update dataset, tools, and Readme.md.

Author: Messi-Q

Performance_evaluation.py

@@ -2,7 +2,7 @@
 from os.path import join as pjoin
 from sklearn.metrics import confusion_matrix
 
-data_dir = "./logs/timestamp_results/"
+data_dir = "./comparison/reentrancy/"
 
 
 # Performance evaluation, tools : SmartCheck, Securify, Mythril, Oyente
@@ -13,11 +13,11 @@ def read_label(fpath, line_parse_fn=None):
     return data
 
 
-ground_truth = read_label("ground_truth_185.txt", line_parse_fn=lambda s: int(float(s.strip())))
-securify_label = read_label("securify_label_185.txt", line_parse_fn=lambda s: int(float(s.strip())))
-smartcheck_label = read_label("smartcheck_label_185.txt", line_parse_fn=lambda s: int(float(s.strip())))
-myth_label = read_label("myth_label_185.txt", line_parse_fn=lambda s: int(float(s.strip())))
-oyente_label = read_label("oyente_label_185.txt", line_parse_fn=lambda s: int(float(s.strip())))
+ground_truth = read_label("ground_truth.txt", line_parse_fn=lambda s: int(float(s.strip())))
+securify_label = read_label("securify_label.txt", line_parse_fn=lambda s: int(float(s.strip())))
+smartcheck_label = read_label("smartcheck_label.txt", line_parse_fn=lambda s: int(float(s.strip())))
+myth_label = read_label("myth_label.txt", line_parse_fn=lambda s: int(float(s.strip())))
+oyente_label = read_label("oyente_label.txt", line_parse_fn=lambda s: int(float(s.strip())))
 
 # tn, fp, fn, tp = confusion_matrix(ground_truth, securify_label).ravel()
 tn, fp, fn, tp = confusion_matrix(ground_truth, smartcheck_label).ravel()

SmConVulDetector.py

@@ -32,12 +32,12 @@
 torch.cuda.manual_seed_all(args.seed)
 rnd_state = np.random.RandomState(args.seed)
 
-print('Loading data...')
+print('Loading training_data...')
 
 
 class DataReader():
     """
-    Class to read the txt files containing all data of the dataset
+    Class to read the txt files containing all training_data of the dataset
     """
 
     def __init__(self, data_dir, rnd_state=None, use_cont_node_attr=False, folds=n_folds):
@@ -111,8 +111,8 @@ def __init__(self, data_dir, rnd_state=None, use_cont_node_attr=False, folds=n_f
         for u in np.unique(features_all):
             print('feature {}, count {}/{}'.format(u, np.count_nonzero(features_all == u), len(features_all)))
 
-        N_graphs = len(labels)  # number of samples (graphs) in data
-        assert N_graphs == len(data['adj_list']) == len(features_onehot), 'invalid data'
+        N_graphs = len(labels)  # number of samples (graphs) in training_data
+        assert N_graphs == len(data['adj_list']) == len(features_onehot), 'invalid training_data'
 
         # Create test sets first
         train_ids, test_ids = split_ids(rnd_state.permutation(N_graphs), folds=folds)
@@ -143,7 +143,7 @@ def read_graph_adj(self, fpath, nodes, graphs):
             node1 = int(edge[0].strip()) - 1  # -1 because of zero-indexing in our code
             node2 = int(edge[1].strip()) - 1
             graph_id = nodes[node1]
-            assert graph_id == nodes[node2], ('invalid data', graph_id, nodes[node2])
+            assert graph_id == nodes[node2], ('invalid training_data', graph_id, nodes[node2])
             if graph_id not in adj_dict:
                 n = len(graphs[graph_id])
                 adj_dict[graph_id] = np.zeros((n, n))
@@ -183,7 +183,7 @@ def read_node_features(self, fpath, nodes, graphs, fn):
         return node_features_lst
 
 
-datareader = DataReader(data_dir='./data/%s/' % args.dataset, rnd_state=rnd_state,
+datareader = DataReader(data_dir='./training_data/%s/' % args.dataset, rnd_state=rnd_state,
                         use_cont_node_attr=args.use_cont_node_attr, folds=args.folds)
 
 # train and test
@@ -239,7 +239,7 @@ def train(train_loader):
             for i in range(len(data)):
                 data[i] = data[i].to(args.device)
             optimizer.zero_grad()
-            # output = model(data[0], data[1])  # when model is gcn_origin or gat, use this
+            # output = model(training_data[0], training_data[1])  # when model is gcn_origin or gat, use this
             output = model(data)  # when model is gcn_modify, use this
             loss = loss_fn(output, data[4])
             loss.backward()
@@ -266,7 +266,7 @@ def test(test_loader):
         for batch_idx, data in enumerate(test_loader):
             for i in range(len(data)):
                 data[i] = data[i].to(args.device)
-            # output = model(data[0], data[1])  # when model is gcn_origin or gat, use this
+            # output = model(training_data[0], training_data[1])  # when model is gcn_origin or gat, use this
             output = model(data)  # when model is gcn_modify, use this
             loss = loss_fn(output, data[4], reduction='sum')
             test_loss += loss.item()

…aph_data/reentrancy/reentrancy_label.txt comparison/reentrancy/ground_truth.txtgraph_data/reentrancy/reentrancy_label.txt renamed to comparison/reentrancy/ground_truth.txt graph_data/reentrancy/reentrancy_label.txt renamed to comparison/reentrancy/ground_truth.txt

@@ -0,0 +1,5 @@
+FALLBACK FUN0 0 NULL
+FUN0 FALLBACK 0 NULL
+FUN1 FUN1 0 NULL
+FUN2 FUN2 0 NULL
+FUN3 FUN3 0 NULL

…gs/reentrancy_results/myth_label_185.txt comparison/reentrancy/myth_label.txtlogs/reentrancy_results/myth_label_185.txt renamed to comparison/reentrancy/myth_label.txt logs/reentrancy_results/myth_label_185.txt renamed to comparison/reentrancy/myth_label.txt

@@ -0,0 +1,5 @@
+FALLBACK FUN0 0 NULL
+FUN0 FALLBACK 0 NULL
+FUN1 FUN1 0 NULL
+FUN3 FUN3 0 NULL
+FUN2 VAR0 1 FOR

…/reentrancy_results/oyente_label_185.txt comparison/reentrancy/oyente_label.txtlogs/reentrancy_results/oyente_label_185.txt renamed to comparison/reentrancy/oyente_label.txt logs/reentrancy_results/oyente_label_185.txt renamed to comparison/reentrancy/oyente_label.txt

@@ -0,0 +1,5 @@
+FALLBACK FUN0 0 NULL
+FUN0 FALLBACK 0 NULL
+FUN1 FUN1 0 NULL
+FUN2 FUN2 0 NULL
+FUN3 FUN3 0 NULL