From bcbd7dbb5a3a5db7856ebc849c08c06ad09bfee9 Mon Sep 17 00:00:00 2001
From: Alex | Kronox <alexander_vogt@outlook.de>
Date: Thu, 28 Nov 2024 08:12:54 +0100
Subject: [PATCH] include basecode in symmetric similarity

---
 core/src/main/java/de/jplag/JPlagComparison.java    | 13 +++++++++++++
 .../java/de/jplag/options/SimilarityMetric.java     |  9 +--------
 2 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/core/src/main/java/de/jplag/JPlagComparison.java b/core/src/main/java/de/jplag/JPlagComparison.java
index 37aa0c7ad..2fa03cef0 100644
--- a/core/src/main/java/de/jplag/JPlagComparison.java
+++ b/core/src/main/java/de/jplag/JPlagComparison.java
@@ -54,6 +54,19 @@ public final double similarity() {
         return 2 * similarity(divisorA + divisorB);
     }
 
+    /**
+     * @return A symmetric similarity in interval [0, 1]. O means no similarity, 1 means maximum similarity.
+     */
+    public final double symmetricSimilarity() {
+        boolean subtractBaseCode = firstSubmission.hasBaseCodeMatches() && secondSubmission.hasBaseCodeMatches();
+        int divisorA = firstSubmission.getSimilarityDivisor(subtractBaseCode);
+        int divisorB = secondSubmission.getSimilarityDivisor(subtractBaseCode);
+        if (divisorA + divisorB == 0) {
+            return 0.0;
+        }
+        return 2.0 * getNumberOfMatchedTokens() / (divisorA + divisorB);
+    }
+
     /**
      * @return Similarity of the first submission in interval [0, 1]. O means no similarity, 1 means maximum similarity.
      */
diff --git a/core/src/main/java/de/jplag/options/SimilarityMetric.java b/core/src/main/java/de/jplag/options/SimilarityMetric.java
index 08d00f78d..4cefb9ff7 100644
--- a/core/src/main/java/de/jplag/options/SimilarityMetric.java
+++ b/core/src/main/java/de/jplag/options/SimilarityMetric.java
@@ -10,14 +10,7 @@ public enum SimilarityMetric implements ToDoubleFunction<JPlagComparison> {
     MIN("minimum similarity", JPlagComparison::minimalSimilarity),
     MAX("maximal similarity", JPlagComparison::maximalSimilarity),
     INTERSECTION("matched tokens", it -> (double) it.getNumberOfMatchedTokens()),
-    SYMMETRIC("symmetric similarity", it -> {
-        int divisor = it.firstSubmission().getNumberOfTokens() + it.secondSubmission().getNumberOfTokens();
-        if (divisor != 0) {
-            return 2.0 * it.getNumberOfMatchedTokens() / divisor;
-        } else {
-            return .0;
-        }
-    }),
+    SYMMETRIC("symmetric similarity", JPlagComparison::symmetricSimilarity),
     LONGEST_MATCH("number of tokens in the longest match", it -> it.matches().stream().mapToInt(Match::length).max().orElse(0)),
     OVERALL("Sum of both submission lengths", it -> it.firstSubmission().getNumberOfTokens() + it.secondSubmission().getNumberOfTokens());