vatin: reject a duplicated country code prefix (#420)

stdnum/vatin.py

@@ -87,11 +87,29 @@ def validate(number: str) -> str:
     This performs the country-specific check for the number.
     """
     number = clean(number, '').strip()
-    module = _get_cc_module(number[:2])
+    cc = number[:2].upper()
+    module = _get_cc_module(cc)
     try:
-        return number[:2].upper() + module.validate(number[2:])
+        # Most country modules accept and strip the optional country code
+        # prefix themselves, so the number can be validated as it is.
+        result = module.validate(number)
+        if not result.upper().startswith(cc):
+            result = cc + result
     except ValidationError:
-        return module.validate(number)
+        # Other country modules only accept the national number, so retry
+        # with the country code prefix removed.
+        national = number[2:]
+        result = module.validate(national)
+        # If the country module stripped a second country code off the
+        # national number then the prefix was duplicated and the number is
+        # not valid (see #420). A national number that legitimately starts
+        # with the country code, such as the Mexican RFC MXDE111111GR2,
+        # keeps its prefix when compacted and is accepted.
+        if (re.sub(r'[^0-9A-Za-z]', '', national)[:2].upper() == cc and
+                not module.compact(national).upper().startswith(cc)):
+            raise InvalidFormat()
+        result = cc + result
+    return result
 
 
 def is_valid(number: str) -> bool:

tests/test_vatin.doctest

@@ -104,7 +104,26 @@ True
 'EU191849184'
 
 
-Check seemingly double country codes are handled correctly:
+A duplicated country code prefix should not be accepted (#420). This used to
+pass because the country code was stripped twice, once here and once by the
+country module:
+
+>>> vatin.is_valid('BE 0308.357.159')
+True
+>>> vatin.is_valid('BE BE 0308.357.159')
+False
+>>> vatin.is_valid('BEBE0308357159')
+False
+>>> vatin.validate('BE BE 0308.357.159')
+Traceback (most recent call last):
+    ...
+InvalidFormat: ...
+
+
+Check seemingly double country codes are handled correctly. A national number
+may legitimately start with the country code, in which case it is kept:
 
 >>> vatin.validate('MX MXDE 111111 GR2')  # the second "MX" is part of the number
 'MXMXDE111111GR2'
+>>> vatin.validate('MXDE 111111 GR2')  # here "MX" is only the country code
+'MXDE111111GR2'