Temporarily disable binary String data type for LIKE pushdown to Column Shards.

4 files changed, 64 insertions, 9 deletions

diff --git a/ydb/core/kqp/opt/physical/kqp_opt_phy_olap_filter_collection.cpp b/ydb/core/kqp/opt/physical/kqp_opt_phy_olap_filter_collection.cpp
index 2719be84242..b9312668a0a 100644
--- a/ydb/core/kqp/opt/physical/kqp_opt_phy_olap_filter_collection.cpp
+++ b/ydb/core/kqp/opt/physical/kqp_opt_phy_olap_filter_collection.cpp
@@ -13,6 +13,33 @@ using namespace NYql::NNodes;
 
 namespace {
 
+bool ColumnHasBinaryStringType(const TExprBase& expr) {
+    if (!expr.Maybe<TCoMember>()) {
+        return false;
+    }
+    auto typeAnn = expr.Ptr()->GetTypeAnn();
+    auto itemType = GetSeqItemType(typeAnn);
+    if (!itemType) {
+        itemType = typeAnn;
+    }
+    if (itemType->GetKind() != ETypeAnnotationKind::Data) {
+        return false;
+    }
+    auto dataTypeInfo = NUdf::GetDataTypeInfo(itemType->Cast<TDataExprType>()->GetSlot());
+    return (std::string(dataTypeInfo.Name.data()) == "String");
+}
+
+bool IsLikeOperator(const TCoCompare& predicate) {
+    if (predicate.Maybe<TCoCmpStringContains>()) {
+        return true;
+    } else if (predicate.Maybe<TCoCmpStartsWith>()) {
+        return true;
+    } else if (predicate.Maybe<TCoCmpEndsWith>()) {
+        return true;
+    }
+    return false;
+}
+
 bool IsSupportedPredicate(const TCoCompare& predicate) {
     if (predicate.Maybe<TCoCmpEqual>()) {
         return true;
@@ -30,13 +57,7 @@ bool IsSupportedPredicate(const TCoCompare& predicate) {
         return true;
     } else if (NKikimr::NSsa::RuntimeVersion >= 2U) {
         // We introduced LIKE pushdown in v2 of SSA program
-        if (predicate.Maybe<TCoCmpStringContains>()) {
-            return true;
-        } else if (predicate.Maybe<TCoCmpStartsWith>()) {
-            return true;
-        } else if (predicate.Maybe<TCoCmpEndsWith>()) {
-            return true;
-        }
+        return IsLikeOperator(predicate);
     }
 
     return false;
@@ -273,6 +294,10 @@ bool CheckComparisonParametersForPushdown(const TCoCompare& compare, const TExpr
         if (!IsComparableTypes(leftList[i], rightList[i], equality, inputType)) {
             return false;
         }
+        if (IsLikeOperator(compare) && (ColumnHasBinaryStringType(leftList[i]) || ColumnHasBinaryStringType(rightList[i]))) {
+            // Currently Column Shard doesn't have LIKE kernel for binary strings
+            return false;
+        }
     }
 
     return true;
diff --git a/ydb/core/kqp/ut/olap/kqp_olap_ut.cpp b/ydb/core/kqp/ut/olap/kqp_olap_ut.cpp
index b3d7b6be54c..683fec54d88 100644
--- a/ydb/core/kqp/ut/olap/kqp_olap_ut.cpp
+++ b/ydb/core/kqp/ut/olap/kqp_olap_ut.cpp
@@ -1406,6 +1406,29 @@ Y_UNIT_TEST_SUITE(KqpOlap) {
     }
 #endif
 
