fix incorrect index calculation in hashmap

Author: maxbachmann

CHANGELOG.rst

@@ -1,14 +1,21 @@
 Changelog
 ---------
 
+[0.3.1] - 2023-11-29
+^^^^^^^^^^^^^^^^^^^^
+Fixed
+~~~~~
+* fixed crash inside hashmap lookup function leading to a crash in the
+  Damerau-Levenshtein implementation
+
 [0.3.0] - 2023-11-27
-^^^^^^^^^^^^^^^^^^^^^
+^^^^^^^^^^^^^^^^^^^^
 Previous versions only existed for testing purposed years ago. This is a complete
 rewrite porting a subset of the features provided in the C++ implementation of
 rapidfuzz. The remaining features will be added in later releases.
 
 Added
-~~~~~~~
+~~~~~
 * added implementations of the following string metrics:
 
   * Levenshtein distance

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 description="rapid fuzzy string matching library"
 name = "rapidfuzz"
-version = "0.3.0"
+version = "0.3.1"
 authors = ["maxbachmann <[email protected]>"]
 edition = "2021"
 readme = "Readme.md"

src/details/growing_hashmap.rs

@@ -100,7 +100,7 @@ where
     /// strategy to `CPython` and `Ruby`
     fn lookup(&self, key: u64) -> usize {
         let hash = key;
-        let mut i = (hash & self.mask as u64) as usize;
+        let mut i = hash as usize & self.mask as usize;
 
         let map = self
             .map
@@ -113,7 +113,7 @@ where
 
         let mut perturb = key;
         loop {
-            i = (i * 5 + perturb as usize + 1) % 128;
+            i = (i * 5 + perturb as usize + 1) & self.mask as usize;
 
             if map[i].value == Default::default() || map[i].key == key {
                 return i;

src/distance/damerau_levenshtein.rs

@@ -628,4 +628,12 @@ mod tests {
             0.0001
         );
     }
+
+    #[test]
+    fn unicode() {
+        assert_eq!(
+            Some(5),
+            _test_distance("Иванко".chars(), "Петрунко".chars(), None, None)
+        );
+    }
 }

src/distance/indel.rs

@@ -790,4 +790,12 @@ mod tests {
 
         assert_eq!(Some(231), test_distance_ascii(s1, s2, None, None));
     }
+
+    #[test]
+    fn unicode() {
+        assert_eq!(
+            Some(8),
+            test_distance("Иванко".chars(), "Петрунко".chars(), None, None)
+        );
+    }
 }

src/distance/jaro.rs

@@ -1111,4 +1111,13 @@ mod tests {
             }
         }
     }
+
+    #[test]
+    fn unicode() {
+        assert_delta!(
+            Some(0.375),
+            _test_distance("Иванко".chars(), "Петрунко".chars(), None, None),
+            0.0001
+        );
+    }
 }

src/distance/jaro_winkler.rs

@@ -715,4 +715,13 @@ mod tests {
             }
         }
     }
+
+    #[test]
+    fn unicode() {
+        assert_delta!(
+            Some(0.375),
+            _test_distance("Иванко".chars(), "Петрунко".chars(), None, None),
+            0.0001
+        );
+    }
 }

src/distance/lcs_seq.rs

@@ -1208,4 +1208,12 @@ mod tests {
         let b = "220";
         assert_eq!(Some(1), test_similarity_ascii(a, b, None, None));
     }
+
+    #[test]
+    fn unicode() {
+        assert_eq!(
+            Some(5),
+            test_distance("Иванко".chars(), "Петрунко".chars(), None, None)
+        );
+    }
 }

src/distance/levenshtein.rs

@@ -2115,4 +2115,12 @@ mod tests {
             distance(OCR_EXAMPLE1.iter(), OCR_EXAMPLE2.iter(), None, None, 0)
         );
     }
+
+    #[test]
+    fn unicode() {
+        assert_eq!(
+            Some(5),
+            _test_distance("Иванко".chars(), "Петрунко".chars(), None, None, None)
+        );
+    }
 }

src/distance/osa.rs

@@ -606,4 +606,12 @@ mod tests {
         let s2 = "b".to_string() + &filler + "AC" + &filler + "b";
         assert_eq!(Some(3), _test_distance_ascii(&s1, &s2, None, None));
     }
+
+    #[test]
+    fn unicode() {
+        assert_eq!(
+            Some(5),
+            _test_distance("Иванко".chars(), "Петрунко".chars(), None, None)
+        );
+    }
 }