Prevent tests from generating unwanted test artifacts (#304)

versioned_hdf5/tests/conftest.py

@@ -1,9 +1,12 @@
 import os
+from typing import Optional, Callable
 import uuid
 
+import h5py
 from pytest import fixture
-from .helpers import setup_vfile
+
 from ..api import VersionedHDF5File
+from ..backend import initialize
 
 # Run tests marked with @pytest.mark.slow last. See
 # https://stackoverflow.com/questions/61533694/run-slow-pytest-commands-at-the-end-of-the-test-suite
@@ -23,7 +26,7 @@ def filepath(tmp_path, request):
     yield file_name
 
 @fixture
-def h5file(tmp_path, request):
+def h5file(setup_vfile, tmp_path, request):
     file_name = os.path.join(tmp_path, 'file.hdf5')
     version_name = None
     m = request.node.get_closest_marker('setup_args')
@@ -308,3 +311,18 @@ def _check_running_version(target):
             f"versioned_hdf5=={__version__} installed; "
             f"this file only generates bad data for DATA_VERSION=={target}"
         )
+
+
+@fixture
+def setup_vfile(tmp_path: str) -> Callable[[Optional[str], Optional[str]], h5py.File]:
+    """Fixture which provides a function that creates an hdf5 file, optionally with groups."""
+    def _setup_vfile(file_name='file.hdf5', *, version_name=None):
+        f = h5py.File(tmp_path / file_name, 'w')
+        initialize(f)
+        if version_name:
+            if isinstance(version_name, str):
+                version_name = [version_name]
+            for name in version_name:
+                f['_version_data/versions'].create_group(name)
+        return f
+    return _setup_vfile

versioned_hdf5/tests/test_api.py

@@ -14,7 +14,6 @@
 import numpy as np
 from numpy.testing import assert_equal
 
-from .helpers import setup_vfile
 from ..backend import DEFAULT_CHUNK_SIZE, DATA_VERSION
 from ..api import VersionedHDF5File
 from ..versions import TIMESTAMP_FMT, all_versions
@@ -1046,24 +1045,27 @@ def test_group_contains(vfile):
     assert '/_version_data/versions/version1/group1/group2' in version1
     assert '/_version_data/versions/version1/group1/group2' not in version2
 
-@mark.setup_args(file_name='test.hdf5')
-def test_moved_file(tmp_path, h5file):
+def test_moved_file(setup_vfile, tmpdir):
+    h5file = setup_vfile()
+    path = pathlib.Path(h5file.filename)
+
     # See issue #28. Make sure the virtual datasets do not hard-code the filename.
     file = VersionedHDF5File(h5file)
     data = np.ones(2*DEFAULT_CHUNK_SIZE)
     with file.stage_version('version1') as group:
         group['dataset'] = data
     file.close()
 
-    with h5py.File('test.hdf5', 'r') as f:
+    with h5py.File(h5file.filename, 'r') as f:
         file = VersionedHDF5File(f)
         assert_equal(file['version1']['dataset'][:], data)
         file.close()
 
     # XXX: os.replace
-    os.rename('test.hdf5', 'test2.hdf5')
+    new_path = path.parent / 'test2.hdf5'
+    os.rename(path, new_path)
 
-    with h5py.File('test2.hdf5', 'r') as f:
+    with h5py.File(new_path, 'r') as f:
         file = VersionedHDF5File(f)
         assert_equal(file['version1']['dataset'][:], data)
         file.close()
@@ -1292,7 +1294,7 @@ def test_closes(vfile):
     raises(ValueError, lambda: vfile['version2'])
     assert repr(vfile) == "<Closed VersionedHDF5File>"
 
-def test_scalar_dataset():
+def test_scalar_dataset(setup_vfile):
     for data1, data2 in [
             (b'baz', b'foo'),
             (np.asarray('baz', dtype='S'), np.asarray('foo', dtype='S')),
@@ -1406,7 +1408,7 @@ def test_InMemoryArrayDataset_chunks(vfile):
         assert data_group['g/bar'].compression_opts == 3
 
 
-def test_string_dtypes():
+def test_string_dtypes(setup_vfile):
 
     # Make sure the fillvalue logic works correctly for custom h5py string
     # dtypes.
@@ -1473,8 +1475,10 @@ def test_empty(vfile):
     assert_equal(vfile['version2']['data2'][()], np.empty((1, 0, 2)))
 
 
-def test_read_only():
-    with setup_vfile('test.hdf5') as f:
+def test_read_only(setup_vfile):
+    vfile = setup_vfile()
+    filename = vfile.filename
+    with vfile as f:
         file = VersionedHDF5File(f)
         timestamp = datetime.datetime.now(datetime.timezone.utc)
         with file.stage_version('version1', timestamp=timestamp) as g:
@@ -1495,7 +1499,7 @@ def test_read_only():
         with raises(ValueError):
             file[timestamp]['data2'] = [1, 2, 3]
 
