def next(self) -> RecordBatch:
while True:
try:
batch = self._reader.read_next_batch()
except StopIteration:
break
yield RecordBatch(batch)
def next(self) -> Iterable[RecordBatch]:
for batch in self._parent_operator.next():
if not self.agg_obj:
self._init_agg_obj(batch)
self.agg_obj.next(batch.get_batch())
if self.agg_obj:
yield RecordBatch(self.agg_obj.result())
del self.agg_obj
def next(self) -> Iterable[RecordBatch]:
for batch in self._parent_operator.next():
self._process_arguments(self._arguments, batch=batch)
col_names = tuple(i[0] for i in self._expressions)
exprs = tuple(i[1] for i in self._expressions)
batch = RecordBatch.from_arrays(
tuple(chain(batch.columns, exprs)),
tuple(chain(batch.column_names, col_names))
)
# Remove, once sorting by boolean columns is supported by Arrow
self._verify_bool_columns(batch.get_schema())
self._sort_op.next(batch.get_batch())
self._expressions.clear()
yield RecordBatch(self._sort_op.sorted())
del self._sort_op
def next(self) -> RecordBatch:
while True:
batch = self._reader.next()
if batch is None:
break
yield RecordBatch(batch)