fix(specifiers): === uses original string, not normalized (#1124)

Author: notatallshaw

src/packaging/specifiers.py

@@ -727,13 +727,21 @@ def filter(
         # Filter versions
         for version in iterable:
             parsed_version = _coerce_version(version if key is None else key(version))
+            match = False
             if parsed_version is None:
                 # === operator can match arbitrary (non-version) strings
                 if self.operator == "===" and self._compare_arbitrary(
                     version, self.version
                 ):
                     yield version
-            elif operator_callable(parsed_version, self.version):
+            elif self.operator == "===":
+                match = self._compare_arbitrary(
+                    version if key is None else key(version), self.version
+                )
+            else:
+                match = operator_callable(parsed_version, self.version)
+
+            if match and parsed_version is not None:
                 # If it's not a prerelease or prereleases are allowed, yield it directly
                 if not parsed_version.is_prerelease or include_prereleases:
                     found_non_prereleases = True
@@ -898,7 +906,13 @@ class SpecifierSet(BaseSpecifier):
     specifiers (``>=3.0,!=3.1``), or no specifier at all.
     """
 
-    __slots__ = ("_canonicalized", "_prereleases", "_resolved_ops", "_specs")
+    __slots__ = (
+        "_canonicalized",
+        "_has_arbitrary",
+        "_prereleases",
+        "_resolved_ops",
+        "_specs",
+    )
 
     def __init__(
         self,
@@ -928,8 +942,13 @@ def __init__(
             split_specifiers = [s.strip() for s in specifiers.split(",") if s.strip()]
 
             self._specs: tuple[Specifier, ...] = tuple(map(Specifier, split_specifiers))
+            # Fast substring check; avoids iterating parsed specs.
+            self._has_arbitrary = "===" in specifiers
         else:
             self._specs = tuple(specifiers)
+            # Substring check works for both Specifier objects and plain
+            # strings (setuptools passes lists of strings).
+            self._has_arbitrary = any("===" in str(s) for s in self._specs)
 
         self._canonicalized = len(self._specs) <= 1
         self._resolved_ops: list[tuple[CallableOperator, str, str]] | None = None
@@ -1025,6 +1044,7 @@ def __and__(self, other: SpecifierSet | str) -> SpecifierSet:
         specifier = SpecifierSet()
         specifier._specs = self._specs + other._specs
         specifier._canonicalized = len(specifier._specs) <= 1
+        specifier._has_arbitrary = self._has_arbitrary or other._has_arbitrary
         specifier._resolved_ops = None
 
         # Combine prerelease settings: use common or non-None value
@@ -1137,7 +1157,12 @@ def contains(
         if version is not None and installed and version.is_prerelease:
             prereleases = True
 
-        check_item = item if version is None else version
+        # When item is a string and === is involved, keep it as-is
+        # so the comparison isn't done against the normalized form.
+        if version is None or (self._has_arbitrary and not isinstance(item, Version)):
+            check_item = item
+        else:
+            check_item = version
         return bool(list(self.filter([check_item], prereleases=prereleases)))
 
     @typing.overload
@@ -1289,6 +1314,11 @@ def _filter_versions(
                     yield item
             elif exclude_prereleases and parsed.is_prerelease:
                 pass
-            elif all(op_fn(parsed, ver) for op_fn, ver, _ in ops):
+            elif all(
+                str(item if key is None else key(item)).lower() == ver.lower()
+                if op == "==="
+                else op_fn(parsed, ver)
+                for op_fn, ver, op in ops
+            ):
                 # Short-circuits on the first failing operator.
                 yield item

tests/test_specifiers.py

@@ -691,6 +691,61 @@ def test_arbitrary_equality(
         spec = Specifier(spec_str)
         assert spec.contains(version) == expected
 
