def test_store_config_map_a(self) -> None:
sc1 = StoreConfig(index_depth=3, columns_depth=3)
sc1m = StoreConfigMap.from_config(sc1)
self.assertEqual(sc1m['a'].index_depth, 3)
self.assertEqual(sc1m['b'].index_depth, 3)
sc2 = StoreConfig(include_index=False)
sc2m = StoreConfigMap.from_config(sc2)
self.assertEqual(sc2m['a'].include_index, False)
self.assertEqual(sc2m['b'].include_index, False)
def test_bus_to_xlsx_a(self) -> None:
f1 = Frame.from_dict(dict(a=(1, 2), b=(3, 4)),
index=('x', 'y'),
name='f1')
f2 = Frame.from_dict(dict(c=(1, 2, 3), b=(4, 5, 6)),
index=('x', 'y', 'z'),
name='f2')
f3 = Frame.from_dict(dict(d=(10, 20), b=(50, 60)),
index=('p', 'q'),
name='f3')
config = StoreConfigMap.from_config(
StoreConfig(index_depth=1,
columns_depth=1,
include_columns=True,
include_index=True))
b1 = Bus.from_frames((f1, f2, f3), config=config)
with temp_file('.xlsx') as fp:
b1.to_xlsx(fp)
b2 = Bus.from_xlsx(fp, config=config)
tuple(b2.items()) # force loading all
for frame in (f1, f2, f3):
self.assertEqualFrames(frame, b2[frame.name])
def test_bus_init_a(self) -> None:
f1 = Frame.from_dict(dict(a=(1, 2), b=(3, 4)),
index=('x', 'y'),
name='foo')
f2 = Frame.from_dict(dict(a=(1, 2, 3), b=(4, 5, 6)),
index=('x', 'y', 'z'),
name='bar')
config = StoreConfigMap.from_config(StoreConfig(index_depth=1))
b1 = Bus.from_frames((f1, f2), config=config)
self.assertEqual(b1.keys().values.tolist(), ['foo', 'bar'])
with temp_file('.zip') as fp:
b1.to_zip_tsv(fp)
b2 = Bus.from_zip_tsv(fp)
f3 = b2['bar']
f4 = b2['foo']
# import ipdb; ipdb.set_trace()
zs = StoreZipTSV(fp)
zs.write(b1.items())
# how to show that this derived getitem has derived type?
f3 = zs.read('foo', config=config['foo'])
self.assertEqual(f3.to_pairs(0), (('a', (('x', 1), ('y', 2))),
('b', (('x', 3), ('y', 4)))))
def test_bus_init_c(self) -> None:
f1 = Frame.from_dict(
dict(a=(1,2), b=(3,4)),
index=('x', 'y'),
name='foo')
f2 = Frame.from_dict(
dict(a=(1,2,3), b=(4,5,6)),
index=('x', 'y', 'z'),
name='bar')
config = StoreConfigMap.from_config(StoreConfig(index_depth=1))
b1 = Bus.from_frames((f1, f2), config=config)
self.assertEqual(b1.keys().values.tolist(),
['foo', 'bar'])
with temp_file('.zip') as fp:
b1.to_zip_csv(fp)
b2 = Bus.from_zip_csv(fp, config=config)
f1_loaded = b2['foo']
f2_loaded = b2['bar']
self.assertEqualFrames(f1, f1_loaded)
self.assertEqualFrames(f2, f2_loaded)
def test_store_xlsx_write_b(self) -> None:
f1 = Frame.from_records(
((None, np.nan, 50, 'a'), (None, -np.inf, -50, 'b'), (None, 60.4, -50, 'c')),
index=('p', 'q', 'r'),
columns=IndexHierarchy.from_product(('I', 'II'), ('a', 'b')),
)
config_map = StoreConfigMap.from_config(
StoreConfig(include_index=True, include_columns=True))
with temp_file('.xlsx') as fp:
st = StoreXLSX(fp)
st.write(((STORE_LABEL_DEFAULT, f1),), config=config_map)
c = StoreConfig(
index_depth=f1.index.depth,
columns_depth=f1.columns.depth
)
f2 = st.read(STORE_LABEL_DEFAULT, config=c)
# just a sample column for now
self.assertEqual(
f1[HLoc[('II', 'a')]].values.tolist(),
f2[HLoc[('II', 'a')]].values.tolist() )
self.assertEqualFrames(f1, f2)
def test_store_sqlite_read_many_a(self) -> None:
