mirror of
https://github.com/langchain-ai/delta-rs.git
synced 2026-07-18 20:34:31 -04:00
Avinash Changrani
dd3b1aabe1
- Refactor delete_dir to delete objects via ObjectStore::delete_stream for improved performance. - Add tests for nested deletes, sibling-prefix safety, and missing-prefix no-op behavior. Signed-off-by: Avinash Changrani <avinashchangrani99@gmail.com>
273 lines
8.3 KiB
Python
273 lines
8.3 KiB
Python
from collections import Counter
|
|
from pathlib import Path
|
|
|
|
import numpy as np
|
|
import pytest
|
|
|
|
|
|
@pytest.fixture
|
|
def file_systems(tmp_path: Path):
|
|
import pyarrow.fs as fs
|
|
|
|
from deltalake.fs import DeltaStorageHandler
|
|
|
|
store = fs.PyFileSystem(DeltaStorageHandler(str(tmp_path.absolute())))
|
|
arrow_fs = fs.SubTreeFileSystem(str(tmp_path.absolute()), fs.LocalFileSystem())
|
|
return (store, arrow_fs)
|
|
|
|
|
|
@pytest.fixture
|
|
def table_data():
|
|
import pyarrow as pa
|
|
|
|
return pa.Table.from_arrays(
|
|
[pa.array([1, 2, 3]), pa.array(["a", "b", "c"])], names=["int", "str"]
|
|
)
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_file_info(file_systems, table_data):
|
|
store, arrow_fs = file_systems
|
|
file_path = "table.parquet"
|
|
|
|
import pyarrow.parquet as pq
|
|
|
|
pq.write_table(table_data, file_path, filesystem=arrow_fs)
|
|
|
|
info = store.get_file_info(file_path)
|
|
arrow_info = arrow_fs.get_file_info(file_path)
|
|
|
|
assert type(info) is type(arrow_info)
|
|
assert info.path == arrow_info.path
|
|
assert info.type == arrow_info.type
|
|
assert info.size == arrow_info.size
|
|
assert info.mtime_ns == arrow_info.mtime_ns
|
|
assert info.mtime == arrow_info.mtime
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_get_file_info_selector(file_systems):
|
|
store, arrow_fs = file_systems
|
|
import pyarrow as pa
|
|
import pyarrow.dataset as ds
|
|
import pyarrow.fs as fs
|
|
|
|
table = pa.table({"a": range(10), "b": np.random.randn(10), "c": [1, 2] * 5})
|
|
partitioning = ds.partitioning(pa.schema([("c", pa.int64())]), flavor="hive")
|
|
ds.write_dataset(
|
|
table,
|
|
"/",
|
|
partitioning=partitioning,
|
|
format="parquet",
|
|
filesystem=arrow_fs,
|
|
)
|
|
|
|
selector = fs.FileSelector("/", recursive=True)
|
|
infos = store.get_file_info(selector)
|
|
arrow_infos = arrow_fs.get_file_info(selector)
|
|
|
|
assert Counter([i.type for i in infos]) == Counter([i.type for i in arrow_infos])
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_open_input_file(file_systems, table_data):
|
|
store, arrow_fs = file_systems
|
|
file_path = "table.parquet"
|
|
import pyarrow.parquet as pq
|
|
|
|
pq.write_table(table_data, file_path, filesystem=arrow_fs)
|
|
|
|
file = store.open_input_file(file_path)
|
|
arrow_file = arrow_fs.open_input_file(file_path)
|
|
|
|
# Check the metadata
|
|
assert file.mode == arrow_file.mode
|
|
assert file.closed == arrow_file.closed
|
|
assert file.size() == arrow_file.size()
|
|
assert file.isatty() == arrow_file.isatty()
|
|
assert file.readable() == arrow_file.readable()
|
|
assert file.seekable() == arrow_file.seekable()
|
|
|
|
# Check reading the same content
|
|
assert file.read() == arrow_file.read()
|
|
# Check subsequent read (should return no data anymore)
|
|
assert file.read() == arrow_file.read()
|
|
assert file.read() == b""
|
|
|
|
file = store.open_input_file(file_path)
|
|
arrow_file = arrow_fs.open_input_file(file_path)
|
|
|
|
# Check seeking works as expected
|
|
assert file.tell() == arrow_file.tell()
|
|
assert file.seek(2) == arrow_file.seek(2)
|
|
assert file.tell() == arrow_file.tell()
|
|
assert file.tell() == 2
|
|
|
|
# check reading works as expected
|
|
assert file.read(10) == arrow_file.read(10)
|
|
assert file.read1(10) == arrow_file.read1(10)
|
|
assert file.read_at(10, 0) == arrow_file.read_at(10, 0)
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_open_input_file_with_size(tmp_path, table_data):
|
|
file_path = "table.parquet"
|
|
input_size = 12345 # incorrect file size for testing purposes
|
|
import pyarrow.fs as fs
|
|
import pyarrow.parquet as pq
|
|
|
|
from deltalake.fs import DeltaStorageHandler
|
|
|
|
# test that injected file size gets stored correctly
|
|
store1 = DeltaStorageHandler(
|
|
str(tmp_path.absolute()), known_sizes={file_path: input_size}
|
|
)
|
|
wrapped_fs = fs.PyFileSystem(store1)
|
|
arrow_fs = fs.SubTreeFileSystem(str(tmp_path.absolute()), fs.LocalFileSystem())
|
|
pq.write_table(table_data, file_path, filesystem=arrow_fs)
|
|
file = wrapped_fs.open_input_file(file_path)
|
|
assert file.size() == input_size
|
|
|
|
# confirm that true size is different
|
|
