def groupby(tsastat, index_keys):
"""
group tsastat by index_keys, which are a subset of the original index_keys
the grouping functions are predefined, it makes no sense to make this variable
parameters:
tsastat <TimeseriesArrayStats>
index_keys <tuple>
returns:
<TimeseriesArrayStats>
"""
group_funcs = {
"sum" : lambda a, b: a + b,
"avg" : lambda a, b: (a + b) / 2,
"min" : min,
"max" : max,
"count" : lambda a, b: a + b,
"std" : lambda a, b: (a + b) / 2,
"median" : lambda a, b: (a + b) / 2,
"mean" : lambda a, b: (a + b) / 2,
"last" : lambda a, b: (a + b) / 2,
"first" : lambda a, b: (a + b) / 2,
}
try:
assert all(index_key in tsastat.index_keys for index_key in index_keys)
except AssertionError:
logging.error("All given index_keys have to be in tsastat.index_keys")
return
# intermediate data
data = {}
for key in tsastat.keys():
key_dict = dict(zip(tsastat.index_keys, key))
group_key = tuple((key_dict[key] for key in index_keys))
if group_key not in data:
data[group_key] = tsastat[key].stats
else:
# there is something to group
for value_key in tsastat[key].keys():
for stat_func, value in tsastat[key][value_key].items():
# group values by function
grouped_value = group_funcs[stat_func](value, data[group_key][value_key][stat_func])
# store
data[group_key][value_key][stat_func] = grouped_value
# get to same format as TimeseriesArrayStats.to_json returns
outdata = [tsastat.index_keys, tsastat.value_keys, ]
outdata.append([(key, json.dumps(value)) for key, value in data.items()])
# use TimeseriesArrayStats.from_json to get to TimeseriesArrayStats
# object
new_tsastat = TimeseriesArrayStats.from_json(json.dumps(outdata))
return new_tsastat
def get_tsastats(self, project, tablename, datestring):
"""
get TimeseriesStatsArray object for this particular project/tablename/datestring combination
parameters:
project <str>
tablename <str>
datestring <str>
returns:
<TimeseriesStatsArray>
"""
uri_params = {
"project" : project,
"tablename" : tablename,
"datestring" : datestring,
}
query_params = {}
data = self.__get_raw_data("get_tsastats", uri_params, query_params)
tsastats = TimeseriesArrayStats.from_json(data)
return tsastats