+    Y_UNIT_TEST(PredicatePushdown_LikeNotPushedDownForStringType) {
+        auto settings = TKikimrSettings()
+            .SetWithSampleTables(false);
+        TKikimrRunner kikimr(settings);
+
+        TStreamExecScanQuerySettings scanSettings;
+        scanSettings.Explain(true);
+
+        TTableWithNullsHelper(kikimr).CreateTableWithNulls();
+        WriteTestDataForTableWithNulls(kikimr, "/Root/tableWithNulls");
+        EnableDebugLogging(kikimr);
+
+        auto tableClient = kikimr.GetTableClient();
+        auto query = R"(SELECT id, binary_str FROM `/Root/tableWithNulls` WHERE binary_str LIKE "5%")";
+        auto it = tableClient.StreamExecuteScanQuery(query, scanSettings).GetValueSync();
+        UNIT_ASSERT_C(it.IsSuccess(), it.GetIssues().ToString());
+
+        auto result = CollectStreamResult(it);
+        auto ast = result.QueryStats->Getquery_ast();
+        UNIT_ASSERT_C(ast.find("KqpOlapFilter") == std::string::npos,
+                        TStringBuilder() << "Predicate pushed down. Query: " << query);
+    }
+
     Y_UNIT_TEST(PredicatePushdown_MixStrictAndNotStrict) {
         auto settings = TKikimrSettings()
             .SetWithSampleTables(false);
diff --git a/ydb/core/testlib/cs_helper.cpp b/ydb/core/testlib/cs_helper.cpp
index 0a7374d8d7c..990f6b7aede 100644
--- a/ydb/core/testlib/cs_helper.cpp
+++ b/ydb/core/testlib/cs_helper.cpp
@@ -356,7 +356,8 @@ std::shared_ptr<arrow::Schema> TTableWithNullsHelper::GetArrowSchema() {
         std::vector<std::shared_ptr<arrow::Field>>{
             arrow::field("id", arrow::int32()),
             arrow::field("resource_id", arrow::utf8()),
-            arrow::field("level", arrow::int32())
+            arrow::field("level", arrow::int32()),
+            arrow::field("binary_str", arrow::binary())
     });
 }
 
@@ -371,28 +372,33 @@ std::shared_ptr<arrow::RecordBatch> TTableWithNullsHelper::TestArrowBatch(ui64,
     arrow::Int32Builder b1;
     arrow::StringBuilder b2;
     arrow::Int32Builder b3;
+    arrow::StringBuilder b4;
 
     for (size_t i = 1; i <= rowCount / 2; ++i) {
         Y_VERIFY(b1.Append(i).ok());
         Y_VERIFY(b2.AppendNull().ok());
         Y_VERIFY(b3.Append(i).ok());
+        Y_VERIFY(b4.AppendNull().ok());
     }
 
     for (size_t i = rowCount / 2 + 1; i <= rowCount; ++i) {
         Y_VERIFY(b1.Append(i).ok());
         Y_VERIFY(b2.Append(std::to_string(i)).ok());
         Y_VERIFY(b3.AppendNull().ok());
+        Y_VERIFY(b4.Append(std::to_string(i)).ok());
     }
 
     std::shared_ptr<arrow::Int32Array> a1;
     std::shared_ptr<arrow::StringArray> a2;
     std::shared_ptr<arrow::Int32Array> a3;
+    std::shared_ptr<arrow::StringArray> a4;
 
     Y_VERIFY(b1.Finish(&a1).ok());
     Y_VERIFY(b2.Finish(&a2).ok());
     Y_VERIFY(b3.Finish(&a3).ok());
+    Y_VERIFY(b4.Finish(&a4).ok());
 
-    return arrow::RecordBatch::Make(schema, rowCount, { a1, a2, a3 });
+    return arrow::RecordBatch::Make(schema, rowCount, { a1, a2, a3, a4 });
 }
 
 }
diff --git a/ydb/core/testlib/cs_helper.h b/ydb/core/testlib/cs_helper.h
index b171f660519..cc76cb33976 100644
--- a/ydb/core/testlib/cs_helper.h
+++ b/ydb/core/testlib/cs_helper.h
@@ -182,6 +182,7 @@ public:
         Columns { Name: "id" Type: "Int32" NotNull: true }
         Columns { Name: "resource_id" Type: "Utf8" }
         Columns { Name: "level" Type: "Int32" }
+        Columns { Name: "binary_str" Type: "String" }
         KeyColumnNames: "id"
     )";