def GenerateNumInclusionsForFilterFunction(self, key_filter_function):
keys = [
k for k in self.including_file_dict.keys()
if key_filter_function(k)
]
output_dict = {}
for key in keys:
output_dict[key] = self.including_file_dict[key]
output_dict = common_utils.DictListValuesToSums(output_dict)
output_dict = common_utils.DictWithTotal(output_dict)
return output_dict
def GenerateGroupSizesForFilters(self,
key_partition_function,
filters=None):
if filters is None:
filters = []
including_file_dict = common_utils.DictFilterKeysMatchingRegex(
self.including_file_dict, filters)
feature_groups = common_utils.DictPartitionKeys(
including_file_dict, key_partition_function)
feature_dict = {}
for feature, including_files in feature_groups.items():
feature_dict[feature] = len(including_files)
feature_dict = common_utils.DictWithTotal(feature_dict)
return feature_dict
def GenerateGroupNumInclusionsForFilters(self,
key_partition_function,
filters=None):
if filters is None:
filters = []
including_file_dict = common_utils.DictFilterKeysMatchingRegex(
self.including_file_dict, filters)
feature_groups = common_utils.DictPartitionKeys(
including_file_dict, key_partition_function)
feature_dict = {}
including_file_dict = common_utils.DictListValuesToSums(
including_file_dict)
for feature, including_files in feature_groups.items():
feature_dict[feature] = 0
for including_file in including_files:
feature_dict[feature] += including_file_dict[including_file]
feature_dict = common_utils.DictWithTotal(feature_dict)
return feature_dict
def test_DictWithTotal(self):
test_dict = {"key1": 2, "key2": 0, "key3": 4}
expected_output = {"key1": 2, "key2": 0, "key3": 4, "total": 6}
output = common_utils.DictWithTotal(test_dict)
self.assertEqual(expected_output, output)
def GenerateAnalysis(self):
output_dict = common_utils.DictListValuesToSums(
self.included_file_dict)
output_dict = common_utils.DictWithTotal(output_dict)
return output_dict