Change string semantic comparison algorithm

Use `fuzzywuzzy`'s Token Sort Ratio instead of Jaro-Winkler.

setup.py

@@ -64,6 +64,6 @@ setup(
     },
     extras_require={
         'taxii': ['taxii2-client'],
-        'semantic': ['haversine', 'pyjarowinkler'],
+        'semantic': ['haversine', 'fuzzywuzzy'],
     },
 )

stix2/environment.py

@@ -363,10 +363,10 @@ def partial_string_based(str1, str2):
         float: Number between 0.0 and 1.0 depending on match criteria.
 
     """
-    from pyjarowinkler import distance
+    from fuzzywuzzy import fuzz
-    result = distance.get_jaro_distance(str1, str2)
+    result = fuzz.token_sort_ratio(str1, str2, force_ascii=False)
     logger.debug("--\t\tpartial_string_based '%s' '%s'\tresult: '%s'", str1, str2, result)
-    return result
+    return result / 100.0
 
 
 def custom_pattern_based(pattern1, pattern2):

stix2/test/v21/test_environment.py

@@ -521,7 +521,7 @@ def test_semantic_equivalence_on_same_vulnerability2():
         ],
     )
     VULN_KWARGS2 = dict(
-        name="Zot",
+        name="Foo",
         external_references=[
             {
                 "url": "https://example2",
@@ -550,7 +550,7 @@ def test_semantic_equivalence_on_unknown_object():
     CUSTOM_KWARGS2 = dict(
         type="x-foobar",
         id="x-foobar--0c7b5b88-8ff7-4a4d-aa9d-feb398cd0061",
-        name="Zot",
+        name="Foo",
         external_references=[
             {
                 "url": "https://example2",
@@ -787,7 +787,7 @@ def test_semantic_equivalence_prop_scores():
     tool2 = stix2.v21.Tool(id=TOOL_ID, **TOOL2_KWARGS)
     stix2.Environment().semantically_equivalent(tool1, tool2, prop_scores)
     assert len(prop_scores) == 4
-    assert round(prop_scores["matching_score"], 1) == 37.6
+    assert round(prop_scores["matching_score"], 1) == 8.8
     assert round(prop_scores["sum_weights"], 1) == 100.0
 
 

tox.ini

@@ -9,7 +9,7 @@ deps =
   pytest-cov
   coverage
   taxii2-client
-  pyjarowinkler
+  fuzzywuzzy
   haversine
   medallion
 commands =