Merge 89d46e8 into 6cec2cc

splink/jar_fallback.py

@@ -1,27 +1,6 @@
 import math
 
 
-def jc_sim_py(str1, str2):
-    """
-    Jaccard`similarity calculated exactly as in stringutils.similarity jaccard in Apache Commons
-    """
-
-    if not str1 or not str2:
-        return 0.0
-
-    k = 2  # default k in stringutil is 2 so leaving it like that for compatibility
-
-    # break strings into sets of rolling k-char syllables
-    a = set([str1[i : i + 1] for i in range(len(str1) - k + 1)])
-    b = set([str2[i : i + 1] for i in range(len(str2) - k + 1)])
-
-    # calculate instersection of two sets
-    c = a.intersection(b)
-
-    # return Jaccard similarity
-    return float(len(c)) / (len(a) + len(b) - len(c))
-
-
 def jw_sim_py(
     first,
     second,  # modification from original to not use other imput parameters

tests/test_jar_fallback.py

@@ -1,5 +1,5 @@
 import pytest
-from splink.jar_fallback import jw_sim_py, jc_sim_py
+from splink.jar_fallback import jw_sim_py
 
 
 def test_fallback_jw_nodata():
@@ -8,13 +8,12 @@ def test_fallback_jw_nodata():
     assert jw_sim_py(None, "Something") == 0.0
 
 
-def test_fallback_jc_nodata():
-    assert jc_sim_py(None, None) == 0.0
-    assert jc_sim_py("something", None) == 0.0
-    assert jc_sim_py(None, "Something") == 0.0
-
-
 def test_fallback_jw_wikipedia_examples():
+    """
+    tests from Apache Commons jarowinkler similarity available at:
+    https://github.com/apache/commons-text/blob/master/src/test/java/org/apache/commons/text/similarity/JaroWinklerSimilarityTest.java
+
+    """
     assert jw_sim_py("fly", "ant") == 0.0
     assert jw_sim_py("elephant", "hippo") == 0.44
     assert jw_sim_py("ABC Corporation", "ABC Corp") == 0.91