bugfix: compare distribution names by parsed forms (#127)

* bugfix: compare distribution names by parsed forms

This fixes an error introduced with #124 -- that PR correctly
dropped our "ultranormalization" of distribution names, but
it didn't account for the fact that we were comparing
distribution names as strings, rather than as parsed
forms.

See pypa/gh-action-pypi-publish#365
for more context.

Signed-off-by: William Woodruff <[email protected]>

* ignore long line

Signed-off-by: William Woodruff <[email protected]>

---------

Signed-off-by: William Woodruff <[email protected]>

Author: woodruffw

src/pypi_attestations/_impl.py

@@ -10,6 +10,10 @@
 from enum import Enum
 from typing import TYPE_CHECKING, Annotated, Any, Literal, NewType, Optional, Union, get_args
 
+import packaging
+import packaging.tags
+import packaging.utils
+import packaging.version
 import sigstore.errors
 from annotated_types import MinLen  # noqa: TCH002
 from cryptography import x509
@@ -291,14 +295,17 @@ def verify(
             # be an exact match for their distribution filename.
             # See: https://github.com/pypi/warehouse/issues/18128
             # See: https://github.com/trailofbits/pypi-attestations/issues/123
-            _check_dist_filename(subject.name)
-            subject_name = subject.name
+            parsed_subject_name = _check_dist_filename(subject.name)
         except ValueError as e:
             raise VerificationError(f"invalid subject: {str(e)}")
 
-        if subject_name != dist.name:
+        # NOTE: Cannot fail, since we validate the `Distribution` name
+        # on construction.
+        parsed_dist_name = _check_dist_filename(dist.name)
+
+        if parsed_subject_name != parsed_dist_name:
             raise VerificationError(
-                f"subject does not match distribution name: {subject_name} != {dist.name}"
+                f"subject does not match distribution name: {subject.name} != {dist.name}"
             )
 
         digest = subject.digest.root.get("sha256")
@@ -392,7 +399,17 @@ def _der_decode_utf8string(der: bytes) -> str:
     return der_decode(der, UTF8String)[0].decode()  # type: ignore[no-any-return]
 
 
-def _check_dist_filename(dist: str) -> None:
+_SdistName = tuple[packaging.utils.NormalizedName, packaging.version.Version]
+_BdistName = tuple[
+    packaging.utils.NormalizedName,
+    packaging.version.Version,
+    packaging.utils.BuildTag,
+    frozenset[packaging.tags.Tag],
+]
+_DistName = Union[_SdistName, _BdistName]
+
+
+def _check_dist_filename(dist: str) -> _DistName:
     """Validate a distribution filename for well-formedness.
 
     This does **not** fully normalize the filename. For example,
@@ -406,10 +423,10 @@ def _check_dist_filename(dist: str) -> None:
     # already rejects non-lowercase variants.
     if dist.endswith(".whl"):
         # `parse_wheel_filename` raises a supertype of ValueError on failure.
-        parse_wheel_filename(dist)
+        return parse_wheel_filename(dist)
     elif dist.endswith((".tar.gz", ".zip")):
         # `parse_sdist_filename` raises a supertype of ValueError on failure.
-        parse_sdist_filename(dist)
+        return parse_sdist_filename(dist)
     else:
         raise ValueError(f"unknown distribution format: {dist}")
 

test/assets/spt3g-1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.publish.attestation

