Fix parsing of wheel version numbers

changelog.d/+wheel-parsing-prerelease-versions.bugfix.rst

@@ -0,0 +1 @@
+Fix a bug where parsing of wheel filenames in test support code would miss prerelease version numbers

setup.cfg

@@ -60,6 +60,7 @@ testing =
 		python_version < "3.8"
 	importlib-metadata; \
 		python_version < "3.8"
+	wheel-filename
 	# no version of pyproject-metadata supports Python 3.6
 	pyproject-metadata; \
 		python_version >= "3.7"

test_support/test_support/distribution.py

@@ -28,7 +28,8 @@
 from abc import ABC, abstractmethod
 from test_support import importlib_metadata, Project
 from test_support.metadata import parse_core_metadata
-from typing import Any, IO, Iterable, List, Optional, Sequence
+from typing import Any, IO, Iterable, List, NamedTuple, Optional, Sequence, Union
+from wheel_filename import ParsedWheelFilename, parse_wheel_filename
 
 try:
     from pyproject_metadata import RFC822Message, StandardMetadata
@@ -95,18 +96,28 @@ def __init__(self, package_url: str, package_hash_spec: str, metadata_hash_spec:
             self.metadata_hasher = None
 
 
+class ParsedSdistFilename(NamedTuple):
+    project: str
+    version: str
+
+
+def _parse_package_filename(filename: str) -> Union[ParsedSdistFilename, ParsedWheelFilename]:
+    if filename.endswith(".whl"):
+        return parse_wheel_filename(filename)
+    elif filename.endswith(".tar.gz"):
+        name, _, version = filename[:-7].partition("-")
+        return ParsedSdistFilename(name, version)
+    else:
+        raise ValueError(f"Not a valid Python package filename: {filename}")
+
+
 class SimplePackageListingParser(html.parser.HTMLParser):
     """
     A bare-bones parser for the list of versions of a given package offered by
     the simple repository API. It will select releases of the given version of
     the package.
     """
 
-    _filename_regex = re.compile(
-        r"(?P<name>[\w.-]+)-(?P<version>{}).+\.(?:whl|tar\.gz)".format(packaging.version.VERSION_PATTERN),
-        flags=re.VERBOSE | re.IGNORECASE,
-    )
-
     @staticmethod
     def _normalize_package_name(name: str):
         """
@@ -139,10 +150,16 @@ def handle_starttag(self, tag: Any, attrs: Any):
         parsed_url: urllib.parse.ParseResult = urllib.parse.urlparse(href)
         filename: str
         _, _, filename = parsed_url.path.rpartition("/")
-        m = self._filename_regex.match(filename)
-        if not m or self.version != packaging.version.parse(m.group("version")):
+        try:
+            parsed_filename = _parse_package_filename(filename)
+        except ValueError:
+            _logger.debug("no filename match")
+            return
+
+        if self.version != (ver := packaging.version.parse(parsed_filename.version)):
+            _logger.debug("version mismatch: %s vs %s", self.version, ver)
             return
-        assert self.normalized_name == self._normalize_package_name(m.group("name"))
+        assert self.normalized_name == self._normalize_package_name(parsed_filename.project)
         no_fragment_url: str = urllib.parse.urlunparse(list(parsed_url[:5]) + [""])
         self.releases.append(PackageInfo(no_fragment_url, parsed_url.fragment, data_dist_info_metadata))
 

tests/test_filename_matching.py

@@ -0,0 +1,82 @@
+"""
+Tests that verify the logic we use to match sdist and wheel filenames is
+consistent with the packaging standards.
+"""
+
+import pytest
+
+from test_support.distribution import ParsedSdistFilename, _parse_package_filename
+from typing import List, Optional, Tuple
+from wheel_filename import ParsedWheelFilename
+
+SDIST: List[Tuple[str, str]] = [
+    ("p", "1.0"),
+    ("package", "1.0"),
+    ("package_name", "1.0"),
+    ("package", "0.1"),
+    ("package", "0.0.1"),
+    ("package", "0.0.100"),
+    ("package", "1.post0"),
+    ("package", "2024.09.06"),
+    ("package", "20240906"),
+    ("package", "1a"),
+    ("package", "1a0"),
+    ("package", "1a1"),
+    ("package", "1a10"),
+    ("package", "1b"),
+    ("package", "1b0"),
+    ("package", "1b1"),
+    ("package", "1rc"),
+    ("package", "1rc0"),
+    ("package", "1rc1"),
+]
+
+
+@pytest.mark.parametrize(("name", "version"), SDIST)
+class TestSDistFilename:
+    @pytest.fixture
+    def match_filename(self, name: str, version: str) -> ParsedSdistFilename:
+        filename = f"{name}-{version}.tar.gz"
+        parsed = _parse_package_filename(filename)
+        assert isinstance(parsed, ParsedSdistFilename)
+        return parsed
+
+    def test_project(self, name: str, version: str, match_filename: ParsedSdistFilename) -> None:
+        assert match_filename.project == name
+
+    def test_version(self, name: str, version: str, match_filename: ParsedSdistFilename) -> None:
+        assert match_filename.version == version
+
+
+WHEEL_TAGS: List[Tuple[Optional[str], str, str, str]] = [
+    (None, "py3", "none", "any"),
+    ("1", "py3", "none", "any"),
+    (None, "py38", "cp38", "any"),
+    (None, "py3", "none", "linux_86_64"),
+]
+
+
+@pytest.mark.parametrize(("name", "version"), SDIST)
+@pytest.mark.parametrize(("build", "python", "abi", "platform"), WHEEL_TAGS)
+class TestWheelFilename:
+    @pytest.fixture
+    def match_filename(
+        self, name: str, version: str, build: str, python: str, abi: str, platform: str
+    ) -> ParsedWheelFilename:
+        filename = f"{name}-{version}"
+        if build:
+            filename += f"-{build}"
+        filename += f"-{python}-{abi}-{platform}.whl"
+        parsed = _parse_package_filename(filename)
+        assert isinstance(parsed, ParsedWheelFilename)
+        return parsed
+
+    def test_name(
+        self, name: str, version: str, python: str, abi: str, platform: str, match_filename: ParsedWheelFilename
+    ) -> None:
+        assert match_filename.project == name
+
+    def test_version(
+        self, name: str, version: str, python: str, abi: str, platform: str, match_filename: ParsedWheelFilename
+    ) -> None:
+        assert match_filename.version == version