Merge pull request #1120 from gruebel/fix-caching

change MAXREPEAT to int

Author: erezsh

lark/utils.py

@@ -137,9 +137,10 @@ def get_regexp_width(expr):
             # we manually test for the most important info (whether the empty string is matched)
             c = regex.compile(regexp_final)
             if c.match('') is None:
-                return 1, sre_constants.MAXREPEAT
+                # MAXREPEAT is a none pickable subclass of int, therefore needs to be converted to enable caching
+                return 1, int(sre_constants.MAXREPEAT)
             else:
-                return 0, sre_constants.MAXREPEAT
+                return 0, int(sre_constants.MAXREPEAT)
 
 ###}
 

tests/test_cache.py

@@ -1,5 +1,6 @@
 from __future__ import absolute_import
 
+import logging
 from unittest import TestCase, main
 
 from lark import Lark, Tree, Transformer
@@ -140,6 +141,24 @@ def test_imports(self):
         res = parser.parse("ab")
         self.assertEqual(res, Tree('startab', [Tree('expr', ['a', 'b'])]))
 
+    def test_recursive_pattern(self):
+        g = """
+        start: recursive+
+        recursive: /\w{3}\d{3}(?R)?/
+        """
+
+        assert len(self.mock_fs.files) == 0
+        Lark(g, parser="lalr", regex=True, cache=True)
+        assert len(self.mock_fs.files) == 1
+
+        with self.assertLogs("lark", level="ERROR") as cm:
+            Lark(g, parser='lalr', regex=True, cache=True)
+            assert len(self.mock_fs.files) == 1
+            # need to add an error log, because 'self.assertNoLogs' was added in Python 3.10
+            logging.getLogger('lark').error("dummy message")
+        # should only have the dummy log
+        self.assertCountEqual(cm.output, ["ERROR:lark:dummy message"])
+