cache "concrete" dists by Distribution instead of InstallRequirement

news/12863.trivial.rst

@@ -0,0 +1 @@
+Cache "concrete" dists by ``Distribution`` instead of ``InstallRequirement``.

src/pip/_internal/distributions/__init__.py

@@ -1,4 +1,5 @@
 from pip._internal.distributions.base import AbstractDistribution
+from pip._internal.distributions.installed import InstalledDistribution
 from pip._internal.distributions.sdist import SourceDistribution
 from pip._internal.distributions.wheel import WheelDistribution
 from pip._internal.req.req_install import InstallRequirement
@@ -8,6 +9,10 @@ def make_distribution_for_install_requirement(
     install_req: InstallRequirement,
 ) -> AbstractDistribution:
     """Returns a Distribution for the given InstallRequirement"""
+    # Only pre-installed requirements will have a .satisfied_by dist.
+    if install_req.satisfied_by:
+        return InstalledDistribution(install_req)
+
     # Editable requirements will always be source distributions. They use the
     # legacy logic until we create a modern standard for them.
     if install_req.editable:

src/pip/_internal/distributions/base.py

@@ -39,11 +39,17 @@ def build_tracker_id(self) -> str | None:
 
         If None, then this dist has no work to do in the build tracker, and
         ``.prepare_distribution_metadata()`` will not be called."""
-        raise NotImplementedError()
+        ...
 
     @abc.abstractmethod
     def get_metadata_distribution(self) -> BaseDistribution:
-        raise NotImplementedError()
+        """Generate a concrete ``BaseDistribution`` instance for this artifact.
+
+        The implementation should also cache the result with
+        ``self.req.cache_concrete_dist()`` so the distribution is available to other
+        users of the ``InstallRequirement``. This method is not called within the build
+        tracker context, so it should not identify any new setup requirements."""
+        ...
 
     @abc.abstractmethod
     def prepare_distribution_metadata(
@@ -52,4 +58,11 @@ def prepare_distribution_metadata(
         build_isolation: bool,
         check_build_deps: bool,
     ) -> None:
-        raise NotImplementedError()
+        """Generate the information necessary to extract metadata from the artifact.
+
+        This method will be executed within the context of ``BuildTracker#track()``, so
+        it needs to fully identify any setup requirements so they can be added to the
+        same active set of tracked builds, while ``.get_metadata_distribution()`` takes
+        care of generating and caching the ``BaseDistribution`` to expose to the rest of
+        the resolve."""
+        ...

src/pip/_internal/distributions/installed.py

@@ -21,8 +21,10 @@ def build_tracker_id(self) -> str | None:
         return None
 
     def get_metadata_distribution(self) -> BaseDistribution:
-        assert self.req.satisfied_by is not None, "not actually installed"
-        return self.req.satisfied_by
+        dist = self.req.satisfied_by
+        assert dist is not None, "not actually installed"
+        self.req.cache_concrete_dist(dist)
+        return dist
 
     def prepare_distribution_metadata(
         self,

src/pip/_internal/distributions/sdist.py

@@ -7,7 +7,7 @@
 from pip._internal.build_env import BuildEnvironment
 from pip._internal.distributions.base import AbstractDistribution
 from pip._internal.exceptions import InstallationError
-from pip._internal.metadata import BaseDistribution
+from pip._internal.metadata import BaseDistribution, get_directory_distribution
 from pip._internal.utils.subprocess import runner_with_spinner_message
 
 if TYPE_CHECKING:
@@ -24,13 +24,19 @@ class SourceDistribution(AbstractDistribution):
     """
 