+    @pytest.mark.parametrize(
+        ("spec_str", "version", "expected"),
+        [
+            # Zero padding: unnormalized spec vs string/Version
+            # Strings preserve their original form, so "1.01" != "1.1"
+            ("===1.1", "1.01", False),
+            ("===1.01", "1.1", False),
+            ("===1.01", "1.01", True),
+            ("===1.1", "1.1", True),
+            # Version objects are normalized, so Version("1.01") -> "1.1"
+            ("===1.1", Version("1.01"), True),
+            ("===1.1", Version("1.1"), True),
+            ("===1.01", Version("1.01"), False),
+            ("===1.01", Version("1.1"), False),
+            # Prerelease separator normalization (issue #766)
+            # "1.a1" is valid PEP 440, normalizes to "1a1"
+            ("===1.a1", "1.a1", True),
+            ("===1a1", "1.a1", False),
+            ("===1.a1", "1a1", False),
+            ("===1a1", "1a1", True),
+            ("===1.a1", Version("1.a1"), False),
+            ("===1a1", Version("1.a1"), True),
+            # Epoch normalization: "0!1.0" normalizes to "1.0"
+            ("===0!1.0", "0!1.0", True),
+            ("===0!1.0", "1.0", False),
+            ("===1.0", "0!1.0", False),
+            ("===0!1.0", Version("1.0"), False),
+            ("===1.0", Version("0!1.0"), True),
+            # Leading zeros in release segments
+            ("===01.0", "01.0", True),
+            ("===01.0", "1.0", False),
+            ("===1.0", "01.0", False),
+            ("===01.0", Version("1.0"), False),
+            ("===1.0", Version("01.0"), True),
+            # Post-release normalization: "post" vs "-" separator
+            ("===1.0.post1", "1.0.post1", True),
+            ("===1.0-1", "1.0-1", True),
+            ("===1.0-1", "1.0.post1", False),
+            ("===1.0.post1", "1.0-1", False),
+            ("===1.0-1", Version("1.0.post1"), False),
+            ("===1.0.post1", Version("1.0-1"), True),
+            # Dev normalization
+            ("===1.0.dev01", "1.0.dev01", True),
+            ("===1.0.dev01", "1.0.dev1", False),
+            ("===1.0.dev1", "1.0.dev01", False),
+            ("===1.0.dev01", Version("1.0.dev1"), False),
+            ("===1.0.dev1", Version("1.0.dev01"), True),
+        ],
+    )
+    def test_arbitrary_equality_normalization(
+        self, spec_str: str, version: str | Version, expected: bool
+    ) -> None:
+        spec = Specifier(spec_str, prereleases=True)
+        assert spec.contains(version) == expected
+
     @pytest.mark.parametrize(
         ("specifier", "expected"),
         [
@@ -830,6 +885,33 @@ def test_specifier_filter(
 
         assert result == expected
 
+    @pytest.mark.parametrize(
+        ("specifier", "input", "expected"),
+        [
+            # Strings preserve original form
+            ("===1.01", ["1.01", "1.1", "1.0.1"], ["1.01"]),
+            ("===1.1", ["1.01", "1.1"], ["1.1"]),
+            # Version objects use normalized form
+            (
+                "===1.1",
+                [Version("1.01"), Version("1.1")],
+                [Version("1.01"), Version("1.1")],
+            ),
+            ("===1.01", [Version("1.01"), Version("1.1")], []),
+            # Mixed strings and Version objects
+            ("===1.1", ["1.01", "1.1", Version("1.01")], ["1.1", Version("1.01")]),
+            ("===1.01", ["1.01", "1.1", Version("1.01")], ["1.01"]),
+            # Prerelease separator
+            ("===1.a1", ["1.a1", "1a1"], ["1.a1"]),
+            ("===1a1", ["1.a1", "1a1", Version("1.a1")], ["1a1", Version("1.a1")]),
+        ],
+    )
+    def test_specifier_filter_arbitrary_equality_normalization(
+        self, specifier: str, input: list[str | Version], expected: list[str | Version]
+    ) -> None:
+        spec = Specifier(specifier, prereleases=True)
+        assert list(spec.filter(input)) == expected
+
     @pytest.mark.parametrize(
         ("prereleases", "expected_indexes"),
         [
@@ -1917,6 +1999,37 @@ def test_contains_arbitrary_equality_contains(
         spec = SpecifierSet(specifier)
         assert spec.contains(version) == expected
 
+    @pytest.mark.parametrize(
+        ("spec_str", "version", "expected"),
+        [
+            # Zero padding: string preserves original, Version normalizes
+            ("===1.1", "1.01", False),
+            ("===1.01", "1.1", False),
+            ("===1.01", "1.01", True),
+            ("===1.1", "1.1", True),
+            ("===1.1", Version("1.01"), True),
+            ("===1.01", Version("1.01"), False),
+            # Prerelease separator normalization
+            ("===1.a1", "1.a1", True),
+            ("===1a1", "1.a1", False),
+            ("===1.a1", "1a1", False),
+            ("===1a1", "1a1", True),
+            ("===1.a1", Version("1.a1"), False),
+            ("===1a1", Version("1.a1"), True),
+            # Combined with other operators
+            (">=1.0,===1.01", "1.01", True),
+            (">=1.0,===1.1", "1.1", True),
+            (">=1.0,===1.1", "1.01", False),
+            (">=1.0,===1.1", Version("1.01"), True),
+            (">=1.0,===1.01", Version("1.01"), False),
+        ],
+    )
+    def test_contains_arbitrary_equality_normalization(
+        self, spec_str: str, version: str | Version, expected: bool
+    ) -> None:
+        spec = SpecifierSet(spec_str, prereleases=True)
+        assert spec.contains(version) == expected
+
     @pytest.mark.parametrize(
         ("specifier", "expected"),
         [