Use parse_wheel_filename to parse wheel filename, with fallback

Author: notatallshaw

news/13229.bugfix.rst

@@ -0,0 +1,4 @@
+Parse wheel filenames according to `binary distribution format specification
+<https://packaging.python.org/en/latest/specifications/binary-distribution-format/#file-format>`_.
+When a filename doesn't match the spec a deprecation warning is emitted and the
+filename is parsed using the old method.

src/pip/_internal/index/package_finder.py

@@ -514,11 +514,7 @@ def _sort_key(self, candidate: InstallationCandidate) -> CandidateSortingKey:
                 )
             if self._prefer_binary:
                 binary_preference = 1
-            if wheel.build_tag is not None:
-                match = re.match(r"^(\d+)(.*)$", wheel.build_tag)
-                assert match is not None, "guaranteed by filename validation"
-                build_tag_groups = match.groups()
-                build_tag = (int(build_tag_groups[0]), build_tag_groups[1])
+            build_tag = wheel.build_tag
         else:  # sdist
             pri = -(support_num)
         has_allowed_hash = int(link.is_hash_allowed(self._hashes))

src/pip/_internal/metadata/importlib/_envs.py

@@ -8,10 +8,14 @@
 import zipimport
 from typing import Iterator, List, Optional, Sequence, Set, Tuple
 
-from pip._vendor.packaging.utils import NormalizedName, canonicalize_name
+from pip._vendor.packaging.utils import (
+    InvalidWheelFilename,
+    NormalizedName,
+    canonicalize_name,
+    parse_wheel_filename,
+)
 
 from pip._internal.metadata.base import BaseDistribution, BaseEnvironment
-from pip._internal.models.wheel import Wheel
 from pip._internal.utils.deprecation import deprecated
 from pip._internal.utils.filetypes import WHEEL_EXTENSION
 
@@ -26,7 +30,9 @@ def _looks_like_wheel(location: str) -> bool:
         return False
     if not os.path.isfile(location):
         return False
-    if not Wheel.wheel_file_re.match(os.path.basename(location)):
+    try:
+        parse_wheel_filename(os.path.basename(location))
+    except InvalidWheelFilename:
         return False
     return zipfile.is_zipfile(location)
 

src/pip/_internal/models/wheel.py

@@ -3,13 +3,13 @@
 """
 
 import re
-from typing import Dict, Iterable, List
+from typing import Dict, Iterable, List, Optional
 
 from pip._vendor.packaging.tags import Tag
+from pip._vendor.packaging.utils import BuildTag, parse_wheel_filename
 from pip._vendor.packaging.utils import (
-    InvalidWheelFilename as PackagingInvalidWheelName,
+    InvalidWheelFilename as _PackagingInvalidWheelFilename,
 )
-from pip._vendor.packaging.utils import parse_wheel_filename
 
 from pip._internal.exceptions import InvalidWheelFilename
 from pip._internal.utils.deprecation import deprecated
@@ -18,54 +18,75 @@
 class Wheel:
     """A wheel file"""
 
-    wheel_file_re = re.compile(
+    legacy_wheel_file_re = re.compile(
         r"""^(?P<namever>(?P<name>[^\s-]+?)-(?P<ver>[^\s-]*?))
         ((-(?P<build>\d[^-]*?))?-(?P<pyver>[^\s-]+?)-(?P<abi>[^\s-]+?)-(?P<plat>[^\s-]+?)
         \.whl|\.dist-info)$""",
         re.VERBOSE,
     )
 
