def _get_null_counts(self, path, columns=None):
reader = FeatherReader(path)
counts = []
for i in range(reader.num_columns):
col = reader.get_column(i)
if columns is None or col.name in columns:
counts.append(col.null_count)
return counts
def test_num_columns_attr(self):
df0 = pd.DataFrame({})
df1 = pd.DataFrame({
'foo': [1, 2, 3, 4, 5]
})
df2 = pd.DataFrame({
'foo': [1, 2, 3, 4, 5],
'bar': [1, 2, 3, 4, 5]
})
for df, ncols in zip([df0, df1, df2], [0, 1, 2]):
path = random_path()
self.test_files.append(path)
write_feather(df, path)
reader = FeatherReader(path)
assert reader.num_columns == ncols
def test_file_not_exist(self):
with self.assertRaises(pa.ArrowIOError):
FeatherReader('test_invalid_file')
def test_file_not_exist(self):
with pytest.raises(pa.ArrowIOError):
FeatherReader('test_invalid_file')
def load(self, gs: Type[GeneSignature]) -> pd.DataFrame:
return FeatherReader(self._fname).read_pandas(
columns=(INDEX_NAME, ) + gs.genes).set_index(INDEX_NAME)
def load_full(self) -> pd.DataFrame:
return FeatherReader(self._fname).read_pandas().set_index(INDEX_NAME)
def genes(self) -> Tuple[str]:
# noinspection PyTypeChecker
reader = FeatherReader(self._fname)
return tuple(
reader.get_column_name(idx) for idx in range(self.total_genes)
if reader.get_column_name(idx) != INDEX_NAME)
def total_genes(self) -> int:
return FeatherReader(self._fname).num_columns - 1
def read_feather(cls, path, columns=None, use_threads=True):
"""Read a pandas.DataFrame from Feather format.
Ray DataFrame only supports pyarrow engine for now.
Args:
path: The filepath of the feather file.
We only support local files for now.
multi threading is set to True by default
columns: not supported by pandas api, but can be passed here to read only
specific columns
use_threads: Whether or not to use threads when reading
Notes:
pyarrow feather is used. Please refer to the documentation here
https://arrow.apache.org/docs/python/api.html#feather-format
"""
if cls.read_feather_remote_task is None:
return super(RayIO, cls).read_feather(
path, columns=columns, use_threads=use_threads
)
if columns is None:
from pyarrow.feather import FeatherReader
fr = FeatherReader(path)
columns = [fr.get_column_name(i) for i in range(fr.num_columns)]
from modin.pandas import DEFAULT_NPARTITIONS
num_partitions = DEFAULT_NPARTITIONS
num_splits = min(len(columns), num_partitions)
# Each item in this list will be a list of column names of the original df
column_splits = (
len(columns) // num_partitions
if len(columns) % num_partitions == 0
else len(columns) // num_partitions + 1
)
col_partitions = [
columns[i : i + column_splits]
for i in range(0, len(columns), column_splits)
]
blk_partitions = np.array(
[
cls.read_feather_remote_task._remote(
args=(path, cols, num_splits), num_return_vals=num_splits + 1
)
for cols in col_partitions
]
).T
remote_partitions = np.array(
[
[cls.frame_partition_cls(obj) for obj in row]
for row in blk_partitions[:-1]
]
)
index_len = ray.get(blk_partitions[-1][0])
index = pandas.RangeIndex(index_len)
new_query_compiler = cls.query_compiler_cls(
cls.frame_cls(remote_partitions, index, columns)
)
return new_query_compiler
def load_full(self) -> pd.DataFrame:
df = FeatherReader(self._fname).read_pandas()
# Avoid copying the whole dataframe by replacing the index in place.
# This makes loading a database twice as fast in case the database file is already in the filesystem cache.
df.set_index(INDEX_NAME, inplace=True)
return df
def total_genes(self) -> int:
# Do not count column 1 as it contains the index with the name of the features.
return FeatherReader(self._fname).num_columns - 1
