def _store_reader(
store: Store,
config: StoreConfigMap,
labels: tp.Iterator[tp.Hashable],
max_persist: tp.Optional[int],
) -> FrameIterType:
'''
Read as many labels as possible from Store, then yield back each one at a time. If max_persist is active, max_persist will set the maximum number of Frame to load per read. Using Store.read_many is shown to have significant performance benefits on large collections of Frame.
'''
if max_persist is None:
for frame in store.read_many(labels, config=config):
yield frame
elif max_persist > 1:
coll = []
for label in labels:
coll.append(label)
# try to collect max_persist-sized bundles in coll, then use read_many to get all at once, then clear if we have more to iter
if len(coll) == max_persist:
for frame in store.read_many(coll, config=config):
yield frame
coll.clear()
if coll: # less than max persist remaining
for frame in store.read_many(coll, config=config):
yield frame
else: # max persist is 1
for label in labels:
yield store.read(label, config=config[labels])
def _from_store(
cls,
store: Store,
config: StoreConfigMapInitializer = None,
) -> 'Batch':
config_map = StoreConfigMap.from_initializer(config)
items = ((label, store.read(label, config=config_map[label]))
for label in store.labels())
return cls(items, config=config)
def _from_store(
cls,
store: Store,
config: StoreConfigMapInitializer = None,
max_persist: tp.Optional[int] = None, # not used
) -> 'Batch':
config_map = StoreConfigMap.from_initializer(config)
items = ((label, store.read(label, config=config_map[label]))
for label in store.labels())
return cls(items, config=config)
def test_store_get_field_names_and_dtypes_d(self) -> None:
from static_frame.core.index_hierarchy import IndexHierarchy
columns = IndexHierarchy.from_labels(((1, 'a'), (1, 'b'), (2, 'c')),
name=('foo', 'bar'))
f1 = Frame.from_records((('a', True, None), ),
index=(('a', )),
columns=columns)
field_names, dtypes = Store.get_field_names_and_dtypes(
frame=f1,
include_index=True,
include_index_name=False,
include_columns=True,
include_columns_name=True,
)
self.assertEqual(field_names,
[('foo', 'bar'), "[1 'a']", "[1 'b']", "[2 'c']"])
self.assertTrue(len(field_names) == len(dtypes))
field_names, dtypes = Store.get_field_names_and_dtypes(
frame=f1,
include_index=True,
include_index_name=False,
include_columns=True,
include_columns_name=True,
force_brackets=True,
)
self.assertEqual(field_names,
["['foo' 'bar']", "[1 'a']", "[1 'b']", "[2 'c']"])
with self.assertRaises(StoreParameterConflict):
field_names, dtypes = Store.get_field_names_and_dtypes(
frame=f1,
include_index=True,
include_index_name=False,
include_columns=True,
include_columns_name=False,
)
with self.assertRaises(StoreParameterConflict):
field_names, dtypes = Store.get_field_names_and_dtypes(
frame=f1,
include_index=False,
include_index_name=False,
include_columns=True,
include_columns_name=True,
)
def _from_store(cls,
store: Store,
config: StoreConfigMapInitializer = None
) -> 'Bus':
return cls(cls._deferred_series(store.labels()),
store=store,
config=config
)
def test_store_get_field_names_and_dtypes_c(self) -> None:
f1 = Frame.from_records((('a', True, None),), index=(('a',)), columns=(('x', 'y', 'z'))).rename(columns='foo')
with self.assertRaises(StoreParameterConflict):
field_names, dtypes = Store.get_field_names_and_dtypes(frame=f1,
include_index=False,
include_index_name=True,
include_columns=True,
include_columns_name=True,
)
field_names, dtypes = Store.get_field_names_and_dtypes(frame=f1,
include_index=True,
include_index_name=False,
include_columns=True,
include_columns_name=True,
)
self.assertEqual(field_names, ['foo', 'x', 'y', 'z'])
self.assertTrue(len(field_names) == len(dtypes))
def test_store_get_field_names_and_dtypes_b(self) -> None:
f1 = Frame.from_records((('a', True, None),), index=(('a',)), columns=(('x', 'y', 'z')))
field_names, dtypes = Store.get_field_names_and_dtypes(frame=f1,
include_index=False,
include_columns=True)
self.assertEqual(field_names.tolist(), ['x', 'y', 'z']) #type: ignore
self.assertEqual(dtypes.tolist(), #type: ignore
[np.dtype('<U1'), np.dtype('bool'), np.dtype('O')])
def _from_store(cls,
store: Store,
*,
config: StoreConfigMapInitializer = None,
max_persist: tp.Optional[int],
) -> 'Bus':
return cls(cls._deferred_series(store.labels()),
store=store,
config=config,
max_persist=max_persist,
)
def _from_store(
cls,
store: Store,
*,
config: StoreConfigMapInitializer = None,
max_workers: tp.Optional[int] = None,
chunksize: int = 1,
use_threads: bool = False,
) -> 'Batch':
config_map = StoreConfigMap.from_initializer(config)
items = ((label, store.read(label, config=config_map[label]))
for label in store.labels(config=config_map))
return cls(
items,
config=config,
max_workers=max_workers,
chunksize=chunksize,
use_threads=use_threads,
)
def test_store_get_field_names_and_dtypes_a(self) -> None:
f1 = Frame.from_records((('a', True, None),), index=(('a',)), columns=(('x', 'y', 'z')))
field_names, dtypes = Store.get_field_names_and_dtypes(frame=f1,
include_index=False,
include_index_name=True,
include_columns=False,
include_columns_name=False,
)
self.assertEqual(field_names, range(0, 3))
self.assertEqual(dtypes,
[np.dtype('<U1'), np.dtype('bool'), np.dtype('O')])
def _from_store(cls,
store: Store,
*,
config: StoreConfigMapInitializer = None,
max_persist: tp.Optional[int] = None,
index_constructor: IndexConstructor = None,
) -> 'Bus':
return cls(cls._deferred_series(
store.labels(config=config),
index_constructor=index_constructor,
),
store=store,
config=config,
max_persist=max_persist,
own_data=True,
)
def _from_store(
cls,
store: Store,
*,
config: StoreConfigMapInitializer = None,
max_persist: tp.Optional[int] = None,
index_constructor: IndexConstructor = None,
) -> 'Bus':
return cls(
None, # will generate FrameDeferred array
index=store.labels(config=config),
index_constructor=index_constructor,
store=store,
config=config,
max_persist=max_persist,
own_data=True,
)
