Tests: Fix Windows compatibility across unit test suite

Enable the full unit test suite to pass on Windows (3803 passed, 0 failures).

Changes:

- Reduce RANDOM_LENGTH from 20 to 8 to stay under Windows MAX_PATH (260 chars)

- Use warehouse.as_posix() in SQLite URIs for cross-platform path formatting

- Close file handles before temp directory cleanup (Windows file locking)

- Use os.path.abspath() in path assertions instead of hardcoded POSIX paths

- Use raw paths instead of file: URIs in pyarrow test fixtures

- Relax Rich box-drawing assertions to check content not formatting

- Skip Kerberos tests on Windows (puresasl C lib unavailable)

- Relax error message assertion ([Errno 2] vs [WinError 2])

Depends on #3721. Related: #2477

Author: qzyu999

tests/catalog/test_hive.py

@@ -18,6 +18,7 @@
 import base64
 import copy
 import struct
+import sys
 import threading
 import uuid
 from collections.abc import Generator
@@ -300,7 +301,7 @@ def test_create_table(
     # it to construct the assert_called_with
     metadata_location: str = called_hive_table.parameters["metadata_location"]
     assert metadata_location.endswith(".metadata.json")
-    assert "/database/table/metadata/" in metadata_location
+    assert "/database/table/metadata/" in metadata_location.replace("\\", "/")
     catalog._client.__enter__().create_table.assert_called_with(
         HiveTable(
             tableName="table",
@@ -481,7 +482,7 @@ def test_create_table_with_given_location_removes_trailing_slash(
     # it to construct the assert_called_with
     metadata_location: str = called_hive_table.parameters["metadata_location"]
     assert metadata_location.endswith(".metadata.json")
-    assert "/database/table-given-location/metadata/" in metadata_location
+    assert "/database/table-given-location/metadata/" in metadata_location.replace("\\", "/")
     catalog._client.__enter__().create_table.assert_called_with(
         HiveTable(
             tableName="table",
@@ -1369,6 +1370,7 @@ def test_create_hive_client_failure() -> None:
         assert mock_hive_client.call_count == 2
 
 
+@pytest.mark.skipif(sys.platform == "win32", reason="Kerberos/puresasl not available on Windows")
 def test_create_hive_client_with_kerberos(
     kerberized_hive_metastore_fake_url: str,
 ) -> None:
@@ -1381,6 +1383,7 @@ def test_create_hive_client_with_kerberos(
     assert client is not None
 
 
+@pytest.mark.skipif(sys.platform == "win32", reason="Kerberos/puresasl not available on Windows")
 def test_create_hive_client_with_kerberos_using_context_manager(
     kerberized_hive_metastore_fake_url: str,
 ) -> None:
@@ -1412,6 +1415,7 @@ def test_create_hive_client_with_kerberos_using_context_manager(
             assert open_client._iprot.trans.isOpen()
 
 
+@pytest.mark.skipif(sys.platform == "win32", reason="Kerberos/puresasl not available on Windows")
 def test_kerberized_client_uses_fresh_transport_on_reuse(
     kerberized_hive_metastore_fake_url: str,
 ) -> None:

tests/catalog/test_sql.py

@@ -61,7 +61,7 @@ def catalog_memory(catalog_name: str, warehouse: Path) -> Generator[SqlCatalog,
 @pytest.fixture(scope="module")
 def catalog_sqlite(catalog_name: str, warehouse: Path) -> Generator[SqlCatalog, None, None]:
     props = {
-        "uri": f"sqlite:////{warehouse}/sql-catalog",
+        "uri": f"sqlite:///{warehouse.as_posix()}/sql-catalog",
         "warehouse": f"file://{warehouse}",
     }
     catalog = SqlCatalog(catalog_name, **props)
@@ -72,7 +72,7 @@ def catalog_sqlite(catalog_name: str, warehouse: Path) -> Generator[SqlCatalog,
 
 @pytest.fixture(scope="module")
 def catalog_uri(warehouse: Path) -> str:
-    return f"sqlite:////{warehouse}/sql-catalog"
+    return f"sqlite:///{warehouse.as_posix()}/sql-catalog"
 
 
 @pytest.fixture(scope="module")
@@ -96,7 +96,7 @@ def test_creation_with_echo_parameter(catalog_name: str, warehouse: Path) -> Non
 
     for echo_param, expected_echo_value in test_cases:
         props = {
-            "uri": f"sqlite:////{warehouse}/sql-catalog",
+            "uri": f"sqlite:///{warehouse.as_posix()}/sql-catalog",
             "warehouse": f"file://{warehouse}",
         }
         # None is for default value
@@ -119,7 +119,7 @@ def test_creation_with_pool_pre_ping_parameter(catalog_name: str, warehouse: Pat
 
     for pool_pre_ping_param, expected_pool_pre_ping_value in test_cases:
         props = {
-            "uri": f"sqlite:////{warehouse}/sql-catalog",
+            "uri": f"sqlite:///{warehouse.as_posix()}/sql-catalog",
             "warehouse": f"file://{warehouse}",
         }
         # None is for default value
@@ -139,7 +139,7 @@ def test_creation_from_impl(catalog_name: str, warehouse: Path) -> None:
             catalog_name,
             **{
                 "py-catalog-impl": "pyiceberg.catalog.sql.SqlCatalog",
-                "uri": f"sqlite:////{warehouse}/sql-catalog",
+                "uri": f"sqlite:///{warehouse.as_posix()}/sql-catalog",
                 "warehouse": f"file://{warehouse}",
             },
         ),
@@ -268,7 +268,7 @@ def get_columns(engine: Engine) -> set[str]:
 
 
 def test_adds_iceberg_type_column_to_old_schema(warehouse: Path) -> None:
-    uri = f"sqlite:////{warehouse}/test-migration-add-col"
+    uri = f"sqlite:///{warehouse.as_posix()}/test-migration-add-col"
     engine = _create_v0_db(uri)
 
