def run_mutation_experiment_fixed_size(project_root,
module_under_test_path,
test_suite_sizes_count=30,
test_suite_sizes=None,
max_trace_size=10,
timeout=None,
coverage_metrics=None,
support=100):
logger.debug("Running mutation experiment (fixed size) for {module}",
module=module_under_test_path)
trace_reader = TraceReader(project_root)
not_failing_node_ids = trace_reader.get_not_failing_node_ids(
module_under_test_path)
mutation_scoring_function, total_mutants_count = get_mutation_scoring_function(
project_root,
module_under_test_path,
not_failing_node_ids=not_failing_node_ids,
timeout=timeout)
points = generic_experiment_size(
project_root,
module_under_test_path, [mutation_scoring_function],
test_suite_sizes_count=test_suite_sizes_count,
test_suite_sizes=test_suite_sizes,
max_trace_size=max_trace_size,
coverage_metrics=coverage_metrics,
node_ids=tuple(not_failing_node_ids),
support=support)
df = pd.DataFrame(
data=points,
columns=[SUITE_SIZE, METRIC, MUTATION_SCORE, SUITE_COVERAGE])
return df, total_mutants_count
def run_mutation_experiment_fixed_coverage(project_root,
module_under_test_path,
coverage_boundaries_count=20,
max_trace_size=10,
timeout=None,
coverage_metrics=None,
support=100):
logger.debug("Running mutation experiment (fixed coverage) for {module}",
module=module_under_test_path)
trace_reader = TraceReader(project_root)
not_failing_node_ids = trace_reader.get_not_failing_node_ids(
module_under_test_path)
scoring_function, total_mutants_count = get_mutation_scoring_function(
project_root,
module_under_test_path,
not_failing_node_ids,
timeout=timeout)
points = generic_experiment_coverage(
project_root,
module_under_test_path, [scoring_function],
coverage_boundaries_count=coverage_boundaries_count,
max_trace_size=max_trace_size,
coverage_metrics=coverage_metrics,
support=support)
df = pd.DataFrame(
data=points,
columns=[SUITE_SIZE, METRIC, MUTATION_SCORE, SUITE_COVERAGE])
df = bin_zero_to_one_column_to_percent(df, SUITE_COVERAGE,
SUITE_COVERAGE_BIN,
coverage_boundaries_count)
return df, total_mutants_count
def test_branch_coverage(self):
project_root = CLEAN_LINKED_LIST_ROOT
# trace_root = create_new_temp_dir()
trace_root = CLEAN_LINKED_LIST_ROOT
exclude_folders = ["venv", "dataset"]
cfg = ProjectCFG.create_from_path(project_root,
exclude_folders=exclude_folders,
use_cached_if_possible=False)
thorough.run_tests(CLEAN_LINKED_LIST_ROOT, trace_root, exclude_folders)
trace_reader = TraceReader(trace_root)
ll_py = str(CLEAN_LINKED_LIST_LL)
ll_py_cfg = cfg.module_cfgs[ll_py]
total_exercised = set()
available_branches = ll_py_cfg.branches
for node_id, path in zip(*trace_reader.get_traces_for(ll_py)):
df, size = read_as_dataframe(path)
covered = find_covered_branches(df, ll_py_cfg.branches)
total_exercised.update(covered)
print("Coverage")
print_percent("Branches covered", total_exercised, ll_py_cfg.branches)
print(available_branches)
print(total_exercised)
not_exercised_branches = set(available_branches) - set(total_exercised)
print("Not exercised branches total ({}): ".format(len(not_exercised_branches)), not_exercised_branches)
self.assertEqual(13, len(not_exercised_branches))
def run_real_bugs_experiment_fixed_size(project_root,
module_under_test_path,
revealing_node_ids,
test_suite_sizes_count=30,
test_suite_sizes=None,
max_trace_size=10,
coverage_metrics=None,
support=100):
logger.debug("Running real project experiment for {module} (fixed size)",
module=module_under_test_path)