     @property
-    def build_tracker_id(self) -> str | None:
+    def build_tracker_id(self) -> str:
         """Identify this requirement uniquely by its link."""
         assert self.req.link
         return self.req.link.url_without_fragment
 
     def get_metadata_distribution(self) -> BaseDistribution:
-        return self.req.get_dist()
+        assert (
+            self.req.metadata_directory
+        ), "Set as part of .prepare_distribution_metadata()"
+        dist = get_directory_distribution(self.req.metadata_directory)
+        self.req.cache_concrete_dist(dist)
+        self.req.validate_sdist_metadata()
+        return dist
 
     def prepare_distribution_metadata(
         self,
@@ -69,7 +75,11 @@ def prepare_distribution_metadata(
                 self._raise_conflicts("the backend dependencies", conflicting)
             if missing:
                 self._raise_missing_reqs(missing)
-        self.req.prepare_metadata()
+
+        # NB: we must still call .cache_concrete_dist() and .validate_sdist_metadata()
+        # before the InstallRequirement itself has been updated with the metadata from
+        # this directory!
+        self.req.prepare_metadata_directory()
 
     def _prepare_build_backend(
         self, build_env_installer: BuildEnvironmentInstaller

src/pip/_internal/distributions/wheel.py

@@ -33,7 +33,9 @@ def get_metadata_distribution(self) -> BaseDistribution:
         assert self.req.local_file_path, "Set as part of preparation during download"
         assert self.req.name, "Wheels are never unnamed"
         wheel = FilesystemWheel(self.req.local_file_path)
-        return get_wheel_distribution(wheel, canonicalize_name(self.req.name))
+        dist = get_wheel_distribution(wheel, canonicalize_name(self.req.name))
+        self.req.cache_concrete_dist(dist)
+        return dist
 
     def prepare_distribution_metadata(
         self,

src/pip/_internal/metadata/base.py

@@ -92,6 +92,15 @@ class RequiresEntry(NamedTuple):
 
 
 class BaseDistribution(Protocol):
+    @property
+    def is_concrete(self) -> bool:
+        """Whether the distribution really exists somewhere on disk.
+
+        If this is false, it has been synthesized from metadata, e.g. via
+        ``.from_metadata_file_contents()``, or ``.from_wheel()`` against
+        a ``MemoryWheel``."""
+        raise NotImplementedError()
+
     @classmethod
     def from_directory(cls, directory: str) -> BaseDistribution:
         """Load the distribution from a metadata directory.
@@ -664,6 +673,10 @@ def iter_installed_distributions(
 class Wheel(Protocol):
     location: str
 
+    @property
+    def is_concrete(self) -> bool:
+        raise NotImplementedError()
+
     def as_zipfile(self) -> zipfile.ZipFile:
         raise NotImplementedError()
 
@@ -672,6 +685,10 @@ class FilesystemWheel(Wheel):
     def __init__(self, location: str) -> None:
         self.location = location
 
+    @property
+    def is_concrete(self) -> bool:
+        return True
+
     def as_zipfile(self) -> zipfile.ZipFile:
         return zipfile.ZipFile(self.location, allowZip64=True)
 
@@ -681,5 +698,9 @@ def __init__(self, location: str, stream: IO[bytes]) -> None:
         self.location = location
         self.stream = stream
 
+    @property
+    def is_concrete(self) -> bool:
+        return False
+
     def as_zipfile(self) -> zipfile.ZipFile:
         return zipfile.ZipFile(self.stream, allowZip64=True)

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

@@ -104,16 +104,22 @@ def __init__(
         dist: importlib.metadata.Distribution,
         info_location: BasePath | None,
         installed_location: BasePath | None,
+        concrete: bool,
     ) -> None:
         self._dist = dist
         self._info_location = info_location
         self._installed_location = installed_location
+        self._concrete = concrete
+
+    @property
+    def is_concrete(self) -> bool:
+        return self._concrete
 
     @classmethod
     def from_directory(cls, directory: str) -> BaseDistribution:
         info_location = pathlib.Path(directory)
         dist = importlib.metadata.Distribution.at(info_location)
-        return cls(dist, info_location, info_location.parent)
+        return cls(dist, info_location, info_location.parent, concrete=True)
 
     @classmethod
     def from_metadata_file_contents(
@@ -130,7 +136,7 @@ def from_metadata_file_contents(
         metadata_path.write_bytes(metadata_contents)
         # Construct dist pointing to the newly created directory.
         dist = importlib.metadata.Distribution.at(metadata_path.parent)
-        return cls(dist, metadata_path.parent, None)
+        return cls(dist, metadata_path.parent, None, concrete=False)
 