     # Verify the column does not exist in the old schema
@@ -334,7 +334,7 @@ def test_list_tables_filters_by_iceberg_type(warehouse: Path) -> None:
 
 
 def test_migration_to_v1_with_property_set(warehouse: Path) -> None:
-    uri = f"sqlite:////{warehouse}/test-v1-migrate"
+    uri = f"sqlite:///{warehouse.as_posix()}/test-v1-migrate"
     engine = _create_v0_db(uri)
 
     assert "iceberg_type" not in get_columns(engine)
@@ -361,7 +361,7 @@ def test_invalid_schema_version_raises(warehouse: Path) -> None:
 
 def test_list_tables_works_on_v0_schema(warehouse: Path) -> None:
     """list_tables should work on V0 schemas without iceberg_type column."""
-    uri = f"sqlite:////{warehouse}/test-v0-list"
+    uri = f"sqlite:///{warehouse.as_posix()}/test-v0-list"
     _create_v0_db(uri)
 
     catalog = SqlCatalog(
@@ -441,15 +441,15 @@ def _make_v0_catalog(uri: str, warehouse: Path) -> SqlCatalog:
 
 def test_namespace_exists_on_v0_schema(warehouse: Path) -> None:
     """namespace_exists should not fail on V0 schema (no iceberg_type column)."""
-    catalog = _make_v0_catalog(f"sqlite:////{warehouse}/test-v0-ns-exists", warehouse)
+    catalog = _make_v0_catalog(f"sqlite:///{warehouse.as_posix()}/test-v0-ns-exists", warehouse)
     catalog.create_namespace("ns")
     assert catalog.namespace_exists("ns")
     assert not catalog.namespace_exists("missing")
 
 
 def test_create_and_load_table_on_v0_schema(warehouse: Path) -> None:
     """create_table and load_table should work on V0 schema without iceberg_type column."""
-    catalog = _make_v0_catalog(f"sqlite:////{warehouse}/test-v0-create", warehouse)
+    catalog = _make_v0_catalog(f"sqlite:///{warehouse.as_posix()}/test-v0-create", warehouse)
     catalog.create_namespace("ns")
     schema = Schema(NestedField(1, "id", StringType(), required=True))
     tbl = catalog.create_table(("ns", "tbl"), schema)

tests/conftest.py

@@ -2388,7 +2388,7 @@ def empty_home_dir_path(tmp_path_factory: pytest.TempPathFactory) -> str:
     return home_path
 
 
-RANDOM_LENGTH = 20
+RANDOM_LENGTH = 8
 NUM_TABLES = 2
 
 
@@ -2445,15 +2445,15 @@ def hierarchical_namespace_list(hierarchical_namespace_name: str) -> list[str]:
 
 BUCKET_NAME = "test_bucket"
 TABLE_METADATA_LOCATION_REGEX = re.compile(
-    r"""s3://test_bucket/my_iceberg_database-[a-z]{20}.db/
-    my_iceberg_table-[a-z]{20}/metadata/
+    r"""s3://test_bucket/my_iceberg_database-[a-z]{8}.db/
+    my_iceberg_table-[a-z]{8}/metadata/
     [0-9]{5}-[a-f0-9]{8}-?[a-f0-9]{4}-?4[a-f0-9]{3}-?[89ab][a-f0-9]{3}-?[a-f0-9]{12}.metadata.json""",
     re.X,
 )
 
 BQ_TABLE_METADATA_LOCATION_REGEX = re.compile(
-    r"""gs://alexstephen-test-bq-bucket/my_iceberg_database_[a-z]{20}.db/
-    my_iceberg_table-[a-z]{20}/metadata/
+    r"""gs://alexstephen-test-bq-bucket/my_iceberg_database_[a-z]{8}.db/
+    my_iceberg_table-[a-z]{8}/metadata/
     [0-9]{5}-[a-f0-9]{8}-?[a-f0-9]{4}-?4[a-f0-9]{3}-?[89ab][a-f0-9]{3}-?[a-f0-9]{12}.metadata.json""",
     re.X,
 )
@@ -3138,7 +3138,7 @@ def _create_sql_without_rowcount_catalog(name: str, warehouse: Path) -> Catalog:
     from pyiceberg.catalog.sql import SqlCatalog
 
     props = {
-        "uri": f"sqlite:////{warehouse}/sql-catalog",
+        "uri": f"sqlite:///{warehouse.as_posix()}/sql-catalog",
         "warehouse": f"file://{warehouse}",
     }
     catalog = SqlCatalog(name, **props)

tests/io/test_io.py

@@ -53,6 +53,7 @@ def test_custom_local_input_file() -> None:
         data = f.read()
         assert data == b"foo"
         assert len(input_file) == 3
+        f.close()
 
 
 def test_custom_local_output_file() -> None:
@@ -91,6 +92,7 @@ def test_pickled_pyarrow_round_trip() -> None:
         data = f.read()
         assert data == b"foo"
         assert len(input_file) == 3
+        f.close()
         deserialized_file_io.delete(location=f"{absolute_file_location}")