def test_ffmpeg_commiters(self):
res_list = feature_extractor.extract_all_measures_from_file(
"test_data" + os.sep + "FFmpeg_FFmpeg.log", None)[1]["authors_ts"]
oracle_ts = self.oracles['FFmpeg']['commiters'].tolist()
self.assertEqual(len(res_list), len(oracle_ts))
for i, res in enumerate(res_list):
self.assertEqual(res, oracle_ts[i], f'It is {res} but should be {oracle_ts[i]} at {i}')
def test_feature_extraction_mysql(self):
fv_obj = feature_extractor.extract_all_measures_from_file("test_data" + os.sep + "mysql_mysql-server.log", None)[0]["integrations"]
fv = fv_obj.to_dict()
self.assertEqual(923, fv["duration"])
self.assertEqual(140096, fv["sum_y"])
self.assertEqual(347, fv["max_y_pos"])
self.assertEqual(485, fv["max_y"])
self.assertEqual(485, fv["PG_count"])
self.assertEqual(437, fv["NG_count"])
self.assertEqual(-46.2173913043478, fv["avg_NG"])
def test_feature_extraction_pyccflex_integrations(self):
fv_obj = feature_extractor.extract_all_measures_from_file("test_data" + os.sep + "py-ccflex.log", None)[0]["integrations"]
fv = fv_obj.to_dict()
measure = 'integrations'
self.assertEqual(self.feature_oracle('py-ccflex', measure,'duration'), fv["duration"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'sum_y'), fv["sum_y"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'max_y_pos'), fv["max_y_pos"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'max_y'), fv["max_y"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'PG_count'), fv["PG_count"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'NG_count'), fv["NG_count"])
self.assertEqual(self.feature_oracle('py-ccflex', measure,'avg_NG'), fv["avg_NG"])
def test_feature_extraction_ffmpeg_merges(self):
fv_obj = feature_extractor.extract_all_measures_from_file("test_data" + os.sep + "FFmpeg_FFmpeg.log", None)[0]["merges"]
fv = fv_obj.to_dict()
measure = 'merges'
self.assertEqual(self.feature_oracle('FFmpeg', measure,'duration'), fv["duration"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'sum_y'), fv["sum_y"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'max_y_pos'), fv["max_y_pos"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'PG_count'), fv["PG_count"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'NG_count'), fv["NG_count"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'max_y'), fv["max_y"])
self.assertEqual(self.feature_oracle('FFmpeg', measure,'avg_NG'), fv["avg_NG"])
def features(args):
if len(args) < 3:
return mustIncludeRepoLocation()
if (len(args) < 4):
return mustIncludeGitLogLocation()
if (len(args) < 5):
return mustIncludeOutputLocation()
if (len(args) < 6):
return mustIncludeTSLocation()
repoDir = args[2]
output_git_file = args[3]
output_file = args[4]
output_ts_file = args[5]
if repoDir != "-log_file":
generateGitLog(repoDir, output_git_file)
arr = feature_extractor.extract_all_measures_from_file(
output_git_file, output_ts_file)
with open(output_file, "w") as file:
csv_writer = csv.writer(file,
quoting=csv.QUOTE_NONE,
lineterminator='\n')
ready_to_write_headers = ["measure"
] + arr[0]["integrations"].keys_order()
csv_writer.writerow(ready_to_write_headers)
for measure in sorted(arr[0].keys()):
feature_vector = arr[0][measure]
feature_vector_values = []
feature_vector_dict = feature_vector.to_dict()
for feature in sorted(feature_vector_dict.keys()):
feature_vector_values.append(feature_vector_dict[feature])
ready_to_write_list = [measure] + feature_vector_values
csv_writer.writerow(ready_to_write_list)
return operationComplete()
