class GpuArrowNodeReader(object):
def __init__(self, table, index):
self._table = table
self._field = table.schema[index]
self._series = Series(column.as_column(table.column(index)))
self._series.name = self.name
def __len__(self):
return len(self._series)
@property
def schema(self):
return self._table.schema
@property
def field_schema(self):
return self._field
@property
def is_dictionary(self):
return pa.types.is_dictionary(self._field.type)
@property
def null_count(self):
return self._series.null_count
@property
def dtype(self):
return arrow_to_pandas_dtype(self._field.type)
@property
def index_dtype(self):
return self._field.type.index_type.to_pandas_dtype()
@property
def name(self):
return self._field.name
@property
def data(self):
"""
Return the data as the expected dtype
and with the padding bytes truncated.
"""
if self.data_raw is not None:
return self.data_raw.view(
self.dtype if not self.is_dictionary else self.index_dtype)
@property
def null(self):
"""
Return the null mask with the padding bytes truncated.
"""
if self.null_raw is not None:
bits = mask_bitsize
itemsize = mask_dtype.itemsize
end = ((len(self) + bits - 1) // bits) * itemsize
return self.null_raw[:end].view(mask_dtype)
@property
def data_raw(self):
"Accessor for the data buffer as a device array"
return self._series._column.data_array_view
@property
def null_raw(self):
"Accessor for the null buffer as a device array"
return self._series._column.mask_array_view
def make_series(self):
"""Make a Series object out of this node
"""
return self._series.copy(deep=False)
def _make_dictionary_series(self):
"""Make a dictionary-encoded series from this node
"""
assert self.is_dictionary
return self._series.copy(deep=False)