f1 = Frame.from_dict(dict(x=(1, 2, -5, 200), y=(3, 4, -5, -3000)),
index=IndexHierarchy.from_product(('I', 'II'),
('a', 'b')),
name='f1')
f2 = Frame.from_dict(dict(a=(1, 2, 3), b=(4, 5, 6)),
index=('x', 'y', 'z'),
name='f2')
f3 = Frame.from_records(((10, 20, 50, 60), (50.0, 60.4, -50, -60)),
index=('p', 'q'),
columns=IndexHierarchy.from_product(
('I', 'II'), ('a', 'b')),
name='f3')
f4 = Frame.from_records(
(
(10, 20, 50, False, 10, 20, 50, False),
(50.0, 60.4, -50, True, 50.0, 60.4, -50, True),
(234, 44452, 0, False, 234, 44452, 0, False),
(4, -4, 2000, True, 4, -4, 2000, True),
(10, 20, 50, False, 10, 20, 50, False),
(50.0, 60.4, -50, True, 50.0, 60.4, -50, True),
(234, 44452, 0, False, 234, 44452, 0, False),
(4, -4, 2000, True, 4, -4, 2000, True),
),
index=IndexHierarchy.from_product(
('top', 'bottom'), ('far', 'near'), ('left', 'right')),
columns=IndexHierarchy.from_product(('I', 'II'), ('a', 'b'),
(1, 2)),
name='f4')
frames = (f1, f2, f3, f4)
config_map_write = StoreConfigMap.from_config(
StoreConfig(include_index=True, include_columns=True))
with temp_file('.sqlite') as fp:
st1 = StoreSQLite(fp)
st1.write(((f.name, f) for f in frames), config=config_map_write)
labels = tuple(
st1.labels()) # this will read from file, not in memory
self.assertEqual(tuple(f.name for f in frames), labels)
config_map_read: tp.Dict[tp.Hashable, StoreConfig] = {}
for i, name in enumerate(labels):
f_src = frames[i]
c = StoreConfig(index_depth=f_src.index.depth,
columns_depth=f_src.columns.depth)
config_map_read[name] = c
for i, f_loaded in enumerate(
st1.read_many(labels, config=config_map_read)):
f_src = frames[i]
self.assertEqualFrames(f_src, f_loaded, compare_dtype=False)
def test_store_xlsx_write_a(self) -> None:
f1 = Frame.from_dict(dict(x=(1, 2, -5, 200), y=(3, 4, -5, -3000)),
index=IndexHierarchy.from_product(('I', 'II'),
('a', 'b')),
name='f1')
f2 = Frame.from_dict(dict(a=(1, 2, 3), b=(4, 5, 6)),
index=('x', 'y', 'z'),
name='f2')
f3 = Frame.from_records(((10, 20, 50, 60), (50.0, 60.4, -50, -60)),
index=('p', 'q'),
columns=IndexHierarchy.from_product(
('I', 'II'), ('a', 'b')),
name='f3')
f4 = Frame.from_records(
(
(10, 20, 50, False, 10, 20, 50, False),
(50.0, 60.4, -50, True, 50.0, 60.4, -50, True),
(234, 44452, 0, False, 234, 44452, 0, False),
(4, -4, 2000, True, 4, -4, 2000, True),
(10, 20, 50, False, 10, 20, 50, False),
(50.0, 60.4, -50, True, 50.0, 60.4, -50, True),
(234, 44452, 0, False, 234, 44452, 0, False),
(4, -4, 2000, True, 4, -4, 2000, True),
),
index=IndexHierarchy.from_product(
('top', 'bottom'), ('far', 'near'), ('left', 'right')),
columns=IndexHierarchy.from_product(('I', 'II'), ('a', 'b'),
(1, 2)),
name='f4')
frames = (f1, f2, f3, f4)
config_map = StoreConfigMap.from_config(
StoreConfig(include_index=True, include_columns=True))
with temp_file('.xlsx') as fp:
st1 = StoreXLSX(fp)
st1.write(((f.name, f) for f in frames), config=config_map)
# import ipdb; ipdb.set_trace()
sheet_names = tuple(
st1.labels()) # this will read from file, not in memory
self.assertEqual(tuple(f.name for f in frames), sheet_names)
for i, name in enumerate(sheet_names):
f_src = frames[i]
