Add the rest of the _convert_schema_if_needed calls

Author: rambleraptor

pyiceberg/catalog/dynamodb.py

@@ -53,7 +53,7 @@
 from pyiceberg.partitioning import UNPARTITIONED_PARTITION_SPEC, PartitionSpec
 from pyiceberg.schema import Schema
 from pyiceberg.serializers import FromInputFile
-from pyiceberg.table import CommitTableResponse, Table
+from pyiceberg.table import CommitTableResponse, Table, TableProperties
 from pyiceberg.table.locations import load_location_provider
 from pyiceberg.table.metadata import new_table_metadata
 from pyiceberg.table.sorting import UNSORTED_SORT_ORDER, SortOrder
@@ -181,7 +181,10 @@ def create_table(
             ValueError: If the identifier is invalid, or no path is given to store metadata.
 
         """
-        schema: Schema = self._convert_schema_if_needed(schema)  # type: ignore
+        schema: Schema = self._convert_schema_if_needed(  # type: ignore
+            schema,
+            int(properties.get(TableProperties.FORMAT_VERSION, TableProperties.DEFAULT_FORMAT_VERSION)),  # type: ignore
+        )
 
         database_name, table_name = self.identifier_to_database_and_table(identifier)
 

pyiceberg/catalog/sql.py

@@ -62,7 +62,7 @@
 from pyiceberg.partitioning import UNPARTITIONED_PARTITION_SPEC, PartitionSpec
 from pyiceberg.schema import Schema
 from pyiceberg.serializers import FromInputFile
-from pyiceberg.table import CommitTableResponse, Table
+from pyiceberg.table import CommitTableResponse, Table, TableProperties
 from pyiceberg.table.locations import load_location_provider
 from pyiceberg.table.metadata import new_table_metadata
 from pyiceberg.table.sorting import UNSORTED_SORT_ORDER, SortOrder
@@ -200,7 +200,10 @@ def create_table(
             ValueError: If the identifier is invalid, or no path is given to store metadata.
 
         """
-        schema: Schema = self._convert_schema_if_needed(schema)  # type: ignore
+        schema: Schema = self._convert_schema_if_needed(  # type: ignore
+            schema,
+            int(properties.get(TableProperties.FORMAT_VERSION, TableProperties.DEFAULT_FORMAT_VERSION)),  # type: ignore
+        )
 
         namespace_identifier = Catalog.namespace_from(identifier)
         table_name = Catalog.table_name_from(identifier)

pyiceberg/io/pyarrow.py

@@ -1453,6 +1453,7 @@ def _task_to_record_batches(
     case_sensitive: bool,
     name_mapping: Optional[NameMapping] = None,
     partition_spec: Optional[PartitionSpec] = None,
+    format_version: TableVersion = TableProperties.DEFAULT_FORMAT_VERSION,
 ) -> Iterator[pa.RecordBatch]:
     arrow_format = ds.ParquetFileFormat(pre_buffer=True, buffer_size=(ONE_MEGABYTE * 8))
     with io.new_input(task.file.file_path).open() as fin:
@@ -1462,7 +1463,9 @@ def _task_to_record_batches(
         # Hence it is reasonable to always cast 'ns' timestamp to 'us' on read.
         # When V3 support is introduced, we will update `downcast_ns_timestamp_to_us` flag based on
         # the table format version.
-        file_schema = pyarrow_to_schema(physical_schema, name_mapping, downcast_ns_timestamp_to_us=True)
+        file_schema = pyarrow_to_schema(
+            physical_schema, name_mapping, downcast_ns_timestamp_to_us=True, format_version=format_version
+        )
 
         # Apply column projection rules: https://iceberg.apache.org/spec/#column-projection
         projected_missing_fields = _get_column_projection_values(
@@ -1691,6 +1694,7 @@ def _record_batches_from_scan_tasks_and_deletes(
                 self._case_sensitive,
                 self._table_metadata.name_mapping(),
                 self._table_metadata.specs().get(task.file.spec_id),
+                self._table_metadata.format_version,
             )
             for batch in batches:
                 if self._limit is not None:

pyiceberg/table/update/schema.py

@@ -34,6 +34,7 @@
     visit,
     visit_with_partner,
 )
+from pyiceberg.table import TableProperties
 from pyiceberg.table.name_mapping import (
     NameMapping,
     update_mapping,
@@ -48,7 +49,7 @@
     UpdatesAndRequirements,
     UpdateTableMetadata,
 )
-from pyiceberg.typedef import L
+from pyiceberg.typedef import L, TableVersion
 from pyiceberg.types import IcebergType, ListType, MapType, NestedField, PrimitiveType, StructType
 
 if TYPE_CHECKING:
@@ -142,11 +143,13 @@ def case_sensitive(self, case_sensitive: bool) -> UpdateSchema:
         self._case_sensitive = case_sensitive
         return self
 
-    def union_by_name(self, new_schema: Union[Schema, "pa.Schema"]) -> UpdateSchema:
+    def union_by_name(
+        self, new_schema: Union[Schema, "pa.Schema"], format_version: TableVersion = TableProperties.DEFAULT_FORMAT_VERSION
+    ) -> UpdateSchema:
         from pyiceberg.catalog import Catalog
 
         visit_with_partner(
-            Catalog._convert_schema_if_needed(new_schema),
+            Catalog._convert_schema_if_needed(new_schema, format_version=format_version),
             -1,
             _UnionByNameVisitor(update_schema=self, existing_schema=self._schema, case_sensitive=self._case_sensitive),
             # type: ignore

tests/integration/test_writes/test_writes.py

@@ -2251,13 +2251,19 @@ def test_branch_py_write_spark_read(session_catalog: Catalog, spark: SparkSessio
 @pytest.mark.integration
 def test_nanosecond_support_on_catalog(session_catalog: Catalog) -> None:
     identifier = "default.test_nanosecond_support_on_catalog"
-    # Create a pyarrow table with a nanosecond timestamp column
-    table = pa.Table.from_arrays(
-        [
-            pa.array([datetime.now()], type=pa.timestamp("ns")),
-            pa.array([datetime.now()], type=pa.timestamp("ns", tz="America/New_York")),
-        ],
-        names=["timestamp_ns", "timestamptz_ns"],
+
+    catalog = load_catalog("default", type="in-memory")
+    catalog.create_namespace("ns")
+
+    table = pa.Table.from_arrays([pa.array([datetime.now()], type=pa.timestamp("ns"))], names=["timestamps"])
+    table2 = pa.Table.from_arrays(
+        [pa.array([datetime.now()], type=pa.timestamp("ns", tz="America/New_York"))], names=["timestamps"]
     )
 
     _create_table(session_catalog, identifier, {"format-version": "3"}, schema=table.schema)
+
+    with pytest.raises(NotImplementedError, match="Writing V3 is not yet supported"):
+        catalog.create_table("ns.table1", schema=table.schema, properties={"format-version": "3"})
+
+    with pytest.raises(NotImplementedError, match="Writing V3 is not yet supported"):
+        catalog.create_table("ns.table2", schema=table2.schema, properties={"format-version": "3"})