Fix rendering of max. quantifier in regexes (#457)

There are many different syntaxes for regular expressions. Some of them
require the min. quantifier to be explicitly specified as 0 even if only
the max. quantifier is given. To comply with such syntaxes, we
explicitly render the max. quantifier without a min. quantifier as
`{0,m}`.

Author: mristin

aas_core_codegen/parse/retree/_render.py

@@ -178,7 +178,7 @@ def transform_quantifier(
                     if node.maximum == 1:
                         quantifier = "?"
                     else:
-                        quantifier = f"{{,{node.maximum}}}"
+                        quantifier = f"{{0,{node.maximum}}}"
                 else:
                     quantifier = f"{{{node.minimum},{node.maximum}}}"
         else:

aas_core_codegen/xsd/main.py

@@ -127,6 +127,12 @@ def _undo_escaping_backslash_x_u_and_U_in_pattern(pattern: str) -> str:
 class _AnchorRemover(parse_retree.BaseVisitor):
     """
     Remove anchors from a regex in-place.
+
+    We need to remove the anchors (``^``, ``$``) since patterns in the XSD are always
+    anchored.
+
+    This is necessary since otherwise the schema validation fails.
+    See: https://stackoverflow.com/questions/4367914/regular-expression-in-xml-schema-definition-fails
     """
 
     def visit_concatenation(self, node: parse_retree.Concatenation) -> None:
@@ -146,15 +152,10 @@ def visit_concatenation(self, node: parse_retree.Concatenation) -> None:
 
 
 @ensure(lambda result: (result[0] is not None) ^ (result[1] is not None))
-def _remove_anchors_in_pattern(pattern: str) -> Tuple[Optional[str], Optional[str]]:
-    """
-    We need to remove the anchors (``^``, ``$``) since schemas are always anchored.
-
-    This is necessary since otherwise the schema validation fails.
-    See: https://stackoverflow.com/questions/4367914/regular-expression-in-xml-schema-definition-fails
+def _translate_pattern(pattern: str) -> Tuple[Optional[str], Optional[str]]:
+    """Translate the pattern to obtain the equivalent in XSD."""
+    pattern = _undo_escaping_backslash_x_in_pattern(pattern)
 
-    Return pattern without anchors, or error message.
-    """
     parsed, error = parse_retree.parse(values=[pattern])
     if error is not None:
         regex_line, pointer_line = parse_retree.render_pointer(error.cursor)
@@ -176,18 +177,6 @@ def _remove_anchors_in_pattern(pattern: str) -> Tuple[Optional[str], Optional[st
     return "".join(parts), None
 
 
-@ensure(lambda result: (result[0] is not None) ^ (result[1] is not None))
-def _translate_pattern(pattern: str) -> Tuple[Optional[str], Optional[str]]:
-    """Translate the pattern to obtain the equivalent in XSD."""
-    result, error = _remove_anchors_in_pattern(
-        _undo_escaping_backslash_x_in_pattern(pattern)
-    )
-    if error is not None:
-        return None, error
-
-    return result, None
-
-
 def _generate_xs_restriction(
     base_type: intermediate.PrimitiveType,
     len_constraint: Optional[infer_for_schema.LenConstraint],

test_data/parse_retree/expected/quantifier/greedy/at_most_3/rendered_regex.txt

@@ -1 +1 @@
-a{,3}
+a{0,3}

test_data/parse_retree/expected/quantifier/non_greedy/at_most_3/rendered_regex.txt

@@ -1 +1 @@
-a{,3}?
+a{0,3}?

tests/xsd/test_main.py

@@ -69,18 +69,36 @@ def test_complex(self) -> None:
         )
 
 
-class Test_removing_anchors_in_patterns(unittest.TestCase):
+class Test_translate_pattern(unittest.TestCase):
     # NOTE (mristin, 2022-06-18):
     # This is relevant since XSD are always anchored.
     # See: https://stackoverflow.com/questions/4367914/regular-expression-in-xml-schema-definition-fails
 
-    def test_table(self) -> None:
+    def test_table_for_removing_anchors(self) -> None:
         for pattern, expected, identifier in [
             ("^$", "", "empty"),
             ("^something$", "something", "simple_literal"),
             ("(^.*$)", "(.*)", "within_a_group"),
         ]:
-            fixed, error = xsd_main._remove_anchors_in_pattern(pattern)
+            fixed, error = xsd_main._translate_pattern(pattern)
+            assert error is None, identifier
+            assert fixed is not None, identifier
+
+            self.assertEqual(expected, fixed, identifier)
+
+    def test_table_for_rendering_quantifiers(self) -> None:
+        # NOTE (mristin, 2024-03-22):
+        # We explicitly test for quantifiers to make sure that they all comply with
+        # XSD patterns. For example, when only the maximum quantifier is given,
+        # the minimum quantifier of 0 must be indicated explicitly.
+        for pattern, expected, identifier in [
+            ("a{1}", "a{1}", "exact repetition"),
+            ("a{1,}", "a+", "min 1 repetition"),
+            ("a{2,}", "a{2,}", "more than 1 min repetition"),
+            ("a{,2}", "a{0,2}", "only max repetition"),
+            ("a{1,2}", "a{1,2}", "min and max repetition"),
+        ]:
+            fixed, error = xsd_main._translate_pattern(pattern)
             assert error is None, identifier
             assert fixed is not None, identifier