c = StoreConfig(index_depth=f_src.index.depth,
columns_depth=f_src.columns.depth)
f_loaded = st1.read(name, config=c)
self.assertEqualFrames(f_src, f_loaded, check_dtypes=False)
def test_bus_update_series_cache_iloc(self) -> None:
f1 = Frame.from_dict(dict(a=(1, 2), b=(3, 4)),
index=('x', 'y'),
name='foo')
config = StoreConfigMap.from_config(StoreConfig(index_depth=1))
# simulating a Bus with a FrameDefferred but no Store, just for testing
s1 = Series((f1, FrameDeferred), index=('p', 'q'))
b1 = Bus(s1, config=config)
self.assertFalse(b1._loaded_all)
with self.assertRaises(RuntimeError):
b1._update_series_cache_iloc(1)
def test_store_zip_pickle_c(self) -> None:
f1 = FrameGO.from_dict(dict(a=(1, 2), b=(3, 4)),
index=('x', 'y'),
name='foo')
config = StoreConfig(index_depth=1, include_index=True)
config_map = StoreConfigMap.from_config(config)
with temp_file('.zip') as fp:
st = StoreZipPickle(fp)
# raise if trying to store a FrameGO
with self.assertRaises(NotImplementedError):
st.write(((f1.name, f1), ))
def test_store_zip_pickle_b(self) -> None:
f1 = Frame.from_dict(dict(a=(1, 2), b=(3, 4)),
index=('x', 'y'),
name='foo')
config = StoreConfig(index_depth=1, include_index=True)
config_map = StoreConfigMap.from_config(config)
with temp_file('.zip') as fp:
st = StoreZipPickle(fp)
st.write(((f1.name, f1), ))
frame_stored = st.read(f1.name)
self.assertEqual(frame_stored.shape, f1.shape)
def test_store_xlsx_read_a(self) -> None:
f1 = Frame.from_elements([1, 2, 3],
index=('a', 'b', 'c'),
columns=('x', ))
config_map = StoreConfigMap.from_config(
StoreConfig(include_index=False, include_columns=True))
with temp_file('.xlsx') as fp:
st = StoreXLSX(fp)
st.write(((None, f1), ), config=config_map)
c = StoreConfig(index_depth=0, columns_depth=f1.columns.depth)
f2 = st.read(None, config=c)
self.assertTrue((f1.values == f2.values).all())
self.assertEqual(f2.to_pairs(0), (('x', ((0, 1), (1, 2), (2, 3))), ))
def test_store_xlsx_read_b(self) -> None:
index = IndexHierarchy.from_product(('left', 'right'), ('up', 'down'))
columns = IndexHierarchy.from_labels(((100, -5, 20), ))
f1 = Frame.from_elements([1, 2, 3, 4], index=index, columns=columns)
config_map = StoreConfigMap.from_config(
StoreConfig(include_index=False, include_columns=True))
with temp_file('.xlsx') as fp:
st = StoreXLSX(fp)
st.write(((None, f1), ), config=config_map)
c = StoreConfig(index_depth=0, columns_depth=f1.columns.depth)
f2 = st.read(None, config=c)
self.assertTrue((f1.values == f2.values).all())
self.assertEqual(f2.to_pairs(0),
(((100, -5, 20), ((0, 1), (1, 2), (2, 3), (3, 4))), ))
def test_bus_mloc_b(self) -> None:
f1 = Frame.from_dict(dict(a=(1, 2, 3)),
index=('x', 'y', 'z'),
name='f1')
f2 = Frame.from_dict(dict(x=(10, 20, 30)),
index=('q', 'r', 's'),
name='f2')
config = StoreConfigMap.from_config(StoreConfig(index_depth=1))
b1 = Bus.from_frames((f1, f2), config=config)
with temp_file('.xlsx') as fp:
b1.to_xlsx(fp)
b2 = Bus.from_xlsx(fp, config=config)
# only show memory locations for loaded Frames
self.assertTrue(b2.iloc[1].equals(f2))
self.assertEqual(
(b2.mloc == None).to_pairs(), #pylint: disable=C0121
(('f1', True), ('f2', False)))