nfa greediness

regexy/compile/nfa.py

@@ -149,9 +149,19 @@ def nfa(nodes: Iterator[Node]) -> Node:
 
         if node.char == Symbols.ZERO_OR_ONE:
             state = states.pop()
-            node.out = [state, EOF]
-            states.append(node)
-            continue
+
+            # todo: refactor the whole thing. Move repetition range to parse.py
+            # todo: add non-greedy symbols
+            if state.char in (
+                    Symbols.ZERO_OR_MORE,
+                    Symbols.ZERO_OR_ONE):
+                state.out = list(reversed(state.out))
+                states.append(state)
+                continue
+            else:
+                node.out = [state, EOF]
+                states.append(node)
+                continue
 
         if node.char == Symbols.GROUP_START:
             state = states.pop()

tests/tests.py

@@ -337,3 +337,23 @@ def test_non_capturing_groups(self):
 
         # self.assertIsNone(match(r'((a)*n?(asd))*', 'aaanasdnasd'))  # fixme
         # should be equal to r'((a)*n?(asd)*)*' (see last capture)
+
+    def test_greediness(self):
+        self.assertEqual(
+            match(r'(a)*(a)*(a)*', 'aaa'),
+            (('a', 'a', 'a'), None, None))
+        self.assertEqual(
+            match(r'(a)*?(a)*(a)*?', 'aaa'),
+            (None, ('a', 'a', 'a'), None))
+        self.assertEqual(
+            match(r'(a)*?(a)*?(a)*', 'aaa'),
+            (None, None, ('a', 'a', 'a')))
+        self.assertEqual(
+            match(r'(a)*?(a)*?(a)*?', 'aaa'),
+            (None, None, ('a', 'a', 'a')))
+
+        self.assertEqual(match(r'(a)?(aa?)', 'aa'), ('a', 'a'))
+        self.assertEqual(match(r'(a)??(a)', 'aa'), ('a', 'a'))
+        self.assertEqual(match(r'(a)??(aa?)', 'aa'), (None, 'aa'))
+
+        # print(to_nfa_str('(a)*?(a)*(a)*'))