@@ -0,0 +1 @@
+{"version":1,"verification_material":{"certificate":"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","transparency_entries":[{"logIndex":"43047182","logId":{"keyId":"0y8wo8MtY5wrdiIFohx7sHeI5oKDpK5vQhGHI6G+pJY="},"kindVersion":{"kind":"dsse","version":"0.0.1"},"integratedTime":"1748916708","inclusionPromise":{"signedEntryTimestamp":"MEUCIQCfM8ovwc04CEOHgX6fGf5Pzz82+vDTsHslOBVUGezYIAIgGn7dk6tO7J39kmf/qnlxmZPydLBJoncbLNp6E8fvgFM="},"inclusionProof":{"logIndex":"11364770","rootHash":"ENMXdOX33Y4lNWu52ZxECkS2DK3mEFZgkfS/kSRm5oA=","treeSize":"11364771","hashes":["AWheyozzascmgR49/VMzWgWt9UmWTNKIf2rc6PuTugA=","kumo1jNbrVlSyW2vbHZ0ZpHoWT2V1JvhDp24nCXPPC8=","JyRDG0Zq0G+9t/GT8bw4OZT1wp5JTuvBg15t12oQYXw=","xIvGcJZcwnG+yR9P1yHBaTTAyorVGqdeoa5jid5x97c=","V48BwQhJSweCIRk7yVu1rhWZ8oFfO9Qxs3SZbzVJ0kw=","58Noh/WgnaJutPtj9YL2QRs6Kp42XcPEibW2RlJltE8=","hiBsgkAwBBrSUXxyhhIXU+eEpNvBb20KnrGpN0zTnkU=","QIeOgI91ZwOrM9y+4z277AZ241pn7POReCZzUU/gM8k=","kPnAtj3bBHoLjYDGPJ3n/tPr3Zuy0BYmlTJOFsUAiHU=","4X3ZcZ5JupJMfhwYrQk+zbHb+9n5mEEw+A6nkd7pL0g=","3itMWeZPQXMyYki8lnPZVbzzneXD3w/xvhQsm7VP1ck=","OdoqbUqBYHhj2W1RLM8APkQOnM2K9gzGm1KPFmwIIeQ="],"checkpoint":{"envelope":"rekor.sigstage.dev - 8202293616175992157\n11364771\nENMXdOX33Y4lNWu52ZxECkS2DK3mEFZgkfS/kSRm5oA=\n\n— rekor.sigstage.dev 0y8wozBFAiAYfxAWPCarnkqNKYuOWM332aYBjoqbunGVZpRwpWbBrAIhAJSmUeuXeFaS1aWT57W7u0o5hQatvLGlYtjUYXFXZc7y\n"}},"canonicalizedBody":"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"}]},"envelope":{"statement":"eyJfdHlwZSI6Imh0dHBzOi8vaW4tdG90by5pby9TdGF0ZW1lbnQvdjEiLCJzdWJqZWN0IjpbeyJuYW1lIjoic3B0M2ctMS4wLWNwMzEwLWNwMzEwLW1hbnlsaW51eF8yXzE3X3g4Nl82NC5tYW55bGludXgyMDE0X3g4Nl82NC53aGwiLCJkaWdlc3QiOnsic2hhMjU2IjoiZDI3NzJmOWE1MTk5ZjA1ZWQxYmU4ZDlhYTc4Yjg3OWU1MTc3MmUzZWFkOWQ3M2ZlODA1NzI1N2IxYWVjN2NmOCJ9fV0sInByZWRpY2F0ZVR5cGUiOiJodHRwczovL2RvY3MucHlwaS5vcmcvYXR0ZXN0YXRpb25zL3B1Ymxpc2gvdjEiLCJwcmVkaWNhdGUiOm51bGx9","signature":"MEUCIQD3neI5rll21OHPazzuLDGfIFYEh904YWdvHdkj1RZmowIgKz5RRk7I1CB6xKHc8oT1UXuLOu2w6g7H4QI7VBThJPU="}}

test/test_impl.py

@@ -458,6 +458,32 @@ def test_certificate_claims(self) -> None:
 
         assert not results ^ set(attestation.certificate_claims.items())
 
+    def test_verify_different_wheel_tag_order(self) -> None:
+        attestation_path = (
+            _ASSETS
+            / "spt3g-1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.publish.attestation"  # noqa: E501
+        )
+
+        attestation = impl.Attestation.model_validate_json(attestation_path.read_bytes())
+
+        pol = policy.Identity(
+            identity="[email protected]", issuer="https://github.com/login/oauth"
+        )
+
+        dist = impl.Distribution(
+            # Distribution intentionally has a different tag order.
+            name="spt3g-1.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl",
+            digest="d2772f9a5199f05ed1be8d9aa78b879e51772e3ead9d73fe8057257b1aec7cf8",
+        )
+
+        attestation.verify(pol, dist, staging=True, offline=True)
+
+        # Distribution names are not string equivalent, but do compare
+        # as equal when parsed.
+        subject_name = attestation.statement["subject"][0]["name"]
+        assert impl._check_dist_filename(subject_name) == impl._check_dist_filename(dist.name)
+        assert subject_name != dist.name
+
 
 def test_from_bundle_missing_signatures() -> None:
     bundle = Bundle.from_json(dist_bundle_path.read_bytes())