Small fixes

Author: ianton-ru

src/Storages/ObjectStorage/DataLakes/Iceberg/IcebergWrites.cpp

@@ -155,9 +155,9 @@ std::vector<uint8_t> dumpFieldToBytes(const Field & field, DataTypePtr type)
         case TypeIndex::Int32:
         case TypeIndex::Date:
         case TypeIndex::Date32:
+        case TypeIndex::Time:
             return dumpValue(field.safeGet<Int32>());
         case TypeIndex::Int64:
-        case TypeIndex::Time:
             return dumpValue(field.safeGet<Int64>());
         case TypeIndex::Time64:
         case TypeIndex::DateTime64:
@@ -423,8 +423,8 @@ void generateManifestFile(
 
                 case Field::Types::Decimal64:
                 {
-                    auto type_id = partition_types[i]->getTypeId();
-                    if (type_id == TypeIndex::Time64 || type_id == TypeIndex::Time)
+                    const WhichDataType which(*partition_types[i]);
+                    if (which.isTime64())
                     { /// Need to write logical type into Avro
                         auto scale = getDecimalScale(*partition_types[i]);
                         if (scale == 0)
@@ -442,7 +442,7 @@ void generateManifestFile(
                         {
                             throw Exception(
                                 ErrorCodes::BAD_ARGUMENTS,
-                                "Avro file supports only seconds, milliseconds and microsecods for time, partition precision: {}", scale);
+                                "Avro file supports only seconds, milliseconds and microseconds for time, partition precision: {}", scale);
                         }
                     }
                     else

src/Storages/ObjectStorage/DataLakes/Iceberg/Utils.cpp

@@ -484,8 +484,17 @@ Poco::Dynamic::Var getAvroType(DataTypePtr type)
         case TypeIndex::Int64:
         case TypeIndex::DateTime:
         case TypeIndex::DateTime64:
-        case TypeIndex::Time64:
             return "long";
+        case TypeIndex::Time64:
+        {
+            auto scale = getDecimalScale(*type);
+            if (scale == 0 || scale == 3)
+                return "int";
+            else if (scale == 6)
+                return "long";
+            else
+                throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unsupported type for iceberg {}", type->getName());
+        }
         case TypeIndex::Float32:
             return "float";
         case TypeIndex::Float64:
@@ -505,8 +514,14 @@ Poco::Dynamic::Var getAvroType(DataTypePtr type)
 
 Poco::Dynamic::Var getAvroLogicalType(DataTypePtr type)
 {
-    auto type_id = type->getTypeId();
-    if (type_id == TypeIndex::Time || type_id == TypeIndex::Time64)
+    if (type->isNullable())
+    {
+        auto type_nullable = std::static_pointer_cast<const DataTypeNullable>(type);
+        return getAvroLogicalType(type_nullable->getNestedType());
+    }
+
+    const WhichDataType which(type);
+    if (which.isTime64())
     {
         auto scale = getDecimalScale(*type);
         switch (scale)

tests/integration/test_database_iceberg/test.py

@@ -1018,8 +1018,7 @@ def _test_cluster_joins(started_cluster):
     assert res == "Jack\tBlack\nJack\tSilver\nJohn\tBlack\nJohn\tSilver\n"
 
 
-@pytest.mark.parametrize("storage_type", ["s3"])
-def test_partitioning_by_time(started_cluster, storage_type):
+def test_partitioning_by_time(started_cluster):
     node = started_cluster.instances["node1"]
 
     test_ref = f"test_partitioning_by_time_{uuid.uuid4()}"
@@ -1067,8 +1066,7 @@ def test_partitioning_by_time(started_cluster, storage_type):
     assert node.query(f"SELECT * FROM {CATALOG_NAME}.`{namespace}.{table_name}` WHERE key <= '13:00:00.000000' ORDER BY key") == "12:00:00.000000\ttest1\n13:00:00.000000\ttest2\n"
 
 
-@pytest.mark.parametrize("storage_type", ["s3"])
-def test_partitioning_by_string(started_cluster, storage_type):
+def test_partitioning_by_string(started_cluster):
     node = started_cluster.instances["node1"]
 
     test_ref = f"test_partitioning_by_string_{uuid.uuid4()}"

tests/integration/test_storage_iceberg_no_spark/test_write_time.py

@@ -3,13 +3,10 @@
 from helpers.iceberg_utils import get_uuid_str
 
 
-@pytest.mark.parametrize("time_type", ["Time", "Time64"])
+@pytest.mark.parametrize("time_type", ["Time", "Time64(0)", "Time64(3)", "Time64(6)"])
 def test_write_time(started_cluster_iceberg_no_spark, time_type):
     node = started_cluster_iceberg_no_spark.instances["node1"]
 
-    if time_type == "Time64":
-        time_type = "Time64(6)"
-
     TABLE_NAME = "test_partitioning_by_time_" + get_uuid_str()
 
     node.query(