Fix erroneous non-ASCII in local version (#469)

* Fix validation in the packaging.version.Version's constructor

* Fix validation in the packaging.specifiers.Specifier's constructor

* Complement relevant tests, including adding cases related to presence
  of non-ASCII whitespace characters in input strings

* Fix docs of packaging.version.VERSION_PATTERN by mentioning necessity
  of the re.ASCII flag

Author: zuo

CHANGELOG.rst

@@ -4,7 +4,10 @@ Changelog
 *unreleased*
 ~~~~~~~~~~~~
 
-No unreleased changes.
+* Fix parsing of ``Version`` and ``Specifier``, to prevent certain
+  non-ASCII letters from being accepted as a part of the local version
+  segment (:issue:`469`); also, fix the docs of ``VERSION_PATTERN``, to
+  mention necessity of the ``re.ASCII`` flag
 
 21.0 - 2021-07-03
 ~~~~~~~~~~~~~~~~~

docs/version.rst

@@ -284,7 +284,7 @@ Reference
     The pattern is not anchored at either end, and is intended for embedding
     in larger expressions (for example, matching a version number as part of
     a file name). The regular expression should be compiled with the
-    ``re.VERBOSE`` and ``re.IGNORECASE`` flags set.
+    ``re.VERBOSE``, ``re.IGNORECASE`` and ``re.ASCII`` flags set.
 
 
 .. _PEP 440: https://www.python.org/dev/peps/pep-0440/

packaging/specifiers.py

@@ -411,6 +411,23 @@ class Specifier(_IndividualSpecifier):
 
     _regex = re.compile(r"^\s*" + _regex_str + r"\s*$", re.VERBOSE | re.IGNORECASE)
 
+    # Note: an additional check, based of the following regular
+    # expression, is necessary because without it the 'a-z'
+    # character ranges in the above regular expression, in
+    # conjunction with re.IGNORECASE, would cause erroneous
+    # acceptance of non-ASCII letters in the local version segment
+    # (see: https://docs.python.org/library/re.html#re.IGNORECASE).
+    _supplementary_restriction_regex = re.compile(
+        r"""
+        \s*===.*         # No restriction in the identity operator case.
+        |
+        [\s\0-\177]*     # In all other cases only whitespace characters
+                         # and ASCII-only non-whitespace characters are
+                         # allowed.
+        """,
+        re.VERBOSE,
+    )
+
     _operators = {
         "~=": "compatible",
         "==": "equal",
@@ -422,6 +439,13 @@ class Specifier(_IndividualSpecifier):
         "===": "arbitrary",
     }
 
+    def __init__(self, spec: str = "", prereleases: Optional[bool] = None) -> None:
+        super().__init__(spec, prereleases)
+
+        match = self._supplementary_restriction_regex.fullmatch(spec)
+        if not match:
+            raise InvalidSpecifier(f"Invalid specifier: '{spec}'")
+
     @_require_version_compare
     def _compare_compatible(self, prospective: ParsedVersion, spec: str) -> bool:
 

packaging/version.py

@@ -256,12 +256,20 @@ def _legacy_cmpkey(version: str) -> LegacyCmpKey:
 
 class Version(_BaseVersion):
 
-    _regex = re.compile(r"^\s*" + VERSION_PATTERN + r"\s*$", re.VERBOSE | re.IGNORECASE)
+    _regex = re.compile(
+        VERSION_PATTERN,
+        # Note: the re.ASCII flag is necessary because without it the
+        # 'a-z' character ranges in VERSION_PATTERN, in conjunction
+        # with re.IGNORECASE, would cause erroneous acceptance of
+        # non-ASCII letters in the local version segment (see:
+        # https://docs.python.org/library/re.html#re.IGNORECASE).
+        re.VERBOSE | re.IGNORECASE | re.ASCII,
+    )
 
     def __init__(self, version: str) -> None:
 
         # Validate the version and parse it into pieces
-        match = self._regex.search(version)
+        match = self._regex.fullmatch(version.strip())
         if not match:
             raise InvalidVersion(f"Invalid version: '{version}'")
 

tests/test_specifiers.py

@@ -81,6 +81,9 @@ def test_specifiers_valid(self, specifier):
             # Cannot use a prefix matching after a .devN version
             "==1.0.dev1.*",
             "!=1.0.dev1.*",
+            # Local version which includes a non-ASCII letter that matches
+            # regex '[a-z]' when re.IGNORECASE is in force and re.ASCII is not
+            "==1.0+\u0130",
         ],
     )
     def test_specifiers_invalid(self, specifier):
@@ -197,6 +200,7 @@ def test_specifiers_invalid(self, specifier):
             # Various other normalizations
             "v1.0",
             "  \r \f \v v1.0\t\n",
+            "  \r\N{NARROW NO-BREAK SPACE}\v v1.0\N{PARAGRAPH SEPARATOR}",
         ],
     )
     def test_specifiers_normalized(self, version):
@@ -221,6 +225,7 @@ def test_specifiers_normalized(self, version):
             ("~=2.0", "~=2.0"),
             # Spaces should be removed
             ("< 2", "<2"),
+            ("<\N{HAIR SPACE}2", "<2"),
         ],
     )
     def test_specifiers_str_and_repr(self, specifier, expected):

tests/test_version.py

@@ -96,6 +96,9 @@ def test_valid_versions(self, version):
             "1.0+_foobar",
             "1.0+foo&asd",
             "1.0+1+1",
+            # Local version which includes a non-ASCII letter that matches
+            # regex '[a-z]' when re.IGNORECASE is in force and re.ASCII is not
+            "1.0+\u0130",
         ],
     )
     def test_invalid_versions(self, version):
@@ -218,6 +221,7 @@ def test_invalid_versions(self, version):
             # Various other normalizations
             ("v1.0", "1.0"),
             ("   v1.0\t\n", "1.0"),
+            ("\N{NARROW NO-BREAK SPACE}1.0\t\N{PARAGRAPH SEPARATOR}\n ", "1.0"),
         ],
     )
     def test_normalized_versions(self, version, normalized):