Optimize backreference

Bump version to 0.9.4

Author: Buba98

regex_enumerator/regex_tree.py

@@ -99,19 +99,14 @@ def __init__(self, reference: RegexTree, min_len: int, max_len: int | None, prec
         self.current: dict[str, list[str]
                            ] = self._calculate() if not self.done else {}
 
-    def update(self):
+    def update_reference(self, new_strings: set[str]) -> None:
         if self._max_len is not None and self._min_len + self._index >= self._max_len and self.reference.done:
             self.done = True
 
-        for string in self.reference.current:
-            if string in self.current:
-                self.current[string].append(
-                    string * (self._min_len + self._index))
-            else:
-                result = []
-                for i in range(self._min_len, self._min_len + self._index + 1):
-                    result.append(string * i)
-                self.current[string] = result
+        for string in new_strings:
+            assert string not in self.current
+            self.current[string] = [
+                string * i for i in range(self._min_len, self._min_len + self._index + 1)]
 
     def _calculate(self) -> dict[str, set[str]]:
         current_ref = self.reference.current
@@ -121,10 +116,8 @@ def _calculate(self) -> dict[str, set[str]]:
         result: dict[str, list[str]] = {}
 
         for string in current_ref:
-            partial = []
-            for i in range(self._min_len, self._min_len + self._index + 1):
-                partial.append(string * i)
-            result[string] = partial
+            result[string] = [
+                string * i for i in range(self._min_len, self._min_len + self._index + 1)]
 
         return result
 
@@ -324,8 +317,8 @@ def next(self) -> set[str]:
             return result
 
         for reference in self.references:
-            reference.update()
-        return self.current
+            reference.update_reference(result)
+        return result
 
     def _calculate(self) -> set[str]:
         assert self._index_repetition != 0

setup.py

@@ -5,7 +5,7 @@
 
 setup(
     name='regex_enumerator',
-    version='0.9.3',
+    version='0.9.4',
     packages=find_packages(include=['regex_enumerator', 'regex_enumerator.*']),
     description='Enumerate all strings that match a given regex',
     author='Vincenzo Greco',

tests/test_backreference.py

@@ -80,3 +80,19 @@ def test_named_group_infinite_repetition_with_backreference():
     ]
 
     f_infinite(regex, possibilities)
+
+
+def test_backreference_with_group_quantifier_and_mismatch():
+    regex = r'(a){1,3}\1{0,2}'
+    possibilities = ['a', 'aa', 'aaa', 'aaaa', 'aaaaaa', 'aaaaaaaaa']
+
+    f_finite(regex, possibilities)
+
+
+def test_backreference_with_group_quantifier_and_mismatch_complex():
+    regex = r'(a){1,3}[c-d]\1{0,2}'
+    possibilities = ['ac', 'ad', 'aca', 'ada', 'acaa', 'adaa',
+                     'aac', 'aad', 'aacaa', 'aadaa', 'aacaaaa', 'aadaaaa',
+                     'aaac', 'aaad', 'aaacaaa', 'aaadaaa', 'aaacaaaaaa', 'aaadaaaaaa']
+
+    f_finite(regex, possibilities)