put functions outside of getLogMetric for increased efficiency

logmetric.py

@@ -97,6 +97,30 @@ class LogMetric(evaluate.Metric):
         # TODO: Download external resources if needed
         pass
 
+    # Jaccard Similarity to measure closeness of two log-messages
+    def get_jaccard_similarity(set1, set2):
+        intersection = set1.intersection(set2)
+        union = set1.union(set2)
+        return len(intersection) / len(union)
+    
+    # A score depending on the difference in length of two sentences
+    def get_length_score(sentence1, sentence2):
+        s1len = len(sentence1)
+        s2len = len(sentence2)
+
+        return 1 - (abs(s1len - s2len) / max(s1len, s2len))
+
+    # Combine a weighted average of different scores
+    def get_overall_similarity(sentence1, sentence2):
+        s1split = sentence1.split()
+        s2split = sentence2.split()
+
+        jaccard_score = get_jaccard_similarity(set(s1split), set(s2split))
+        length_score = get_length_score(s1split, s2split)
+
+        return (jaccard_score * 0.7 + length_score * 0.3) * 100.0
+
+
     def getLogMetric(self, pred : str, ref : str):
         ref = ref.strip(' \t\n\r')
         pred = pred.strip(' \t\n\r')
@@ -170,30 +194,6 @@ class LogMetric(evaluate.Metric):
                 # e.g. date format not parsable by dateutil.parser
                 matchesPatternScore = 0.0
                 monotonicallyIncreasingScore = 0.0
-            
-        # Jaccard Similarity to measure closeness of two log-messages
-        def get_jaccard_similarity(set1, set2):
-            intersection = set1.intersection(set2)
-            union = set1.union(set2)
-            return len(intersection) / len(union)
-        
-        # A score depending on the difference in length of two sentences
-        def get_length_score(sentence1, sentence2):
-            s1len = len(sentence1)
-            s2len = len(sentence2)
-
-            return 1 - (abs(s1len - s2len) / max(s1len, s2len))
-
-        # Combine a weighted average of different scores
-        def get_overall_similarity(sentence1, sentence2):
-            s1split = sentence1.split()
-            s2split = sentence2.split()
-
-            jaccard_score = get_jaccard_similarity(set(s1split), set(s2split))
-            length_score = get_length_score(s1split, s2split)
-
-            return (jaccard_score * 0.7 + length_score * 0.3) * 100.0
-
     
         # apply jaccard-similarity to every pred-ref pair and then take mean score * 100
         local_score = np.mean([get_overall_similarity(p, r) for p,r in