def _seek_internal(self, target_index):
if (self._raw_data_iter is not None
and self._raw_data_iter.get_index() == target_index):
return target_index, self._raw_data_iter.get_item()
assert len(self._raw_data_reps) == len(self._raw_data_start_index)
idx = bisect.bisect_left(self._raw_data_start_index, target_index)
if (len(self._raw_data_reps) > 0 and
(len(self._raw_data_reps) == idx
or self._raw_data_reps[idx].index.start_index > target_index)):
idx -= 1
if idx < 0:
raise IndexError(
"target_index {} is too samll".format(target_index))
if idx < len(self._raw_data_reps):
if self._raw_data_iter is None:
self._raw_data_iter = create_raw_data_iter()
self._raw_data_iter.reset_iter(self._raw_data_reps[idx])
self._raw_data_iter.seek_to_target(target_index)
assert self._raw_data_iter.get_index() <= target_index
if self._raw_data_iter.get_index() == target_index:
item = self._raw_data_iter.get_item()
return target_index, item
idx += 1
if idx != len(self._raw_data_reps):
raise IndexError('raw data index is not consecutive')
next_rep = self._raw_data_manager.get_raw_data_rep_by_index(idx)
if next_rep is None:
self._finished = True
raise StopIteration("all raw data finished")
if next_rep.HasField('unindexed'):
start_index = 0
if self._raw_data_iter is not None:
start_index = self._raw_data_iter.get_index() + 1
self._index_raw_data_rep(idx, start_index)
return self._seek_internal(target_index)
def _next_internal(self):
if not self._finished:
try:
item = None
if self._fiter is None:
self._fiter = create_raw_data_iter(self._reader_options)
meta = visitor.IndexMeta(0, 0, self._fpath)
self._fiter.reset_iter(meta, True)
item = self._fiter.get_item()
else:
_, item = next(self._fiter)
assert item is not None
return SortRunReader.MergeItem(item, self._reader_index,
self._comparator)
except StopIteration:
self._finished = True
raise StopIteration("%s has been iter finished" % self._fpath)
def _new_iter(self):
return create_raw_data_iter(self._raw_data_options)