     def __init__(self, filename: str) -> None:
-        """
-        :raises InvalidWheelFilename: when the filename is invalid for a wheel
-        """
-        wheel_info = self.wheel_file_re.match(filename)
-        if not wheel_info:
-            raise InvalidWheelFilename(f"{filename} is not a valid wheel filename.")
         self.filename = filename
-        self.name = wheel_info.group("name").replace("_", "-")
-        _version = wheel_info.group("ver")
-        if "_" in _version:
-            try:
-                parse_wheel_filename(filename)
-            except PackagingInvalidWheelName as e:
-                deprecated(
-                    reason=(
-                        f"Wheel filename {filename!r} is not correctly normalised. "
-                        "Future versions of pip will raise the following error:\n"
-                        f"{e.args[0]}\n\n"
-                    ),
-                    replacement=(
-                        "to rename the wheel to use a correctly normalised "
-                        "name (this may require updating the version in "
-                        "the project metadata)"
-                    ),
-                    gone_in="25.1",
-                    issue=12938,
-                )
-
-            _version = _version.replace("_", "-")
-
-        self.version = _version
-        self.build_tag = wheel_info.group("build")
-        self.pyversions = wheel_info.group("pyver").split(".")
-        self.abis = wheel_info.group("abi").split(".")
-        self.plats = wheel_info.group("plat").split(".")
-
-        # All the tag combinations from this file
-        self.file_tags = {
-            Tag(x, y, z) for x in self.pyversions for y in self.abis for z in self.plats
-        }
+
+        # To make mypy happy specify type hints that can come from either
+        # parse_wheel_filename or the legacy_wheel_file_re match.
+        self.name: str
+        self._build_tag: Optional[BuildTag] = None
+
+        try:
+            wheel_info = parse_wheel_filename(filename)
+            self.name, _version, self._build_tag, self.file_tags = wheel_info
+            self.version = str(_version)
+        except _PackagingInvalidWheelFilename as e:
+            # Check if the wheel filename is in the legacy format
+            legacy_wheel_info = self.legacy_wheel_file_re.match(filename)
+            if not legacy_wheel_info:
+                raise InvalidWheelFilename(e.args[0]) from None
+
+            deprecated(
+                reason=(
+                    f"Wheel filename {filename!r} is not correctly normalised. "
+                    "Future versions of pip will raise the following error:\n"
+                    f"{e.args[0]}\n\n"
+                ),
+                replacement=(
+                    "to rename the wheel to use a correctly normalised "
+                    "name (this may require updating the version in "
+                    "the project metadata)"
+                ),
+                gone_in="25.3",
+                issue=12938,
+            )
+
+            self.name = legacy_wheel_info.group("name").replace("_", "-")
+            self.version = legacy_wheel_info.group("ver").replace("_", "-")
+
+            # Generate the file tags from the legacy wheel filename
+            pyversions = legacy_wheel_info.group("pyver").split(".")
+            abis = legacy_wheel_info.group("abi").split(".")
+            plats = legacy_wheel_info.group("plat").split(".")
+            self.file_tags = frozenset(
+                Tag(interpreter=py, abi=abi, platform=plat)
+                for py in pyversions
+                for abi in abis
+                for plat in plats
+            )
+
+    @property
+    def build_tag(self) -> BuildTag:
+        if self._build_tag is not None:
+            return self._build_tag
+
+        # Parse the build tag from the legacy wheel filename
+        legacy_wheel_info = self.legacy_wheel_file_re.match(self.filename)
+        assert legacy_wheel_info is not None, "guaranteed by filename validation"
+        build_tag = legacy_wheel_info.group("build")
+        match = re.match(r"^(\d+)(.*)$", build_tag)
+        assert match is not None, "guaranteed by filename validation"
+        build_tag_groups = match.groups()
+        self._build_tag = (int(build_tag_groups[0]), build_tag_groups[1])
+
+        return self._build_tag
 
     def get_formatted_file_tags(self) -> List[str]:
         """Return the wheel's tags as a sorted list of strings."""

tests/functional/test_install_wheel.py

@@ -190,7 +190,7 @@ def test_install_from_wheel_with_headers(script: PipTestEnvironment) -> None:
     dist_info_folder = script.site_packages / "headers.dist-0.1.dist-info"
     result.did_create(dist_info_folder)
 
-    header_scheme_path = get_header_scheme_path_for_script(script, "headers.dist")
+    header_scheme_path = get_header_scheme_path_for_script(script, "headers-dist")
     header_path = header_scheme_path / "header.h"
     assert header_path.read_text() == header_text
 

tests/unit/test_models_wheel.py

@@ -12,34 +12,43 @@ def test_std_wheel_pattern(self) -> None:
         w = Wheel("simple-1.1.1-py2-none-any.whl")
         assert w.name == "simple"
         assert w.version == "1.1.1"
-        assert w.pyversions == ["py2"]
-        assert w.abis == ["none"]
-        assert w.plats == ["any"]
+        assert w.build_tag == ()
+        assert w.file_tags == frozenset(
+            [Tag(interpreter="py2", abi="none", platform="any")]
+        )
 
     def test_wheel_pattern_multi_values(self) -> None:
         w = Wheel("simple-1.1-py2.py3-abi1.abi2-any.whl")
         assert w.name == "simple"
         assert w.version == "1.1"
-        assert w.pyversions == ["py2", "py3"]
-        assert w.abis == ["abi1", "abi2"]
-        assert w.plats == ["any"]
+        assert w.build_tag == ()
+        assert w.file_tags == frozenset(
+            [
+                Tag(interpreter="py2", abi="abi1", platform="any"),
+                Tag(interpreter="py2", abi="abi2", platform="any"),
+                Tag(interpreter="py3", abi="abi1", platform="any"),
+                Tag(interpreter="py3", abi="abi2", platform="any"),
+            ]
+        )
 
     def test_wheel_with_build_tag(self) -> None:
         # pip doesn't do anything with build tags, but theoretically, we might
         # see one, in this case the build tag = '4'
         w = Wheel("simple-1.1-4-py2-none-any.whl")
         assert w.name == "simple"
         assert w.version == "1.1"
-        assert w.pyversions == ["py2"]
-        assert w.abis == ["none"]
-        assert w.plats == ["any"]
+        assert w.build_tag == (4, "")
+        assert w.file_tags == frozenset(
+            [Tag(interpreter="py2", abi="none", platform="any")]
+        )
 
     def test_single_digit_version(self) -> None:
         w = Wheel("simple-1-py2-none-any.whl")
         assert w.version == "1"
 
     def test_non_pep440_version(self) -> None:
-        w = Wheel("simple-_invalid_-py2-none-any.whl")
+        with pytest.warns(deprecation.PipDeprecationWarning):
+            w = Wheel("simple-_invalid_-py2-none-any.whl")
         assert w.version == "-invalid-"
 
     def test_missing_version_raises(self) -> None: