More pkg_resources refactoring

src/pip/_internal/commands/show.py

@@ -1,8 +1,6 @@
-import csv
 import logging
-import pathlib
 from optparse import Values
-from typing import Iterator, List, NamedTuple, Optional, Tuple
+from typing import Iterator, List, NamedTuple, Optional
 
 from pip._vendor.packaging.utils import canonicalize_name
 
@@ -69,33 +67,6 @@ class _PackageInfo(NamedTuple):
     files: Optional[List[str]]
 
 
-def _convert_legacy_entry(entry: Tuple[str, ...], info: Tuple[str, ...]) -> str:
-    """Convert a legacy installed-files.txt path into modern RECORD path.
-
-    The legacy format stores paths relative to the info directory, while the
-    modern format stores paths relative to the package root, e.g. the
-    site-packages directory.
-
-    :param entry: Path parts of the installed-files.txt entry.
-    :param info: Path parts of the egg-info directory relative to package root.
-    :returns: The converted entry.
-
-    For best compatibility with symlinks, this does not use ``abspath()`` or
-    ``Path.resolve()``, but tries to work with path parts:
-
-    1. While ``entry`` starts with ``..``, remove the equal amounts of parts
-       from ``info``; if ``info`` is empty, start appending ``..`` instead.
-    2. Join the two directly.
-    """
-    while entry and entry[0] == "..":
-        if not info or info[-1] == "..":
-            info += ("..",)
-        else:
-            info = info[:-1]
-        entry = entry[1:]
-    return str(pathlib.Path(*info, *entry))
-
-
 def search_packages_info(query: List[str]) -> Iterator[_PackageInfo]:
     """
     Gather details from installed distributions. Print distribution name,
@@ -121,34 +92,6 @@ def _get_requiring_packages(current_dist: BaseDistribution) -> Iterator[str]:
             in {canonicalize_name(d.name) for d in dist.iter_dependencies()}
         )
 
-    def _files_from_record(dist: BaseDistribution) -> Optional[Iterator[str]]:
-        try:
-            text = dist.read_text("RECORD")
-        except FileNotFoundError:
-            return None
-        # This extra Path-str cast normalizes entries.
-        return (str(pathlib.Path(row[0])) for row in csv.reader(text.splitlines()))
-
-    def _files_from_legacy(dist: BaseDistribution) -> Optional[Iterator[str]]:
-        try:
-            text = dist.read_text("installed-files.txt")
-        except FileNotFoundError:
-            return None
-        paths = (p for p in text.splitlines(keepends=False) if p)
-        root = dist.location
-        info = dist.info_directory
-        if root is None or info is None:
-            return paths
-        try:
-            info_rel = pathlib.Path(info).relative_to(root)
-        except ValueError:  # info is not relative to root.
-            return paths
-        if not info_rel.parts:  # info *is* root.
-            return paths
-        return (
-            _convert_legacy_entry(pathlib.Path(p).parts, info_rel.parts) for p in paths
-        )
-
     for query_name in query_names:
         try:
             dist = installed[query_name]
@@ -164,7 +107,7 @@ def _files_from_legacy(dist: BaseDistribution) -> Optional[Iterator[str]]:
         except FileNotFoundError:
             entry_points = []
 
-        files_iter = _files_from_record(dist) or _files_from_legacy(dist)
+        files_iter = dist.iter_declared_entries()
         if files_iter is None:
             files: Optional[List[str]] = None
         else:

src/pip/_internal/distributions/installed.py

@@ -11,10 +11,8 @@ class InstalledDistribution(AbstractDistribution):
     """
 
     def get_metadata_distribution(self) -> BaseDistribution:
-        from pip._internal.metadata.pkg_resources import Distribution as _Dist
-
         assert self.req.satisfied_by is not None, "not actually installed"
-        return _Dist(self.req.satisfied_by)
+        return self.req.satisfied_by
 
     def prepare_distribution_metadata(
         self, finder: PackageFinder, build_isolation: bool

src/pip/_internal/distributions/sdist.py

@@ -19,9 +19,7 @@ class SourceDistribution(AbstractDistribution):
     """
 
     def get_metadata_distribution(self) -> BaseDistribution:
-        from pip._internal.metadata.pkg_resources import Distribution as _Dist
-
-        return _Dist(self.req.get_dist())
+        return self.req.get_dist()
 
