Merge pull request #1450 from mabel-dev/#1447

#1447

opteryx/__version__.py

@@ -1,4 +1,4 @@
-__build__ = 302
+__build__ = 304
 
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

opteryx/compiled/levenshtein/clevenshtein.pyx

@@ -0,0 +1,47 @@
+# cython: language_level=3
+
+import numpy as np  # Required for array allocation
+
+cdef int min3(int x, int y, int z):
+    """Utility function to find the minimum of three integers."""
+    cdef int m = x
+    if y < m:
+        m = y
+    if z < m:
+        m = z
+    return m
+
+def levenshtein(str string1, str string2):
+    """
+    Calculate the Levenshtein distance between two strings.
+
+    Parameters:
+        string1 (str): The first string to compare.
+        string2 (str): The second string to compare.
+
+    Returns:
+        int: The Levenshtein distance between string1 and string2.
+    """
+    cdef int len1 = len(string1)
+    cdef int len2 = len(string2)
+    cdef int i, j
+
+    # Allocate a numpy array and create a memory view from it
+    cdef int[:] dp = np.zeros((len1 + 1) * (len2 + 1), dtype=np.int32)
+
+    for i in range(len1 + 1):
+        for j in range(len2 + 1):
+            if i == 0:
+                dp[i * (len2 + 1) + j] = j  # First string is empty
+            elif j == 0:
+                dp[i * (len2 + 1) + j] = i  # Second string is empty
+            elif string1[i - 1] == string2[j - 1]:
+                dp[i * (len2 + 1) + j] = dp[(i - 1) * (len2 + 1) + (j - 1)]
+            else:
+                dp[i * (len2 + 1) + j] = 1 + min3(
+                    dp[(i - 1) * (len2 + 1) + j],      # Remove
+                    dp[i * (len2 + 1) + (j - 1)],      # Insert
+                    dp[(i - 1) * (len2 + 1) + (j - 1)] # Replace
+                )
+
+    return dp[len1 * (len2 + 1) + len2]

opteryx/functions/string_functions.py

@@ -300,10 +300,11 @@ def rtrim(*args):
 
 
 def levenshtein(a, b):
-    from opteryx.third_party.levenshtein import levenshtein as lev
+    from opteryx.compiled.levenshtein import levenshtein as lev
 
-    def _outer():
-        for index, value in enumerate(a):
-            yield lev(value, b[index])
+    # Convert numpy arrays to lists
+    a_list = a.tolist()
+    b_list = b.tolist()
 
-    return list(_outer())
+    # Use zip to iterate over pairs of elements from a and b
+    return [lev(value_a, value_b) for value_a, value_b in zip(a_list, b_list)]

setup.py

@@ -63,7 +63,7 @@ def rust_build(setup_kwargs: Dict[str, Any]) -> None:
     ),
     Extension(
         name="clevenshtein",
-        sources=["opteryx/third_party/levenshtein/clevenshtein.pyx"],
+        sources=["opteryx/compiled/levenshtein/clevenshtein.pyx"],
         extra_compile_args=COMPILE_FLAGS,
     ),
     Extension(

tests/misc/test_levenshtien.py

@@ -5,7 +5,7 @@
 
 import pytest
 
-from opteryx.third_party.levenshtein import levenshtein
+from opteryx.compiled.levenshtein import levenshtein
 
 # fmt:off
 TESTS = [