revealing_node_ids = set(revealing_node_ids)
total_revealing_node_ids = len(revealing_node_ids)
trace_reader = TraceReader(project_root)
not_failing_node_ids = trace_reader.get_not_failing_node_ids(
module_under_test_path)
scoring_function = get_bugs_scoring_function(revealing_node_ids)
points = generic_experiment_size(
project_root,
module_under_test_path, [scoring_function],
test_suite_sizes_count=test_suite_sizes_count,
test_suite_sizes=test_suite_sizes,
max_trace_size=max_trace_size,
coverage_metrics=coverage_metrics,
node_ids=set(not_failing_node_ids) | set(revealing_node_ids),
support=support)
df = pd.DataFrame(
data=points,
columns=[SUITE_SIZE, METRIC, BUG_REVEALED_SCORE, SUITE_COVERAGE])
return df, total_revealing_node_ids
def test_intra_method_pairs(self):
len_pairs = [
9, 8, 13, 19, 3, 20, 8, 5, 18, 16, 12, 19, 3, 18, 18, 24, 9, 2, 8
]
project_root = CLEAN_LINKED_LIST_ROOT
trace_root = create_new_temp_dir()
exclude_folders = ["venv"]
cfg = ProjectCFG.create_from_path(project_root,
exclude_folders=exclude_folders)
thorough.run_tests(CLEAN_LINKED_LIST_ROOT, trace_root, exclude_folders)
trace_reader = TraceReader(trace_root)
cppvi = VarIndexFactory.new_cpp_index(project_root, trace_root)
ll_py = str(CLEAN_LINKED_LIST_LL)
def get_pairs(trace_file_path):
np_array, _ = read_as_np_array(trace_file_path)
idx_pairs = analyze_trace_w_index(trace_file_path, cppvi)
def rename_vars(s):
return {(el[0], el[1]) for el in s}
idx_pairs = rename_vars(idx_pairs)
return idx_pairs
node_ids, paths = trace_reader.get_traces_for(ll_py)
for node_id, path, expected_pairs_count in zip(node_ids, paths,
len_pairs):
pairs = get_pairs(path)
self.assertEqual(
expected_pairs_count, len(pairs),
"Pairs count don't match for test case: {}".format(node_id))
def trace_this(function,
project_root=PROJECT_ROOT,
trace_root=TEMP_DIRECTORY,
args=None,
kwargs=None):
if args is None:
args = []
if kwargs is None:
kwargs = {}
target_source_file = inspect.getsourcefile(function)
keep = [str(project_root)]
exclude = [str(project_root / "venv")]
trace_folder = trace_root
tracer = Tracer(keep, exclude, trace_folder_parent=str(trace_folder))
trace_name = function.__name__
tracer.start(trace_name)
function(*args, **kwargs)
tracer.stop()
tracer.full_stop()
trace_reader = TraceReader(trace_root)
return trace_reader.trace_path(trace_name, target_source_file)
def test_failed_tests_recorded(self):
project_root = BUGGY_LINKED_LIST_ROOT
exclude_folders = ["venv"]
thorough_location = str(PROJECT_ROOT / "thorough.py")
subprocess.run(f"python3 {thorough_location} -t --trace_dir {str(project_root)}", cwd=project_root, shell=True)
trace_root = project_root
trace_reader = TraceReader(trace_root)
failed_cases = trace_reader.read_failed_test_cases()
self.assertEqual(1, len(failed_cases))
expected_failed_test_case = "tests/test_linked_list_module.py::LinkedListTest::test_append_on_removed"
self.assertEqual(expected_failed_test_case, failed_cases[0])
def __init__(self,
trace_root,
project_root,
exclude_folders=None,
max_trace_size=None):
self.project_cfg = ProjectCFG.create_from_path(
project_root,
exclude_folders=exclude_folders,
use_cached_if_possible=True)
self.max_trace_size = max_trace_size
self.trace_root = trace_root
self.trace_reader = TraceReader(trace_root)
self.vi = VarIndexFactory.new_py_index(project_root, trace_root)
self.cppvi = VarIndexFactory.new_cpp_index(project_root, trace_root)
def test_inter_method_pairs_test_interclass_pairs(self):
