def test_should_add_all_misuse_ids(self):
uut = DataEntityLists(["p.v1.m1", "p.v2.m1", "p.v2.m2", "p"], [])
actual = uut.get_misuse_white_list("p.v2")
expected = ["p.v2.m1", "p.v2.m2"]
assert_equals(expected, actual)
def test_should_add_all_project_ids(self):
uut = DataEntityLists(["p1.v.m", "p2.v", "p3"], [])
actual = uut.get_project_white_list()
expected = ["p1", "p2", "p3"]
assert_equals(expected, actual)
def test_should_add_all_version_ids(self):
uut = DataEntityLists(["p.v1.m", "p.v2", "p"], [])
actual = uut.get_version_white_list("p")
expected = ["p.v1", "p.v2"]
assert_equals(expected, actual)
def __is_filtered(project_id: str,
data_entity_lists: DataEntityLists) -> bool:
white_list = data_entity_lists.get_project_white_list()
black_list = data_entity_lists.black_list
is_whitelisted = not white_list or project_id in white_list
is_blacklisted = project_id in black_list
return is_blacklisted or not is_whitelisted
def __is_filtered(version_id: str, misuse_id: str,
data_entity_lists: DataEntityLists) -> bool:
white_list = data_entity_lists.get_misuse_white_list(version_id)
black_list = data_entity_lists.black_list
is_whitelisted = not white_list or misuse_id in white_list
is_blacklisted = misuse_id in black_list
return is_blacklisted or not is_whitelisted
def test_filters_blacklisted(self):
project = create_project("-id-", base_path=self.temp_dir)
create_file(project._project_file)
uut = CollectProjectsTask(self.temp_dir)
actual = uut.run(DataEntityLists([], ["-id-"]))
assert_equals([], actual)
def __is_filtered(project_id: str, version: ProjectVersion,
data_entity_lists: DataEntityLists) -> bool:
version_id = version.id
white_list = data_entity_lists.get_version_white_list(project_id)
black_list = data_entity_lists.black_list
is_whitelisted = not white_list or version_id in white_list
is_blacklisted = version_id in black_list
return is_blacklisted or not is_whitelisted
def test_finds_all_projects(self):
p1 = create_project("p1", base_path=self.temp_dir)
create_file(p1._project_file)
p2 = create_project("p2", base_path=self.temp_dir)
create_file(p2._project_file)
uut = CollectProjectsTask(self.temp_dir)
actual = uut.run(DataEntityLists([], []))
assert_equals([p1, p2], actual)
def test_filters_non_whitelisted(self):
project = create_project("-project-")
version = create_version("-version-", misuses=[], project=project)
create_misuse("-id-", version=version)
uut = CollectMisusesTask()
actual = uut.run(
version, DataEntityLists(["-project-.-version-.-other-id-"], []))
assert_equals([], actual)
def test_whitelisted_version(self):
misuse = create_misuse("-id-")
project = create_project("-project-")
version = create_version("-version-",
misuses=[misuse],
project=project)
uut = CollectMisusesTask()
actual = uut.run(version, DataEntityLists(["-project-.-version-"], []))
assert_equals([misuse], actual)
def test_finds_all_misuses(self):
m1 = create_misuse("-m1-")
m2 = create_misuse("-m2-")
project = create_project("-project-")
version = create_version("-version-",
misuses=[m1, m2],
project=project)
uut = CollectMisusesTask()
actual = uut.run(version, DataEntityLists([], []))
assert_equals([m1, m2], actual)
def __init__(self, config):
self.config = config
white_list = []
black_list = []
if 'white_list' in config:
white_list.extend(config.white_list)
if 'black_list' in config:
black_list.extend(config.black_list)
if 'datasets' in config:
for dataset in config.datasets:
white_list.extend(
get_white_list(config.datasets_file_path, dataset))
self.data_entity_lists = DataEntityLists(white_list, black_list)
_INDEX_PATH = os.path.join(__MUBENCH_ROOT_PATH, "checkouts-xp", "index.csv")
class PrintIndexTask:
def run(self, project: Project, version: ProjectVersion, misuse: Misuse):
print("{}\t{}\t{}\t{}\t{}\t{}\t{}".format(
project.id, version.version_id, misuse.misuse_id,
':'.join(version.source_dirs), misuse.location.file,
misuse.location.method, "\t".join(misuse.apis)),
file=open(_INDEX_PATH, "a"))
datasets = sys.argv[1:]
white_list = []
for dataset in datasets:
white_list.extend(get_white_list(__MUBENCH_DATASETS_FILE, dataset.lower()))
initial_parameters = [DataEntityLists(white_list, [])]
runner = TaskRunner([
CollectProjectsTask(__MUBENCH_DATA_PATH),
CollectVersionsTask(False),
CollectMisusesTask(),
PrintIndexTask()
])
if os.path.exists(_INDEX_PATH):
os.remove(_INDEX_PATH)
runner.run(*initial_parameters)