Skip to content

Commit

Permalink
Switch python tests to rely on the new test path
Browse files Browse the repository at this point in the history
  • Loading branch information
rtyler committed Dec 30, 2023
1 parent 1d64783 commit eb898bf
Show file tree
Hide file tree
Showing 4 changed files with 34 additions and 34 deletions.
2 changes: 1 addition & 1 deletion python/tests/conftest.py
Original file line number Diff line number Diff line change
Expand Up @@ -58,7 +58,7 @@ def s3_localstack_creds():
"s3",
"sync",
"--quiet",
"../crates/deltalake-core/tests/data/simple_table",
"../crates/deltalake-test/tests/data/simple_table",
"s3://deltars/simple",
"--endpoint-url",
endpoint_url,
Expand Down
6 changes: 3 additions & 3 deletions python/tests/test_schema.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@


def test_table_schema():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
schema = dt.schema()
assert json.loads(schema.to_json()) == {
Expand All @@ -37,7 +37,7 @@ def test_table_schema():


def test_table_schema_pyarrow_simple():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
schema = dt.schema().to_pyarrow()
field = schema.field(0)
Expand All @@ -49,7 +49,7 @@ def test_table_schema_pyarrow_simple():


def test_table_schema_pyarrow_020():
table_path = "../crates/deltalake-core/tests/data/delta-0.2.0"
table_path = "../crates/deltalake-test/tests/data/delta-0.2.0"
dt = DeltaTable(table_path)
schema = dt.schema().to_pyarrow()
field = schema.field(0)
Expand Down
58 changes: 29 additions & 29 deletions python/tests/test_table_read.py
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@


def test_read_table_with_edge_timestamps():
table_path = "../crates/deltalake-core/tests/data/table_with_edge_timestamps"
table_path = "../crates/deltalake-test/tests/data/table_with_edge_timestamps"
dt = DeltaTable(table_path)
dataset = dt.to_pyarrow_dataset(
parquet_read_options=ParquetReadOptions(coerce_int96_timestamp_unit="ms")
Expand All @@ -46,19 +46,19 @@ def test_read_table_with_edge_timestamps():


def test_read_simple_table_to_dict():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
assert dt.to_pyarrow_dataset().to_table().to_pydict() == {"id": [5, 7, 9]}


def test_read_simple_table_by_version_to_dict():
table_path = "../crates/deltalake-core/tests/data/delta-0.2.0"
table_path = "../crates/deltalake-test/tests/data/delta-0.2.0"
dt = DeltaTable(table_path, version=2)
assert dt.to_pyarrow_dataset().to_table().to_pydict() == {"value": [1, 2, 3]}


def test_read_simple_table_using_options_to_dict():
table_path = "../crates/deltalake-core/tests/data/delta-0.2.0"
table_path = "../crates/deltalake-test/tests/data/delta-0.2.0"
dt = DeltaTable(table_path, version=2, storage_options={})
assert dt.to_pyarrow_dataset().to_table().to_pydict() == {"value": [1, 2, 3]}

Expand All @@ -72,7 +72,7 @@ def test_read_simple_table_using_options_to_dict():
],
)
def test_load_as_version_datetime(date_value: str, expected_version):
log_dir = "../crates/deltalake-core/tests/data/simple_table/_delta_log"
log_dir = "../crates/deltalake-test/tests/data/simple_table/_delta_log"
log_mtime_pair = [
("00000000000000000000.json", 1588398451.0),
("00000000000000000001.json", 1588484851.0),
Expand All @@ -84,7 +84,7 @@ def test_load_as_version_datetime(date_value: str, expected_version):
file_path = os.path.join(log_dir, file_name)
os.utime(file_path, (dt_epoch, dt_epoch))

table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
dt.load_as_version(date_value)
assert dt.version() == expected_version
Expand All @@ -94,7 +94,7 @@ def test_load_as_version_datetime(date_value: str, expected_version):


def test_load_as_version_datetime_bad_format():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)

for bad_format in [
Expand All @@ -107,15 +107,15 @@ def test_load_as_version_datetime_bad_format():


def test_read_simple_table_update_incremental():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path, version=0)
assert dt.to_pyarrow_dataset().to_table().to_pydict() == {"id": [0, 1, 2, 3, 4]}
dt.update_incremental()
assert dt.to_pyarrow_dataset().to_table().to_pydict() == {"id": [5, 7, 9]}


def test_read_simple_table_file_sizes_failure():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
add_actions = dt.get_add_actions().to_pydict()

Expand All @@ -132,7 +132,7 @@ def test_read_simple_table_file_sizes_failure():


def test_read_partitioned_table_to_dict():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
expected = {
"value": ["1", "2", "3", "6", "7", "5", "4"],
Expand All @@ -144,7 +144,7 @@ def test_read_partitioned_table_to_dict():


def test_read_partitioned_table_with_partitions_filters_to_dict():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
partitions = [("year", "=", "2021")]
expected = {
Expand All @@ -158,15 +158,15 @@ def test_read_partitioned_table_with_partitions_filters_to_dict():


def test_read_empty_delta_table_after_delete():
table_path = "../crates/deltalake-core/tests/data/delta-0.8-empty"
table_path = "../crates/deltalake-test/tests/data/delta-0.8-empty"
dt = DeltaTable(table_path)
expected = {"column": []}

