Merge pull request #5 from qedsoftware/match

Optionally return match operations in get_all_paths

Author: DrKomodo

editdistance/_edit_distance_osa.cpp

@@ -91,19 +91,30 @@ std::vector<std::vector<CppEditop>> backtrack_all_paths(
         all_paths.insert(all_paths.end(), paths.begin(), paths.end());
         current_path.pop_back();
     }
-    
-    if (i > 0 && j > 0) {
-        double sub_cost = (a[i-1] == b[j-1]) ? 0.0 : replace_weight;
-        if (std::abs((dp[i-1][j-1] + sub_cost) - current_cost) < tol) {
-            std::string out_char = (sub_cost == 0.0) ? std::string(1, a[i-1]) : std::string(1, b[j-1]);
-            CppEditop op(REPLACE, i-1, j-1, sub_cost, out_char);
+
+    if (i > 0 && j > 0 && a[i-1] != b[j-1]) {
+        if (std::abs((dp[i-1][j-1] + replace_weight) - current_cost) < tol) {
+            std::string out_char = std::string(1, b[j-1]);
+            CppEditop op(REPLACE, i-1, j-1, replace_weight, out_char);
             current_path.push_back(op);
             auto paths = backtrack_all_paths(a, b, dp, i-1, j-1, current_path, replace_weight, insert_weight, delete_weight, swap_weight);
             all_paths.insert(all_paths.end(), paths.begin(), paths.end());
             current_path.pop_back();
         }
     }
-    
+
+    if (i > 0 && j > 0 && a[i-1] == b[j-1]) {
+        double match_weight = 0.0; // We might want to make this non-zero in the future
+        if (std::abs((dp[i-1][j-1] + match_weight) - current_cost) < tol) {
+            std::string out_char = std::string(1, a[i-1]);
+            CppEditop op(MATCH, i-1, j-1, match_weight, out_char);
+            current_path.push_back(op);
+            auto paths = backtrack_all_paths(a, b, dp, i-1, j-1, current_path, replace_weight, insert_weight, delete_weight, swap_weight);
+            all_paths.insert(all_paths.end(), paths.begin(), paths.end());
+            current_path.pop_back();
+        }
+    }
+
     if (i > 1 && j > 1 &&
         a[i-1] == b[j-2] && a[i-2] == b[j-1] &&
         std::abs((dp[i-2][j-2] + swap_weight) - current_cost) < tol) {
@@ -161,6 +172,7 @@ std::string editop_name_to_string(CppEditopName name) {
         case DELETE: return "DELETE";
         case REPLACE: return "REPLACE";
         case SWAP: return "SWAP";
+        case MATCH: return "MATCH";
         default: return "UNKNOWN";
     }
 }

editdistance/_edit_distance_osa.hpp

@@ -11,7 +11,8 @@ enum CppEditopName {
     INSERT,
     DELETE,
     REPLACE,
-    SWAP
+    SWAP,
+    MATCH
 };
 
 struct CppEditop {

editdistance/edit_distance_osa.pyx

@@ -1,6 +1,7 @@
 # distutils: language = c++
 # distutils: sources = ./editdistance/_edit_distance_osa.cpp
 
+from libcpp cimport bool
 from libcpp.map cimport map
 from libcpp.string cimport string
 from libcpp.vector cimport vector
@@ -14,6 +15,7 @@ cdef extern from "_edit_distance_osa.hpp":
         DELETE
         REPLACE
         SWAP
+        MATCH
 
     cdef struct CppEditop:
         CppEditopName name
@@ -38,6 +40,7 @@ class EditopName(Enum):
     DELETE = 1
     REPLACE = 2
     SWAP = 3
+    MATCH = 4
 
 
 cdef class Editop:
@@ -64,7 +67,8 @@ def get_all_paths(
     double replace_weight=1.0,
     double insert_weight=1.0,
     double delete_weight=1.0,
-    double swap_weight=1.0
+    double swap_weight=1.0,
+    bool return_matches=False,
 ):
     cdef string cpp_a = a.encode("utf-8")
     cdef string cpp_b = b.encode("utf-8")
@@ -76,8 +80,6 @@ def get_all_paths(
     for cpp_path in cpp_paths:
         python_path = []
         for cpp_op in cpp_path:
-            if cpp_op.cost == 0:
-                continue
             if cpp_op.name == INSERT:
                 py_name = EditopName.INSERT
             elif cpp_op.name == DELETE:
@@ -86,6 +88,11 @@ def get_all_paths(
                 py_name = EditopName.REPLACE
             elif cpp_op.name == SWAP:
                 py_name = EditopName.SWAP
+            elif cpp_op.name == MATCH:
+                if return_matches:
+                    py_name = EditopName.MATCH
+                else:
+                    continue
             else:
                 py_name = None
             python_path.append(Editop(
@@ -99,12 +106,11 @@ def get_all_paths(
     return python_paths
 
 def apply_editops(src, dst, editops):
+    # assumes editops are sorted from left to right
+    # assumes match operations are included
     src_idx = 0
     s = ""
     for op in editops:
-        while src_idx < op.src_idx:
-            s += src[src_idx]
-            src_idx += 1
         if op.name == EditopName.INSERT:
             s += dst[op.dst_idx]
         elif op.name == EditopName.DELETE:
@@ -116,9 +122,9 @@ def apply_editops(src, dst, editops):
             s += src[op.src_idx + 1]        
             s += src[op.src_idx]
             src_idx += 2
-    while src_idx < len(src):
-        s += src[src_idx]
-        src_idx += 1
+        elif op.name == EditopName.MATCH:
+            s += src[op.src_idx]
+            src_idx += 1
     return s
 
 

examples/osa_example.py

@@ -5,10 +5,7 @@
 """
 
 try:
-    from editdistance.osa import (
-        compute_distance,
-        get_all_paths,
-    )
+    from editdistance.osa import compute_distance, get_all_paths
 
     def main():
         # Test case from original Python code
@@ -25,12 +22,20 @@ def main():
             print(f"Distance: {distance}")
 
             paths = get_all_paths(source, target)
+            paths_with_matches = get_all_paths(source, target, return_matches=True)
             print(f"Number of optimal edit sequences: {len(paths)}")
 
+            print("Paths without match editops:")
             for i, path in enumerate(paths, 1):
-                print(f"Path {i}:")
+                print(f"    Path {i}:")
                 for op in path:
-                    print(f"  {op}")
+                    print(f"        {op}")
+            print()
+            print("Paths with match editops:")
+            for i, path in enumerate(paths_with_matches, 1):
+                print(f"    Path {i}:")
+                for op in path:
+                    print(f"        {op}")
             print()
 
     if __name__ == "__main__":

tests/__init__.py

@@ -1,10 +1,6 @@
 import unittest
 
-from editdistance.osa import (
-    apply_editops,
-    compute_distance,
-    get_all_paths,
-)
+from editdistance.osa import apply_editops, compute_distance, get_all_paths
 
 COMPUTE_DISTANCE_TEST_CASES = [
     ("single character", "a", "b", 1.0),
@@ -76,7 +72,7 @@ def test_get_all_paths(self):
     def test_editops_transform(self):
         for src, dst in EDITOPS_TRANSFORM_TEST_CASES:
             with self.subTest(src=src, dst=dst):
-                paths = get_all_paths(src, dst)
+                paths = get_all_paths(src, dst, return_matches=True)
                 self.assertTrue(paths, f"No paths found for {src} -> {dst}")
                 for path in paths:
                     result = apply_editops(src, dst, path)