndexOptional(const std::string& name) const {
     auto it = std::lower_bound(ColumnIdxSortedByName.begin(), ColumnIdxSortedByName.end(), name, [this](const ui32 idx, const std::string name) {
-        return SchemaWithSpecials->GetFieldByIndexVerified(idx)->name() < name;
+        AFL_VERIFY(idx < SchemaColumnIdsWithSpecials.size());
+        return GetColumnName(SchemaColumnIdsWithSpecials[idx]) < name;
     });
     if (it != ColumnIdxSortedByName.end() && SchemaWithSpecials->GetFieldByIndexVerified(*it)->name() == name) {
         return *it;
@@ -244,18 +245,20 @@ bool TIndexInfo::DeserializeFromProto(const NKikimrSchemeOp::TColumnTableSchema&
     AFL_VERIFY(PKColumnIds.empty());
     {
         TMemoryProfileGuard g("TIndexInfo::DeserializeFromProto::Columns");
+        THashMap<TString, ui32> columnIds;
         for (const auto& col : schema.GetColumns()) {
             auto tableCol = BuildColumnFromProto(col, cache);
             auto id = tableCol.Id;
+            AFL_VERIFY(columnIds.emplace(tableCol.Name, id).second);
             AFL_VERIFY(columns.emplace(id, std::move(tableCol)).second);
         }
-        BuildColumnIndexByName();
         for (const auto& keyName : schema.GetKeyColumnNames()) {
-            const ui32 columnId = GetColumnIdVerified(keyName);
-            auto it = columns.find(columnId);
+            const ui32* findColumnId = columnIds.FindPtr(keyName);
+            AFL_VERIFY(findColumnId);
+            auto it = columns.find(*findColumnId);
             AFL_VERIFY(it != columns.end());
             it->second.KeyOrder = PKColumnIds.size();
-            PKColumnIds.push_back(columnId);
+            PKColumnIds.push_back(*findColumnId);
         }
     }
     InitializeCaches(operators, columns, cache, false);
@@ -287,6 +290,7 @@ bool TIndexInfo::DeserializeFromProto(const NKikimrSchemeOp::TColumnTableSchema&
     }
 
     Version = schema.GetVersion();
+    BuildColumnIndexByName();
     Precalculate();
     Validate();
     return true;
@@ -562,6 +566,21 @@ void TIndexInfo::Precalculate() {
     }
 }
 
+void TIndexInfo::BuildColumnIndexByName() {
+    const ui32 columnCount = SchemaColumnIdsWithSpecials.size() - SpecialColumnsCount;
+    std::erase_if(ColumnIdxSortedByName, [columnCount](const ui32 idx) {
+        return idx >= columnCount;
+    });
+    ColumnIdxSortedByName.reserve(columnCount);
+    for (ui32 i = 0; i < columnCount; ++i) {
+        ColumnIdxSortedByName.push_back(i);
+    }
+
+    std::sort(ColumnIdxSortedByName.begin(), ColumnIdxSortedByName.end(), [this](const ui32 lhs, const ui32 rhs) {
+        return CompareColumnIdxByName(lhs, rhs);
+    });
+}
+
 void TIndexInfo::Validate() const {
     AFL_VERIFY(!!UsedStorageIds);
     AFL_VERIFY(ColumnFeatures.size() == SchemaColumnIdsWithSpecials.size());
@@ -575,11 +594,12 @@ void TIndexInfo::Validate() const {
             ++idx;
         }
     }
+    AFL_VERIFY(std::is_sorted(SchemaColumnIdsWithSpecials.begin(), SchemaColumnIdsWithSpecials.end()));
 
     AFL_VERIFY(ColumnFeatures.size() == ColumnIdxSortedByName.size() + SpecialColumnsCount);
-    std::is_sorted(ColumnIdxSortedByName.begin(), ColumnIdxSortedByName.end(), [this](const ui32 lhs, const ui32 rhs) {
-        return GetColumnName(lhs) < GetColumnName(rhs);
-    });
+    AFL_VERIFY(std::is_sorted(ColumnIdxSortedByName.begin(), ColumnIdxSortedByName.end(), [this](const ui32 lhs, const ui32 rhs) {
+        return CompareColumnIdxByName(lhs, rhs);
+    }));
 
     {
         ui32 pkIdx = 0;

ydb/core/tx/columnshard/engines/scheme/index_info.h

@@ -111,20 +111,7 @@ struct TIndexInfo: public IIndexInfo {
 
     void Validate() const;
     void Precalculate();
-    void BuildColumnIndexByName() {
-        const ui32 columnCount = SchemaColumnIdsWithSpecials.size() - SpecialColumnsCount;
-        std::erase_if(ColumnIdxSortedByName, [columnCount](const ui32 idx) {
-            return idx >= columnCount;
-        });
-        ColumnIdxSortedByName.reserve(columnCount);
-        for (ui32 i = 0; i < columnCount; ++i) {
-            ColumnIdxSortedByName.push_back(i);
-        }
-
-        std::sort(ColumnIdxSortedByName.begin(), ColumnIdxSortedByName.end(), [this](const ui32 lhs, const ui32 rhs) {
-            return GetColumnName(lhs) < GetColumnName(rhs);
-        });
-    }
+    void BuildColumnIndexByName();
 
     bool DeserializeFromProto(const NKikimrSchemeOp::TColumnTableSchema& schema, const std::shared_ptr<IStoragesManager>& operators,
         const std::shared_ptr<TSchemaObjectsCache>& cache);
@@ -159,6 +146,12 @@ struct TIndexInfo: public IIndexInfo {
 
     void SetAllKeys(const std::shared_ptr<IStoragesManager>& operators, const THashMap<ui32, NTable::TColumn>& columns);
 
+    bool CompareColumnIdxByName(const ui32 lhs, const ui32 rhs) const {
+        AFL_VERIFY(lhs < SchemaColumnIdsWithSpecials.size());
+        AFL_VERIFY(rhs < SchemaColumnIdsWithSpecials.size());
+        return GetColumnName(SchemaColumnIdsWithSpecials[lhs]) < GetColumnName(SchemaColumnIdsWithSpecials[rhs]);
+    }
+
 public:
     NSplitter::TEntityGroups GetEntityGroupsByStorageId(const TString& specialTier, const IStoragesManager& storages) const;
     std::optional<ui32> GetPKColumnIndexByIndexVerified(const ui32 columnIndex) const {