revert back to nl2bash

nl2bash_m.py

@@ -107,45 +107,39 @@ class nl2bash_m(evaluate.Metric):
             predictions = np.char.translate(predictions, table=repl_table)
             references = np.char.translate(references, table=repl_table)
 
-
+        
         final_score = 0
-        for pred, refs in zip(predictions, references):
+        for pred, refs in zip(predictions, references): 
             best_score = 0
-            for ref in refs:
-                if pred == ref:
-                    best_score = 1
-            final_score += best_score
-
-            # if len(pred) == 0 and min([len(ref) for ref in refs]) == 0:
-            #     score = 1
-            # elif len(pred) == 0 or min([len(ref) for ref in refs]) == 0:
-            #     score = 0
-            # else:
-            #     best_score = 0
-            #     for ref in refs:
-            #         pred_words, ref_words = pred.split(), ref.split()
+            if len(pred) == 0 and min([len(ref) for ref in refs]) == 0:
+                best_score = 1
+            elif len(pred) == 0 or min([len(ref) for ref in refs]) == 0:
+                best_score = 0
+            else: 
+                for ref in refs:
+                    pred_words, ref_words = pred.split(), ref.split()
 
                     
-            #         # Get the cmd of predicted and ref 
-            #         cmd_corr = 1 if pred_words.pop(0)==ref_words.pop(0) else 0
+                    # Get the cmd of predicted and ref 
+                    cmd_corr = 1 if pred_words.pop(0)==ref_words.pop(0) else 0
 
-            #         # Get the option of predicted and ref
-            #         pred_option = [ x for x in pred_words if x[0] == '-']
-            #         ref_option = [ x for x in ref_words if x[0] == '-']
+                    # Get the option of predicted and ref
+                    pred_option = [ x for x in pred_words if x[0] == '-']
+                    ref_option = [ x for x in ref_words if x[0] == '-']
                     
-            #         # Get the arguments of predicted and ref
-            #         pred_args = [ x for x in pred_words if x[0] != '-']
-            #         ref_args = [ x for x in ref_words if x[0] != '-']
+                    # Get the arguments of predicted and ref
+                    pred_args = [ x for x in pred_words if x[0] != '-']
+                    ref_args = [ x for x in ref_words if x[0] != '-']
 
-            #         # Calculate scores
-            #         cmd_score = cmd_weight * cmd_corr
-            #         opt_score = opt_weight * self.get_score(pred_option, ref_option)
-            #         arg_score = arg_weight * self.get_score(pred_args, ref_args)
+                    # Calculate scores
+                    cmd_score = cmd_weight * cmd_corr
+                    opt_score = opt_weight * self.get_score(pred_option, ref_option)
+                    arg_score = arg_weight * self.get_score(pred_args, ref_args)
 
-            #         score = cmd_score + opt_score + arg_score 
-            #         best_score = max(best_score, score)
+                    score = cmd_score + opt_score + arg_score 
+                    best_score = max(best_score, score)
 
-            # final_score += best_score
+            final_score += best_score
 
         final_score = final_score/len(predictions)