fix potentially incorrect results of JaroWinkler when using high prefix weights

Author: maxbachmann

CHANGELOG.md

@@ -1,5 +1,9 @@
 ## Changelog
 
+## [3.0.3] - 2023-04-06
+### Fixed
+- fix potentially incorrect results of JaroWinkler when using high prefix weights
+
 ## [3.0.2] - 2023-03-04
 ### Fixed
 - fix assert leading to compilation failures

rapidfuzz/distance/JaroWinkler.hpp

@@ -137,6 +137,7 @@ struct MultiJaroWinkler : public detail::MultiSimilarityBase<MultiJaroWinkler<Ma
                     if (static_cast<uint64_t>(s2[prefix]) != prefixes[i][prefix]) break;
 
                 scores[i] += static_cast<double>(prefix) * prefix_weight * (1.0 - scores[i]);
+                scores[i] = std::min(scores[i], 1.0);
             }
 
             if (scores[i] < score_cutoff) scores[i] = 0.0;

rapidfuzz/distance/JaroWinkler_impl.hpp

@@ -29,7 +29,10 @@ double jaro_winkler_similarity(const Range<InputIt1>& P, const Range<InputIt2>&
     }
 
     double Sim = jaro_similarity(P, T, jaro_score_cutoff);
-    if (Sim > 0.7) Sim += static_cast<double>(prefix) * prefix_weight * (1.0 - Sim);
+    if (Sim > 0.7) {
+        Sim += static_cast<double>(prefix) * prefix_weight * (1.0 - Sim);
+        Sim = std::min(Sim, 1.0);
+    }
 
     return (Sim >= score_cutoff) ? Sim : 0;
 }
@@ -58,7 +61,10 @@ double jaro_winkler_similarity(const BlockPatternMatchVector& PM, const Range<In
     }
 
     double Sim = jaro_similarity(PM, P, T, jaro_score_cutoff);
-    if (Sim > 0.7) Sim += static_cast<double>(prefix) * prefix_weight * (1.0 - Sim);
+    if (Sim > 0.7) {
+        Sim += static_cast<double>(prefix) * prefix_weight * (1.0 - Sim);
+        Sim = std::min(Sim, 1.0);
+    }
 
     return (Sim >= score_cutoff) ? Sim : 0;
 }