     @classmethod
     def from_wheel(cls, wheel: Wheel, name: str) -> BaseDistribution:
@@ -139,7 +145,14 @@ def from_wheel(cls, wheel: Wheel, name: str) -> BaseDistribution:
                 dist = WheelDistribution.from_zipfile(zf, name, wheel.location)
         except zipfile.BadZipFile as e:
             raise InvalidWheel(wheel.location, name) from e
-        return cls(dist, dist.info_location, pathlib.PurePosixPath(wheel.location))
+        except UnsupportedWheel as e:
+            raise UnsupportedWheel(f"{name} has an invalid wheel, {e}")
+        return cls(
+            dist,
+            dist.info_location,
+            pathlib.PurePosixPath(wheel.location),
+            concrete=wheel.is_concrete,
+        )
 
     @property
     def location(self) -> str | None:

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

@@ -86,7 +86,7 @@ def find(self, location: str) -> Iterator[BaseDistribution]:
                 installed_location: BasePath | None = None
             else:
                 installed_location = info_location.parent
-            yield Distribution(dist, info_location, installed_location)
+            yield Distribution(dist, info_location, installed_location, concrete=True)
 
     def find_legacy_editables(self, location: str) -> Iterator[BaseDistribution]:
         """Read location in egg-link files and return distributions in there.
@@ -110,7 +110,7 @@ def find_legacy_editables(self, location: str) -> Iterator[BaseDistribution]:
                 continue
             target_location = str(path.joinpath(target_rel))
             for dist, info_location in self._find_impl(target_location):
-                yield Distribution(dist, info_location, path)
+                yield Distribution(dist, info_location, path, concrete=True)
 
 
 class Environment(BaseEnvironment):

src/pip/_internal/metadata/pkg_resources.py

@@ -78,8 +78,9 @@ def run_script(self, script_name: str, namespace: str) -> None:
 
 
 class Distribution(BaseDistribution):
-    def __init__(self, dist: pkg_resources.Distribution) -> None:
+    def __init__(self, dist: pkg_resources.Distribution, concrete: bool) -> None:
         self._dist = dist
+        self._concrete = concrete
         # This is populated lazily, to avoid loading metadata for all possible
         # distributions eagerly.
         self.__extra_mapping: Mapping[NormalizedName, str] | None = None
@@ -93,6 +94,10 @@ def _extra_mapping(self) -> Mapping[NormalizedName, str]:
 
         return self.__extra_mapping
 
+    @property
+    def is_concrete(self) -> bool:
+        return self._concrete
+
     @classmethod
     def from_directory(cls, directory: str) -> BaseDistribution:
         dist_dir = directory.rstrip(os.sep)
@@ -111,7 +116,7 @@ def from_directory(cls, directory: str) -> BaseDistribution:
             dist_name = os.path.splitext(dist_dir_name)[0].split("-")[0]
 
         dist = dist_cls(base_dir, project_name=dist_name, metadata=metadata)
-        return cls(dist)
+        return cls(dist, concrete=True)
 
     @classmethod
     def from_metadata_file_contents(
@@ -128,7 +133,7 @@ def from_metadata_file_contents(
             metadata=InMemoryMetadata(metadata_dict, filename),
             project_name=project_name,
         )
-        return cls(dist)
+        return cls(dist, concrete=False)
 
     @classmethod
     def from_wheel(cls, wheel: Wheel, name: str) -> BaseDistribution:
@@ -149,7 +154,7 @@ def from_wheel(cls, wheel: Wheel, name: str) -> BaseDistribution:
             metadata=InMemoryMetadata(metadata_dict, wheel.location),
             project_name=name,
         )
-        return cls(dist)
+        return cls(dist, concrete=wheel.is_concrete)
 
     @property
     def location(self) -> str | None:
@@ -261,7 +266,7 @@ def from_paths(cls, paths: list[str] | None) -> BaseEnvironment:
 
     def _iter_distributions(self) -> Iterator[BaseDistribution]:
         for dist in self._ws:
-            yield Distribution(dist)
+            yield Distribution(dist, concrete=True)
 
     def _search_distribution(self, name: str) -> BaseDistribution | None:
         """Find a distribution matching the ``name`` in the environment.