len_im_pairs = [
0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
]
len_ic_pairs = [
5, 2, 5, 4, 1, 4, 2, 1, 5, 5, 7, 4, 1, 10, 7, 5, 9, 1, 4
]
project_root = CLEAN_LINKED_LIST_ROOT
trace_root = create_new_temp_dir()
exclude_folders = ["venv"]
cfg = ProjectCFG.create_from_path(project_root,
exclude_folders=exclude_folders)
thorough.run_tests(CLEAN_LINKED_LIST_ROOT, trace_root, exclude_folders)
trace_reader = TraceReader(trace_root)
cppvi = VarIndexFactory.new_py_index(project_root, trace_root)
ll_py = str(CLEAN_LINKED_LIST_LL)
def get_pairs(trace_file_path):
np_array, _ = read_as_np_array(trace_file_path)
scopes = read_scopes_for_trace_file(trace_file_path)
im_pairs, ic_pairs = analyze(trace_file_path, cppvi, scopes)
def rename_vars(s):
return {(el[0], el[1]) for el in s}
im_pairs = rename_vars(im_pairs)
ic_pairs = rename_vars(ic_pairs)
return im_pairs, ic_pairs
node_ids, paths = trace_reader.get_traces_for(ll_py)
for node_id, path, expected_im_len, expected_ic_len in zip(
node_ids, paths, len_im_pairs, len_ic_pairs):
im_pairs, ic_pairs = get_pairs(path)
self.assertEqual(
expected_im_len, len(im_pairs),
"Intermethod pairs count don't match for test case: {}".format(
node_id))
self.assertEqual(
expected_ic_len, len(ic_pairs),
"Intermethod pairs count don't match for test case: {}".format(
node_id))
def combined_experiment_fixed_coverage(project_root,
module_under_test_path,
revealing_node_ids,
coverage_boundaries_count=20,
max_trace_size=10,
coverage_metrics=None,
support=100,
timeout=None):
logger.debug(
"Running real project experiment for {module} (fixed coverage)",
module=module_under_test_path)
trace_reader = TraceReader(project_root)
not_failing_node_ids = trace_reader.get_not_failing_node_ids(
module_under_test_path)
mutation_scoring_function, total_mutants_count = get_mutation_scoring_function(
project_root,
module_under_test_path,
not_failing_node_ids=not_failing_node_ids,
timeout=timeout)
bugs_scoring_function = get_bugs_scoring_function(revealing_node_ids)
total_bugs_count = len(revealing_node_ids)
points = generic_experiment_coverage(
project_root,
module_under_test_path,
[mutation_scoring_function, bugs_scoring_function],
coverage_boundaries_count=coverage_boundaries_count,
max_trace_size=max_trace_size,
coverage_metrics=coverage_metrics,
node_ids=set(revealing_node_ids) | set(not_failing_node_ids),
support=support)
columns = [
SUITE_SIZE, METRIC, MUTATION_SCORE, BUG_REVEALED_SCORE, SUITE_COVERAGE
]
df = pd.DataFrame(data=points, columns=columns)
df = bin_zero_to_one_column_to_percent(df, SUITE_COVERAGE,
SUITE_COVERAGE_BIN,
coverage_boundaries_count)
return df, total_mutants_count, total_bugs_count
def test_mutate_linked_list_module(self):
project_root = Path(
__file__).parent.parent.parent / "dataset/linked_list_clean"
module_under_test_path = project_root / "core" / "ll.py"
trace_reader = TraceReader(project_root)
not_failing_node_ids = trace_reader.get_not_failing_node_ids(
module_under_test_path)
killed, total = killed_mutants(
project_root=str(project_root),
module_under_test_path=str(module_under_test_path),
not_failing_node_ids=not_failing_node_ids,
timeout=None)
s = set()
for m in killed.values():
s.update(m)
self.assertEqual(46, total)
self.assertEqual(30, len(s))
if len(failing_on_assertion_node_ids) > 0:
repo.status = STATUS_NO_MODULE_CFG
repo.save()
repo_stat.mark_repo_as_good(manager)
logger.warning("Found node ids which failed on assertion!")