-    with h5py.File('test.hdf5', 'r+') as f:
+    with h5py.File(filename, 'r+') as f:
         file = VersionedHDF5File(f)
 
         with raises(ValueError):
@@ -1596,21 +1600,23 @@ def test_auto_create_group(vfile):
 
     assert_equal(vfile['version1']['a']['b']['c'][:], [0, 1, 2])
 
-def test_scalar():
-    with setup_vfile('test.hdf5') as f:
+def test_scalar(setup_vfile):
+    file = setup_vfile()
+    filename = file.filename
+    with file as f:
         vfile = VersionedHDF5File(f)
         with vfile.stage_version('version1') as g:
             dtype = h5py.special_dtype(vlen=bytes)
             g.create_dataset('bar', data=np.array(['aaa'], dtype='O'), dtype=dtype)
 
-    with h5py.File('test.hdf5', 'r+') as f:
+    with h5py.File(filename, 'r+') as f:
         vfile = VersionedHDF5File(f)
         assert isinstance(vfile['version1']['bar'], DatasetWrapper)
         assert isinstance(vfile['version1']['bar'].dataset, InMemoryDataset)
         # Should return a scalar, not a shape () array
         assert isinstance(vfile['version1']['bar'][0], bytes)
 
-    with h5py.File('test.hdf5', 'r') as f:
+    with h5py.File(filename, 'r') as f:
         vfile = VersionedHDF5File(f)
         assert isinstance(vfile['version1']['bar'], h5py.Dataset)
         # Should return a scalar, not a shape () array
@@ -1825,20 +1831,22 @@ def test_mask_reading_read_only(tmp_path):
         b = sv['bar'][mask]
         assert_equal(b, [1, 2])
 
-def test_read_only_no_wrappers():
+def test_read_only_no_wrappers(setup_vfile):
+    file = setup_vfile()
+    filename = file.filename
     # Read-only files should not use the wrapper classes
-    with setup_vfile('test.hdf5') as f:
+    with file as f:
         vfile = VersionedHDF5File(f)
         with vfile.stage_version('version1') as g:
             g.create_dataset('bar', data=np.array([0, 1, 2]))
 
-    with h5py.File('test.hdf5', 'r+') as f:
+    with h5py.File(filename, 'r+') as f:
         vfile = VersionedHDF5File(f)
         assert isinstance(vfile['version1'], InMemoryGroup)
         assert isinstance(vfile['version1']['bar'], DatasetWrapper)
         assert isinstance(vfile['version1']['bar'].dataset, InMemoryDataset)
 
-    with h5py.File('test.hdf5', 'r') as f:
+    with h5py.File(filename, 'r') as f:
         vfile = VersionedHDF5File(f)
         assert isinstance(vfile['version1'], h5py.Group)
         assert isinstance(vfile['version1']['bar'], h5py.Dataset)

versioned_hdf5/tests/test_backend.py

@@ -8,16 +8,14 @@
 
 import itertools
 
-from .helpers import setup_vfile
-
 from ..backend import (create_base_dataset, write_dataset,
                        create_virtual_dataset, DEFAULT_CHUNK_SIZE,
                        write_dataset_chunks)
 
 CHUNK_SIZE_3D = 2**4  # = cbrt(DEFAULT_CHUNK_SIZE)
 
 
-def test_initialize():
+def test_initialize(setup_vfile):
     with setup_vfile() as f:
         pass
     f.close()

versioned_hdf5/tests/test_hashtable.py

@@ -6,7 +6,6 @@
 
 from ..backend import create_base_dataset
 from ..hashtable import Hashtable
-from .helpers import setup_vfile
 from .. import VersionedHDF5File
 
 def test_hashtable(h5file):
@@ -30,7 +29,7 @@ def test_hashtable(h5file):
         with raises(ValueError):
             h[b'\x01'*32] = slice(0, 4, 2)
 
-def test_from_raw_data():
+def test_from_raw_data(setup_vfile):
     with setup_vfile('test.h5') as f:
         vf = VersionedHDF5File(f)
         with vf.stage_version('0') as sv:
@@ -50,13 +49,15 @@ def test_hashtable_multidimension(h5file):
     h = Hashtable(h5file, 'test_data')
     assert h.hash(np.ones((1, 2, 3,))) != h.hash(np.ones((3, 2, 1)))
 
-def test_issue_208():
-    with setup_vfile('test.h5') as f:
+def test_issue_208(setup_vfile):
+    file = setup_vfile()
+    filename = file.filename
+    with file as f:
         vf = VersionedHDF5File(f)
         with vf.stage_version('0') as sv:
             sv.create_dataset('bar', data=np.arange(10))
 
-    with h5py.File('test.h5', 'r+') as f:
+    with h5py.File(filename, 'r+') as f:
         vf = VersionedHDF5File(f)
         with vf.stage_version('1') as sv:
             sv['bar'].resize((12,))