model._string_constraints: implement constraint AASd-130

basyx/aas/model/_string_constraints.py

@@ -26,6 +26,14 @@
 
 
 _T = TypeVar("_T")
+AASD130_RE = re.compile("[\x09\x0A\x0D\x20-\uD7FF\uE000-\uFFFD\U00010000-\U0010FFFF]*")
+
+
+def _unicode_escape(value: str) -> str:
+    """
+    Escapes unicode characters such as \uD7FF, that may be used in regular expressions, for better error messages.
+    """
+    return value.encode("unicode_escape").decode("utf-8")
 
 
 # Functions to verify the constraints for a given value.
@@ -36,7 +44,16 @@ def check(value: str, type_name: str, min_length: int = 0, max_length: Optional[
     if max_length is not None and len(value) > max_length:
         raise ValueError(f"{type_name} has a maximum length of {max_length}! (length: {len(value)})")
     if pattern is not None and not pattern.fullmatch(value):
-        raise ValueError(f"{type_name} must match the pattern '{pattern.pattern}'! (value: '{value}')")
+        raise ValueError(f"{type_name} must match the pattern '{_unicode_escape(pattern.pattern)}'! "
+                         f"(value: '{_unicode_escape(value)}')")
+    # Constraint AASd-130: an attribute with data type "string" shall consist of these characters only:
+    if not AASD130_RE.fullmatch(value):
+        # It's easier to implement this as a ValueError, because otherwise AASConstraintViolation would need to be
+        # imported from `base` and the ConstrainedLangStringSet would need to except AASConstraintViolation errors
+        # as well, while only re-raising ValueErrors. Thus, even if an AASConstraintViolation would be raised here,
+        # in case of a ConstrainedLangStringSet it would be re-raised as a ValueError anyway.
+        raise ValueError(f"Every string must match the pattern '{_unicode_escape(AASD130_RE.pattern)}'! "
+                         f"(value: '{_unicode_escape(value)}')")
 
 
 def check_content_type(value: str, type_name: str = "ContentType") -> None:

test/model/test_string_constraints.py

@@ -41,6 +41,25 @@ def test_version_type(self) -> None:
         version = "0"
         _string_constraints.check_version_type(version)
 
+    def test_aasd_130(self) -> None:
+        name: model.NameType = "\0"
+        with self.assertRaises(ValueError) as cm:
+            _string_constraints.check_name_type(name)
+        self.assertEqual(r"Every string must match the pattern '[\t\n\r -\ud7ff\ue000-\ufffd\U00010000-\U0010ffff]*'! "
+                         r"(value: '\x00')", cm.exception.args[0])
+        name = "\ud800"
+        with self.assertRaises(ValueError) as cm:
+            _string_constraints.check_name_type(name)
+        self.assertEqual(r"Every string must match the pattern '[\t\n\r -\ud7ff\ue000-\ufffd\U00010000-\U0010ffff]*'! "
+                         r"(value: '\ud800')", cm.exception.args[0])
+        name = "\ufffe"
+        with self.assertRaises(ValueError) as cm:
+            _string_constraints.check_name_type(name)
+        self.assertEqual(r"Every string must match the pattern '[\t\n\r -\ud7ff\ue000-\ufffd\U00010000-\U0010ffff]*'! "
+                         r"(value: '\ufffe')", cm.exception.args[0])
+        name = "this\ris\na\tvalid täst\uffdd\U0010ab12"
+        _string_constraints.check_name_type(name)
+
 
 class StringConstraintsDecoratorTest(unittest.TestCase):
     @_string_constraints.constrain_path_type("some_attr")