graphs_path = graphs_path_parent
module_under_test_path = manager.test_module_path
logger.info("Running the experiment for module {p}",
p=module_under_test_path)
st = time()
buggy_project.run_command(
f"python3 {thorough_path} -t --trace_dir={buggy_project.path}",
extra_requirements=extra_requirements)
elapsed_tracing = int(time() - st)
trace_reader = TraceReader(buggy_project.path)
module_under_test_absolute = os.path.join(
buggy_project.path, module_under_test_path)
covering_node_ids, paths = trace_reader.get_traces_for(
module_under_test_absolute)
covering_node_ids = set(covering_node_ids)
bad_node_ids = after - failing_on_assertion_node_ids
covering_node_ids -= bad_node_ids
not_failing_node_ids = covering_node_ids - failing_on_assertion_node_ids
not_failing_node_ids_as_params = " ".join(
not_failing_node_ids)
return_code = buggy_project.run_command(
f"pytest {not_failing_node_ids_as_params}")
if return_code == 0:
logger.warning("Test cases can be run individually")
out_folder = maybe_expand(args.out)
module = args.module
timeout = args.timeout
revealing_node_ids = args.revealing_node_ids
exclude_folders_tracing = COMMON_EXCLUDE
thorough.run_tests(project_root,
trace_root,
exclude_folders_tracing=exclude_folders_tracing,
exclude_folders_collection=None,
show_time_per_test=False,
quiet=False,
deselect_tests=None)
trace_reader = TraceReader(trace_root)
failed_node_ids = trace_reader.read_failed_test_cases()
module = maybe_expand(module)
coverage_metrics = (CoverageMetric.STATEMENT, CoverageMetric.BRANCH,
CoverageMetric.ALL_PAIRS)
# coverage_metrics = (CoverageMetric.ALL_PAIRS, CoverageMetric.ALL_C_USES, CoverageMetric.ALL_P_USES)
new_module_path = module_path(out_folder, project_root, module)
shutil.copy(module, new_module_path)
# fixed size
df, total_mutants_count, total_bugs_count = combined_experiment_fixed_size(
project_root=project_root,
module_under_test_path=module,
revealing_node_ids=revealing_node_ids,
def create(project_root, traces_root):
trace_reader = TraceReader(traces_root)
return VarIndex(read_du_index(project_root),
trace_reader.files_mapping.index)
def generic_experiment_size(project_root,
module_under_test_path,
scoring_functions,
test_suite_sizes_count=30,
test_suite_sizes=None,
max_trace_size=10,
coverage_metrics=None,
node_ids=None,
support=100):
logger.debug("Running mutation experiment for {module}",
module=module_under_test_path)
if coverage_metrics is None:
coverage_metrics = DEFAULT_METRICS
cfg = ProjectCFG.create_from_path(project_root,
exclude_folders=COMMON_EXCLUDE)
if node_ids is None:
trace_reader = TraceReader(project_root)
node_ids, paths = trace_reader.get_traces_for(module_under_test_path)
total_cases = len(node_ids)
if test_suite_sizes is None:
if total_cases < test_suite_sizes_count:
test_suite_sizes_count = total_cases
test_suite_sizes = np.arange(test_suite_sizes_count) + 1
test_suite_sizes = [
int(c * total_cases / test_suite_sizes_count)
for c in test_suite_sizes
]
logger.debug("Testing test suites of sizes {ss}", ss=test_suite_sizes)
generator = SuiteGenerator(project_root,
project_root,
COMMON_EXCLUDE,
max_trace_size=max_trace_size)
points = []
no_suites_created_count_max = 3
for metric in coverage_metrics:
no_suites_created_count = 0
for sub_test_suites_size in test_suite_sizes:
logger.debug(
"Test suite size: {sub_test_suites_size}, metric: {metric}",
sub_test_suites_size=sub_test_suites_size,
metric=metric)
suites = generator.fix_sized_suites(
module_under_test_path=module_under_test_path,
coverage_metric=metric,
exact_size=sub_test_suites_size,
n=support,
check_unique_items_covered=False,
test_cases=node_ids)
if not suites:
no_suites_created_count += 1
else:
no_suites_created_count = 0
if no_suites_created_count > no_suites_created_count_max:
break
for suite in suites:
scores = [
scoring_function(suite)
for scoring_function in scoring_functions
]
point = (len(suite.test_cases), str(metric), *scores,
suite.coverage)
points.append(point)
return points