Fix column order (ydb-platform#7839)

yumkam · Aug 16, 2024 · 277ed62 · 277ed62
1 parent af1afea
commit 277ed62
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 19 deletions.
diff --git a/ydb/library/yql/core/type_ann/type_ann_pg.cpp b/ydb/library/yql/core/type_ann/type_ann_pg.cpp
@@ -1823,7 +1823,7 @@ bool ScanColumns(TExprNode::TPtr root, TInputs& inputs, const THashSet<TString>&
                             }
                         }
 
-                        if (x.Order && x.Order->IsDuplicated(TString(node->Tail().Content()))) {
+                        if (x.Order && x.Order->IsDuplicatedIgnoreCase(TString(node->Tail().Content()))) {
                             ctx.Expr.AddError(TIssue(ctx.Expr.GetPosition(node->Pos()),
                                     TStringBuilder() << "Column reference is ambiguous: " << node->Tail().Content()));
                                 isError = true;

diff --git a/ydb/library/yql/core/ut/yql_column_order_ut.cpp b/ydb/library/yql/core/ut/yql_column_order_ut.cpp
@@ -9,12 +9,6 @@ using namespace NYql;
 
 Y_UNIT_TEST_SUITE(TYqlColumnOrder) {
 
-    Y_UNIT_TEST(ColumnOrderIgnoreCase) {
-        TColumnOrder order;
-        UNIT_ASSERT_EQUAL(order.AddColumn("a"), "a");
-        UNIT_ASSERT_EQUAL(order.AddColumn("A"), "A_generated_2");
-    }
-
     Y_UNIT_TEST(ColumnOrderShrink) {
         TColumnOrder order;
         UNIT_ASSERT_EQUAL(order.AddColumn("a"), "a");
@@ -32,6 +26,13 @@ Y_UNIT_TEST_SUITE(TYqlColumnOrder) {
         UNIT_ASSERT_EQUAL(order.AddColumn("a_generated_2_generated_2"), "a_generated_2_generated_2_generated_2");
     }
 
+    Y_UNIT_TEST(ColumnOrderCaseSensetive) {
+        TColumnOrder order;
+        UNIT_ASSERT_EQUAL(order.AddColumn("a"), "a");
+        UNIT_ASSERT_EQUAL(order.AddColumn("A"), "A");
+        UNIT_ASSERT_EQUAL(order.IsDuplicatedIgnoreCase("a"), true);
+    }
+
     Y_UNIT_TEST(ColumnOrderGeneratedMatchOverVectorCtor) {
         TColumnOrder order(TVector<TString>{"a", "a", "a_generated_2", "a_generated_2_generated_2"});
         TVector<TColumnOrder::TOrderedItem> got(order.begin(), order.end());

diff --git a/ydb/library/yql/core/yql_type_annotation.cpp b/ydb/library/yql/core/yql_type_annotation.cpp
@@ -124,6 +124,7 @@ TString TColumnOrder::Find(const TString& name) const {
 TColumnOrder& TColumnOrder::operator=(const TColumnOrder& rhs) {
     GeneratedToOriginal_ = rhs.GeneratedToOriginal_;
     Order_ = rhs.Order_;
+    UseCountLcase_ = rhs.UseCountLcase_;
     UseCount_ = rhs.UseCount_;
     return *this;
 }
@@ -137,32 +138,32 @@ TColumnOrder::TColumnOrder(const TVector<TString>& order) {
 
 TString TColumnOrder::AddColumn(const TString& name) {
     auto lcase = to_lower(name);
-    if (uint64_t count = ++UseCount_[lcase]; count > 1) {
+    ++UseCountLcase_[lcase];
+    if (uint64_t count = ++UseCount_[name]; count > 1) {
         TString generated = name + "_generated_" + ToString(count);
         GeneratedToOriginal_[generated] = name;
         Order_.emplace_back(name, generated);
-        ++UseCount_[to_lower(generated)];
+        ++UseCount_[generated];
         return generated;
     }
     Order_.emplace_back(name, name);
     GeneratedToOriginal_[name] = name;
     return name;
 }
 
-bool TColumnOrder::IsDuplicated(const TString& name) const {
-    auto it = UseCount_.find(to_lower(name));
-    return it != UseCount_.end() && it->second > 1;
+bool TColumnOrder::IsDuplicatedIgnoreCase(const TString& name) const {
+    auto it = UseCountLcase_.find(to_lower(name));
+    return it != UseCountLcase_.end() && it->second > 1;
 }
 
 void TColumnOrder::Shrink(size_t remain) {
     for (size_t i = remain; i < Order_.size(); ++i) {
-        auto logicalLcase = to_lower(Order_[i].LogicalName);
-        if (!--UseCount_[logicalLcase]) {
-            UseCount_.erase(logicalLcase);
+        --UseCountLcase_[to_lower(Order_[i].LogicalName)];
+        if (!--UseCount_[Order_[i].LogicalName]) {
+            UseCount_.erase(Order_[i].LogicalName);
         }
-        auto physicalLcase = to_lower(Order_[i].PhysicalName);
-        if (!--UseCount_[physicalLcase]) {
-            UseCount_.erase(physicalLcase);
+        if (!--UseCount_[Order_[i].PhysicalName]) {
+            UseCount_.erase(Order_[i].PhysicalName);
         }
         GeneratedToOriginal_.erase(Order_[i].PhysicalName);
     }
@@ -179,6 +180,7 @@ void TColumnOrder::Clear() {
     Order_.clear();
     GeneratedToOriginal_.clear();
     UseCount_.clear();
+    UseCountLcase_.clear();
 }
 
 void TColumnOrder::EraseIf(const std::function<bool(const TString&)>& fn) {

diff --git a/ydb/library/yql/core/yql_type_annotation.h b/ydb/library/yql/core/yql_type_annotation.h
@@ -170,7 +170,7 @@ class TColumnOrder {
     explicit TColumnOrder(const TVector<TString>& order);
     TString AddColumn(const TString& name);
 
-    bool IsDuplicated(const TString& name) const;
+    bool IsDuplicatedIgnoreCase(const TString& name) const;
 
     void Shrink(size_t remain);
 
@@ -213,6 +213,7 @@ class TColumnOrder {
 private:
     THashMap<TString, TString> GeneratedToOriginal_;
     THashMap<TString, uint64_t> UseCount_;
+    THashMap<TString, uint64_t> UseCountLcase_;
     // (name, generated_name)
     TVector<TOrderedItem> Order_;
 };