improve performance and memory usage of editops

Author: maxbachmann

VERSION

@@ -1 +1 @@
-1.8.3
+1.9.0

docs/conf.py

@@ -22,7 +22,7 @@
 author = 'Max Bachmann'
 
 # The full version, including alpha/beta/rc tags
-release = '1.8.3'
+release = '1.9.0'
 
 
 # -- General configuration ---------------------------------------------------

extern/rapidfuzz-cpp

@@ -3,6 +3,6 @@
 """
 __author__ = "Max Bachmann"
 __license__ = "MIT"
-__version__ = "1.8.3"
+__version__ = "1.9.0"
 
 from rapidfuzz import process, fuzz, utils, levenshtein, string_metric

src/rapidfuzz/__init__.py

@@ -109,6 +109,32 @@ def extract_scorer(s1, s2, scorer, processor=None, **kwargs):
 def extract_iter_scorer(s1, s2, scorer, processor=None, **kwargs):
     return list(process.extract_iter(s1, [s2], processor=processor, scorer=scorer, **kwargs))[0][1]
 
+def apply_editops(s1, s2, ops):
+    new_str = ''
+    s1_pos = 0
+    for op in ops:
+        j = op[1] - s1_pos
+        while j:
+            new_str += s1[s1_pos]
+            s1_pos += 1
+            j -= 1
+
+        if op[0] == 'delete':
+            s1_pos += 1
+        elif op[0] == 'insert':
+            new_str += s2[op[2]]
+        elif op[0] == 'replace':
+            new_str += s2[op[2]]
+            s1_pos += 1
+
+    j = len(s1) - s1_pos
+    while j:
+        new_str += s1[s1_pos]
+        s1_pos += 1
+        j -= 1
+
+    return new_str
+
 
 HYPOTHESIS_ALPHABET = ascii_letters + digits + punctuation
 
@@ -137,12 +163,13 @@ def extract_iter_scorer(s1, s2, scorer, processor=None, **kwargs):
 ]
 
 @given(s1=st.text(), s2=st.text())
-@settings(max_examples=50, deadline=1000)
+@settings(max_examples=100, deadline=None)
 def test_levenshtein_editops(s1, s2):
     """
     test levenshtein_editops. Currently this only tests, so there are no exceptions.
     """
-    string_metric.levenshtein_editops(s1, s2)
+    ops = string_metric.levenshtein_editops(s1, s2)
+    assert apply_editops(s1, s2, ops) == s2
 
 @given(s1=st.text(max_size=64), s2=st.text())
 @settings(max_examples=50, deadline=1000)
@@ -322,4 +349,4 @@ def test_cdist(queries, choices):
 
     reference_matrix = cdist_distance(queries, queries, scorer=string_metric.levenshtein)
     matrix = process.cdist(queries, queries, scorer=string_metric.levenshtein)
-    assert (matrix == reference_matrix).all() 
+    assert (matrix == reference_matrix).all()

tests/test_hypothesis.py

@@ -143,7 +143,7 @@ def testWithScorer(self):
         self.assertEqual(best[0], choices_mapping[1])
         best = process.extract(query, choices_mapping, scorer=fuzz.QRatio)[0]
         self.assertEqual(best[0], choices_mapping[1])
-        
+
     def testWithCutoff(self):
         choices = [
             "new york mets vs chicago cubs",

tests/test_process.py

@@ -59,6 +59,10 @@ def test_levenshtein_editops():
     """
     assert string_metric.levenshtein_editops("0", "") == [("delete", 0, 0)]
     assert string_metric.levenshtein_editops("", "0") == [("insert", 0, 0)]
+
+    assert string_metric.levenshtein_editops("00", "0") == [("delete", 1, 1)]
+    assert string_metric.levenshtein_editops("0", "00") == [("insert", 1, 1)]
+
     assert string_metric.levenshtein_editops("qabxcd", "abycdf") == [
         ("delete", 0, 0), ("replace", 3, 2), ("insert", 6, 5)
     ]