store2 = DeltaStorageHandler(str(tmp_path.absolute()))
|
|
wrapped_fs = fs.PyFileSystem(store2)
|
|
arrow_fs = fs.SubTreeFileSystem(str(tmp_path.absolute()), fs.LocalFileSystem())
|
|
pq.write_table(table_data, file_path, filesystem=arrow_fs)
|
|
file = wrapped_fs.open_input_file(file_path)
|
|
assert file.size() != input_size
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_read_table(file_systems, table_data):
|
|
store, arrow_fs = file_systems
|
|
file_path = "table.parquet"
|
|
import pyarrow as pa
|
|
import pyarrow.parquet as pq
|
|
|
|
pq.write_table(table_data, file_path, filesystem=arrow_fs)
|
|
|
|
table = pq.read_table(file_path, filesystem=store)
|
|
arrow_table = pq.read_table(file_path, filesystem=arrow_fs)
|
|
|
|
assert isinstance(table, pa.Table)
|
|
assert table.equals(arrow_table)
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_read_dataset(file_systems):
|
|
store, arrow_fs = file_systems
|
|
import pyarrow as pa
|
|
import pyarrow.dataset as ds
|
|
import pyarrow.parquet as pq
|
|
|
|
table = pa.table({"a": range(10), "b": np.random.randn(10), "c": [1, 2] * 5})
|
|
|
|
pq.write_table(table.slice(0, 5), "data1.parquet", filesystem=arrow_fs)
|
|
pq.write_table(table.slice(5, 10), "data2.parquet", filesystem=arrow_fs)
|
|
|
|
dataset = ds.dataset("/", format="parquet", filesystem=store)
|
|
ds_table = dataset.to_table()
|
|
|
|
assert table.schema == dataset.schema
|
|
assert table.equals(ds_table)
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_write_table(file_systems):
|
|
store, _ = file_systems
|
|
import pyarrow as pa
|
|
import pyarrow.dataset as ds
|
|
import pyarrow.parquet as pq
|
|
|
|
table = pa.table({"a": range(10), "b": np.random.randn(10), "c": [1, 2] * 5})
|
|
|
|
pq.write_table(table.slice(0, 5), "data1.parquet", filesystem=store)
|
|
pq.write_table(table.slice(5, 10), "data2.parquet", filesystem=store)
|
|
|
|
dataset = ds.dataset("/", format="parquet", filesystem=store)
|
|
ds_table = dataset.to_table()
|
|
|
|
assert table.schema == ds_table.schema
|
|
assert table.equals(ds_table)
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_write_partitioned_dataset(file_systems):
|
|
store, arrow_fs = file_systems
|
|
import pyarrow as pa
|
|
import pyarrow.dataset as ds
|
|
|
|
table = pa.table({"a": range(10), "b": np.random.randn(10), "c": [1, 2] * 5})
|
|
|
|
partitioning = ds.partitioning(pa.schema([("c", pa.int64())]), flavor="hive")
|
|
ds.write_dataset(
|
|
table,
|
|
"/",
|
|
partitioning=partitioning,
|
|
format="parquet",
|
|
filesystem=store,
|
|
)
|
|
|
|
dataset = ds.dataset(
|
|
"/", format="parquet", filesystem=arrow_fs, partitioning=partitioning
|
|
)
|
|
ds_table = dataset.to_table().select(["a", "b", "c"])
|
|
|
|
dataset = ds.dataset(
|
|
"/", format="parquet", filesystem=store, partitioning=partitioning
|
|
)
|
|
ds_table2 = dataset.to_table().select(["a", "b", "c"])
|
|
|
|
assert table.schema == ds_table.schema
|
|
assert table.schema == ds_table2.schema
|
|
assert table.shape == ds_table.shape
|
|
assert table.shape == ds_table2.shape
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_delete_dir_nested(file_systems):
|
|
"""delete_dir removes all objects under the prefix, including nested paths."""
|
|
store, arrow_fs = file_systems
|
|
import pyarrow.fs as fs
|
|
|
|
arrow_fs.create_dir("mydir/sub")
|
|
with arrow_fs.open_output_stream("mydir/b.txt") as f:
|
|
f.write(b"b")
|
|
with arrow_fs.open_output_stream("mydir/sub/a.txt") as f:
|
|
f.write(b"a")
|
|
|
|
store.delete_dir("mydir")
|
|
|
|
info_b = arrow_fs.get_file_info("mydir/b.txt")
|
|
info_a = arrow_fs.get_file_info("mydir/sub/a.txt")
|
|
assert info_b.type == fs.FileType.NotFound
|
|
assert info_a.type == fs.FileType.NotFound
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_delete_dir_no_sibling_prefix_collision(file_systems):
|
|
"""delete_dir('foo') must not delete objects under 'foo2/'."""
|
|
store, arrow_fs = file_systems
|
|
import pyarrow.fs as fs
|
|
|
|
arrow_fs.create_dir("foo")
|
|
arrow_fs.create_dir("foo2")
|
|
with arrow_fs.open_output_stream("foo/x.txt") as f:
|
|
f.write(b"x")
|
|
with arrow_fs.open_output_stream("foo2/y.txt") as f:
|
|
f.write(b"y")
|
|
|
|
store.delete_dir("foo")
|
|
|
|
info_x = arrow_fs.get_file_info("foo/x.txt")
|
|
info_y = arrow_fs.get_file_info("foo2/y.txt")
|
|
assert info_x.type == fs.FileType.NotFound
|
|
assert info_y.type == fs.FileType.File
|
|
|
|
|
|
@pytest.mark.pyarrow
|
|
def test_delete_dir_missing_prefix_is_noop(file_systems):
|
|
"""delete_dir on a non-existent prefix should not raise."""
|
|
store, _ = file_systems
|
|
store.delete_dir("does_not_exist")
|