src/pip/_internal/operations/prepare.py

@@ -16,7 +16,6 @@
 
 from pip._internal.build_env import BuildEnvironmentInstaller
 from pip._internal.distributions import make_distribution_for_install_requirement
-from pip._internal.distributions.installed import InstalledDistribution
 from pip._internal.exceptions import (
     DirectoryUrlHashUnsupported,
     HashMismatch,
@@ -200,6 +199,8 @@ def _check_download_dir(
 ) -> str | None:
     """Check download_dir for previously downloaded file with correct hash
     If a correct file is found return its path else None
+
+    If a file is found at the given path, but with an invalid hash, the file is deleted.
     """
     download_path = os.path.join(download_dir, link.filename)
 
@@ -530,7 +531,9 @@ def prepare_linked_requirement(
                 # The file is not available, attempt to fetch only metadata
                 metadata_dist = self._fetch_metadata_only(req)
                 if metadata_dist is not None:
-                    req.needs_more_preparation = True
+                    # These reqs now have the dependency information from the downloaded
+                    # metadata, without having downloaded the actual dist at all.
+                    req.cache_virtual_metadata_only_dist(metadata_dist)
                     return metadata_dist
 
             # None of the optimizations worked, fully prepare the requirement
@@ -540,27 +543,27 @@ def prepare_linked_requirements_more(
         self, reqs: Iterable[InstallRequirement], parallel_builds: bool = False
     ) -> None:
         """Prepare linked requirements more, if needed."""
-        reqs = [req for req in reqs if req.needs_more_preparation]
+        partially_downloaded_reqs: list[InstallRequirement] = []
         for req in reqs:
+            if req.is_concrete:
+                continue
+
             # Determine if any of these requirements were already downloaded.
             if self.download_dir is not None and req.link.is_wheel:
                 hashes = self._get_linked_req_hashes(req)
-                file_path = _check_download_dir(req.link, self.download_dir, hashes)
+                # If the file is there, but doesn't match the hash, delete it and print
+                # a warning. We will be downloading it again via
+                # partially_downloaded_reqs.
+                file_path = _check_download_dir(
+                    req.link, self.download_dir, hashes, warn_on_hash_mismatch=True
+                )
                 if file_path is not None:
+                    # If the hash does match, then we still need to generate a concrete
+                    # dist, but we don't have to download the wheel again.
                     self._downloaded[req.link.url] = file_path
-                    req.needs_more_preparation = False
 
-        # Prepare requirements we found were already downloaded for some
-        # reason. The other downloads will be completed separately.
-        partially_downloaded_reqs: list[InstallRequirement] = []
-        for req in reqs:
-            if req.needs_more_preparation:
-                partially_downloaded_reqs.append(req)
-            else:
-                self._prepare_linked_requirement(req, parallel_builds)
+            partially_downloaded_reqs.append(req)
 
-        # TODO: separate this part out from RequirementPreparer when the v1
-        # resolver can be removed!
         self._complete_partial_requirements(
             partially_downloaded_reqs,
             parallel_builds=parallel_builds,
@@ -661,6 +664,7 @@ def _prepare_linked_requirement(
     def save_linked_requirement(self, req: InstallRequirement) -> None:
         assert self.download_dir is not None
         assert req.link is not None
+        assert req.is_concrete
         link = req.link
         if link.is_vcs or (link.is_existing_dir() and req.editable):
             # Make a .zip of the source_dir we already created.
@@ -715,6 +719,8 @@ def prepare_editable_requirement(
 
             req.check_if_exists(self.use_user_site)
 
+        # This should already have been populated by the preparation of the source dist.
+        assert req.is_concrete
         return dist
 
     def prepare_installed_requirement(
@@ -739,4 +745,13 @@ def prepare_installed_requirement(
                     "completely repeatable environment, install into an "
                     "empty virtualenv."
                 )
-            return InstalledDistribution(req).get_metadata_distribution()
+            dist = _get_prepared_distribution(
+                req,
+                self.build_tracker,
+                self.build_env_installer,
+                self.build_isolation,
+                self.check_build_deps,
+            )
+
+        assert req.is_concrete
+        return dist