assert dt.to_pyarrow_dataset().to_table().to_pydict() == expected


def test_read_table_with_column_subset():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
expected = {
"value": ["1", "2", "3", "6", "7", "5", "4"],
Expand All @@ -179,7 +179,7 @@ def test_read_table_with_column_subset():


def test_read_table_as_category():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)

assert dt.schema().to_pyarrow().field("value").type == pa.string()
Expand All @@ -193,7 +193,7 @@ def test_read_table_as_category():


def test_read_table_with_filter():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
expected = {
"value": ["6", "7", "5"],
Expand All @@ -210,7 +210,7 @@ def test_read_table_with_filter():


def test_read_table_with_stats():
table_path = "../crates/deltalake-core/tests/data/COVID-19_NYT"
table_path = "../crates/deltalake-test/tests/data/COVID-19_NYT"
dt = DeltaTable(table_path)
dataset = dt.to_pyarrow_dataset()

Expand All @@ -236,7 +236,7 @@ def test_read_table_with_stats():


def test_read_special_partition():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-special-partition"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-special-partition"
dt = DeltaTable(table_path)

file1 = (
Expand All @@ -258,7 +258,7 @@ def test_read_special_partition():


def test_read_partitioned_table_metadata():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
metadata = dt.metadata()
assert metadata.id == "fe5a3c11-30d4-4dd7-b115-a1c121e66a4e"
Expand All @@ -270,15 +270,15 @@ def test_read_partitioned_table_metadata():


def test_read_partitioned_table_protocol():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
protocol = dt.protocol()
assert protocol.min_reader_version == 1
assert protocol.min_writer_version == 2


def test_read_table_with_cdc():
table_path = "../crates/deltalake-core/tests/data/simple_table_with_cdc"
table_path = "../crates/deltalake-test/tests/data/simple_table_with_cdc"
dt = DeltaTable(table_path)
assert dt.to_pyarrow_table().to_pydict() == {
"id": [0],
Expand All @@ -287,7 +287,7 @@ def test_read_table_with_cdc():


def test_history_partitioned_table_metadata():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
history = dt.history()
commit_info = history[0]
Expand All @@ -312,7 +312,7 @@ def test_history_partitioned_table_metadata():

@pytest.mark.parametrize("flatten", [True, False])
def test_add_actions_table(flatten: bool):
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
actions_df = dt.get_add_actions(flatten)
# RecordBatch doesn't have a sort_by method yet
Expand Down Expand Up @@ -356,10 +356,10 @@ def assert_correct_files(dt: DeltaTable, partition_filters, expected_paths):


def test_get_files_partitioned_table():
table_path = "../crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
table_path = "../crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
dt = DeltaTable(table_path)
table_path = (
Path.cwd().parent / "crates/deltalake-core/tests/data/delta-0.8.0-partitioned"
Path.cwd().parent / "crates/deltalake-test/tests/data/delta-0.8.0-partitioned"
).as_posix()

partition_filters = [("day", "=", "3")]
Expand Down Expand Up @@ -432,22 +432,22 @@ def test_get_files_partitioned_table():

@pytest.mark.pandas
def test_delta_table_to_pandas():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
assert dt.to_pandas().equals(pd.DataFrame({"id": [5, 7, 9]}))


@pytest.mark.pandas
def test_delta_table_with_filesystem():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
filesystem = SubTreeFileSystem(table_path, LocalFileSystem())
assert dt.to_pandas(filesystem=filesystem).equals(pd.DataFrame({"id": [5, 7, 9]}))


@pytest.mark.pandas
def test_delta_table_with_filters():
table_path = "../crates/deltalake-core/tests/data/COVID-19_NYT"
table_path = "../crates/deltalake-test/tests/data/COVID-19_NYT"
dt = DeltaTable(table_path)
dataset = dt.to_pyarrow_dataset()

Expand Down Expand Up @@ -489,7 +489,7 @@ def test_delta_table_with_filters():


def test_writer_fails_on_protocol():
table_path = "../crates/deltalake-core/tests/data/simple_table"
table_path = "../crates/deltalake-test/tests/data/simple_table"
dt = DeltaTable(table_path)
dt.protocol = Mock(return_value=ProtocolVersions(2, 1))
with pytest.raises(DeltaProtocolError):
Expand Down Expand Up @@ -547,7 +547,7 @@ def join(self, timeout=None):
@pytest.mark.timeout(timeout=5, method="thread")
def test_read_multiple_tables_from_s3(s3_localstack):
"""Should be able to create multiple cloud storage based DeltaTable instances
without blocking on async crates/deltalake-core function calls.
without blocking on async crates/deltalake-test function calls.
"""
for path in ["s3://deltars/simple", "s3://deltars/simple"]:
t = DeltaTable(path)
Expand Down
2 changes: 1 addition & 1 deletion python/tests/test_vacuum.py
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@


def test_vacuum_dry_run_simple_table():
table_path = "../crates/deltalake-core/tests/data/delta-0.2.0"
table_path = "../crates/deltalake-test/tests/data/delta-0.2.0"
dt = DeltaTable(table_path)
retention_periods = 169
tombstones = dt.vacuum(retention_periods)
Expand Down

0 comments on commit eb898bf

Please sign in to comment.