Validate Locale parameter formats while constructing

babel/core.py

@@ -96,17 +96,29 @@ class UnknownLocaleError(Exception):
     is available.
     """
 
-    def __init__(self, identifier):
+    def __init__(self, identifier, message=None):
         """Create the exception.
 
         :param identifier: the identifier string of the unsupported locale
+        :param message: an optional message string to override the default
         """
-        Exception.__init__(self, 'unknown locale %r' % identifier)
+        if not message:
+            message = 'unknown locale %r' % identifier
+        Exception.__init__(self, message)
 
         #: The identifier of the locale that could not be found.
         self.identifier = identifier
 
 
+class InvalidLocaleSpecificationError(ValueError):
+    def __init__(self, identifier, part_name, value):
+        ValueError.__init__(
+            self,
+            "The %s (%r) of the locale identifier %r is invalid." % (part_name, value, identifier)
+        )
+        self.identifier = identifier
+
+
 class Locale(object):
     """Representation of a specific locale.
 
@@ -156,16 +168,34 @@ def __init__(self, language, territory=None, script=None, variant=None):
                                      requested locale
         """
         #: the language code
-        self.language = language
+        self.language = str(language)
         #: the territory (country or region) code
-        self.territory = territory
+        self.territory = (str(territory) if territory else None)
         #: the script code
-        self.script = script
+        self.script = (str(script) if script else None)
         #: the variant code
-        self.variant = variant
+        self.variant = (str(variant) if variant else None)
         self.__data = None
 
+        self._validate()
+
+    def _validate(self):
+        """
+        Validate that the locale parameters seem sane, and that the locale is known.
+        """
         identifier = str(self)
+        if not self.language.isalpha():
+            raise InvalidLocaleSpecificationError(identifier, "language", self.language)
+
+        if self.territory and not (self.territory.isdigit() or (self.territory.isalpha() and self.territory.isupper())):
+            raise InvalidLocaleSpecificationError(identifier, "territory", self.territory)
+
+        if self.script and not self.script.isalpha():
+            raise InvalidLocaleSpecificationError(identifier, "script", self.script)
+
+        if self.variant and not self.variant.isalpha():
+            raise InvalidLocaleSpecificationError(identifier, "variant", self.variant)
+
         if not localedata.exists(identifier):
             raise UnknownLocaleError(identifier)
 
@@ -808,7 +838,7 @@ def interval_formats(self):
         How to format date intervals in Finnish when the day is the
         smallest changing component:
 
-        >>> Locale('fi_FI').interval_formats['MEd']['d']
+        >>> Locale('fi', 'FI').interval_formats['MEd']['d']
         [u'E d. \u2013 ', u'E d.M.']
 
         .. seealso::
@@ -846,7 +876,7 @@ def list_patterns(self):
         u'{0}, {1}'
         >>> Locale('en').list_patterns['end']
         u'{0}, and {1}'
-        >>> Locale('en_GB').list_patterns['end']
+        >>> Locale('en', 'GB').list_patterns['end']
         u'{0} and {1}'
         """
         return self._data['list_patterns']

tests/conftest.py

@@ -7,3 +7,9 @@ def os_environ(monkeypatch):
     mock_environ = dict(os.environ)
     monkeypatch.setattr(os, 'environ', mock_environ)
     return mock_environ
+
+
+def pytest_generate_tests(metafunc):
+    if hasattr(metafunc.function, "all_locales"):
+        from babel.localedata import locale_identifiers
+        metafunc.parametrize("locale", list(locale_identifiers()))

tests/messages/test_checkers.py

@@ -95,7 +95,7 @@ def test_2_num_plurals_checkers(self):
                 num_plurals  = PLURALS[_locale][0]
                 plural_expr  = PLURALS[_locale][1]
             try:
-                locale = Locale(_locale)
+                locale = Locale.parse(_locale)
                 date = format_datetime(datetime.now(LOCALTZ),
                                        'yyyy-MM-dd HH:mmZ',
                                        tzinfo=LOCALTZ, locale=_locale)

tests/test_core.py

@@ -16,7 +16,7 @@
 import pytest
 
 from babel import core, Locale
-from babel.core import default_locale, Locale
+from babel.core import default_locale, Locale, InvalidLocaleSpecificationError
 
 
 def test_locale_provides_access_to_cldr_locale_data():
@@ -35,14 +35,11 @@ def test_locale_comparison():
     en_US = Locale('en', 'US')
     en_US_2 = Locale('en', 'US')
     fi_FI = Locale('fi', 'FI')
-    bad_en_US = Locale('en_US')
     assert en_US == en_US
     assert en_US == en_US_2
     assert en_US != fi_FI
     assert not (en_US != en_US_2)
     assert None != en_US
-    assert en_US != bad_en_US
-    assert fi_FI != bad_en_US
 
 
 def test_can_return_default_locale(os_environ):
@@ -315,3 +312,25 @@ def find_class(self, module, name):
 
     with open(filename, 'rb') as f:
         return Unpickler(f).load()
+
+
+@pytest.mark.parametrize("parts", [
+    ("en", "US", "Hans", "BAHAMAS"),
+    ("yi", "001"),
+])
+def test_locale_ctor_validation(parts):
+    part_names = ("language", "territory", "script", "variant")
+    n = len(parts)
+    for i in range(n):
+        mangled_parts = list(parts)
+        for x in range(i, n):
+            mangled_parts[x] = "~~"
+        assert len(mangled_parts) == n
+        with pytest.raises(InvalidLocaleSpecificationError) as ei:
+            Locale(*mangled_parts)
+        assert part_names[i] in str(ei.value)  # assert the complaint was about the first mangled part
+
+
+@pytest.mark.all_locales
+def test_locale_load(locale):
+    assert str(Locale.parse(locale)) == locale