def test_DictFilterKeysMatchingRegex(self):
test_dict = {
"foo.h": ["bar.h"],
"foo.cc": ["foo.h", "bar.h"],
"bar.h": ["baz.h", "qux.h"],
"bad/bad.h": ["baz.h", "foo.h"]
}
test_regexes = [r"foo\..*", r"bad/.*"]
expected_output = {"bar.h": ["baz.h", "qux.h"]}
output = common_utils.DictFilterKeysMatchingRegex(
test_dict, test_regexes)
self.assertEqual(expected_output, output)
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 __init__(self,
database_filename,
including_files_filters,
included_files_to_limit_to=None):
self.inclusions_db = inclusions_database.ReadInclusionsDbFromDisk(
database_filename)
if included_files_to_limit_to:
inclusions_database.LimitToSpecifiedIncludedFiles(
self.inclusions_db, included_files_to_limit_to)
# TODO: Maybe this function should move to this class?
including_file_dict = inclusions_database.FilterOutIncludedFilesAsKeys(
self.inclusions_db)
including_file_dict = common_utils.DictFilterKeysMatchingRegex(
including_file_dict, including_files_filters)
self.including_file_dict = including_file_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 FilterOutIncludedFilesAsKeys(inclusions_db):
output_dict = common_utils.DictFilterKeysMatchingRegex(
inclusions_db["including_to_included"],
IncludedFilesRegexes(inclusions_db))
return output_dict