def aggregate(dataset, timeline, aggregator=None, **aggregator_param):
# nothing to aggregate if dataset and/or timeline are/is empty
if dataset.isEmpty() or timeline.isEmpty():
return PYODataset.Empty()
# params will be passed to the aggregator function
# it is meant to contain the user-provided parameters
# and additional parameters (such as the current timerange)
params = {}
for p, param in enumerate(aggregator_param):
params[param] = aggregator_param[param]
nTimeranges = timeline.getNumberOfTimeranges()
datas = [None] * nTimeranges
number = np.zeros((nTimeranges, ), dtype=np.int32)
for t, timerange in enumerate(timeline):
subdataset = sub(dataset, timerange)
# try and aggregate subdataset:
# some aggregator may fail in case of empty dataset
# some other may return something
try:
if aggregator == None:
datas[t] = np.copy(subdataset.data)
else:
params['timerange'] = timerange
datas[t] = aggregator(subdataset, **params)
number[t] = datas[t].shape[0]
except Exception, e:
raise e
datas[t] = None
number[t] = 0
def __init__(self, data, number, timeline):
super(PYODataset, self).__init__()
if data == None or data.shape == (0,) or number == None or timeline == None or number.shape[0] != timeline.getNumberOfTimeranges():
self._data = np.array(())
self._number = np.array((), dtype=np.int32)
self._position = np.array((), dtype=np.int32)
self._timeline = PYOTimeline.Empty()
return
self._data = data
self._timeline = timeline
stored, dimension = self._data.shape
self._number = np.array(np.copy(number), dtype=np.int32)
self._position = np.copy(self._number)
self._position[1:] = self._number[:-1].cumsum()
self._position[0] = 0