     def prepare_distribution_metadata(
         self, finder: PackageFinder, build_isolation: bool

src/pip/_internal/index/package_finder.py

@@ -881,7 +881,7 @@ def find_requirement(
 
         installed_version: Optional[_BaseVersion] = None
         if req.satisfied_by is not None:
-            installed_version = parse_version(req.satisfied_by.version)
+            installed_version = req.satisfied_by.version
 
         def _format_versions(cand_iter: Iterable[InstallationCandidate]) -> str:
             # This repeated parse_version and str() conversion is needed to

src/pip/_internal/locations/__init__.py

@@ -64,8 +64,8 @@ def _looks_like_bpo_44860() -> bool:
 
 def _looks_like_red_hat_patched_platlib_purelib(scheme: Dict[str, str]) -> bool:
     platlib = scheme["platlib"]
-    if "/$platlibdir/" in platlib and hasattr(sys, "platlibdir"):
-        platlib = platlib.replace("/$platlibdir/", f"/{sys.platlibdir}/")
+    if "/$platlibdir/" in platlib:
+        platlib = platlib.replace("/$platlibdir/", f"/{_PLATLIBDIR}/")
     if "/lib64/" not in platlib:
         return False
     unpatched = platlib.replace("/lib64/", "/lib/")

src/pip/_internal/metadata/__init__.py

@@ -38,6 +38,17 @@ def get_environment(paths: Optional[List[str]]) -> BaseEnvironment:
     return Environment.from_paths(paths)
 
 
+def get_directory_distribution(directory: str) -> BaseDistribution:
+    """Get the distribution metadata representation in the specified directory.
+
+    This returns a Distribution instance from the chosen backend based on
+    the given on-disk ``.dist-info`` directory.
+    """
+    from .pkg_resources import Distribution
+
+    return Distribution.from_directory(directory)
+
+
 def get_wheel_distribution(wheel: Wheel, canonical_name: str) -> BaseDistribution:
     """Get the representation of the specified wheel's distribution metadata.
 

src/pip/_internal/metadata/base.py

@@ -1,6 +1,8 @@
+import csv
 import email.message
 import json
 import logging
+import pathlib
 import re
 import zipfile
 from typing import (
@@ -12,6 +14,7 @@
     Iterator,
     List,
     Optional,
+    Tuple,
     Union,
 )
 
@@ -36,6 +39,8 @@
 
 DistributionVersion = Union[LegacyVersion, Version]
 
+InfoPath = Union[str, pathlib.PurePosixPath]
+
 logger = logging.getLogger(__name__)
 
 
@@ -53,6 +58,36 @@ def group(self) -> str:
         raise NotImplementedError()
 
 
+def _convert_installed_files_path(
+    entry: Tuple[str, ...],
+    info: Tuple[str, ...],
+) -> str:
+    """Convert a legacy installed-files.txt path into modern RECORD path.
+
+    The legacy format stores paths relative to the info directory, while the
+    modern format stores paths relative to the package root, e.g. the
+    site-packages directory.
+
+    :param entry: Path parts of the installed-files.txt entry.
+    :param info: Path parts of the egg-info directory relative to package root.
+    :returns: The converted entry.
+
+    For best compatibility with symlinks, this does not use ``abspath()`` or
+    ``Path.resolve()``, but tries to work with path parts:
+
+    1. While ``entry`` starts with ``..``, remove the equal amounts of parts
+       from ``info``; if ``info`` is empty, start appending ``..`` instead.
+    2. Join the two directly.
+    """
+    while entry and entry[0] == "..":
+        if not info or info[-1] == "..":
+            info += ("..",)
+        else:
+            info = info[:-1]
+        entry = entry[1:]
+    return str(pathlib.Path(*info, *entry))
+
+
 class BaseDistribution(Protocol):
     def __repr__(self) -> str:
         return f"{self.raw_name} {self.version} ({self.location})"
@@ -97,8 +132,8 @@ def editable_project_location(self) -> Optional[str]:
         return None
 
     @property
-    def info_directory(self) -> Optional[str]:
-        """Location of the .[egg|dist]-info directory.
+    def info_location(self) -> Optional[str]:
+        """Location of the .[egg|dist]-info directory or file.
 
         Similarly to ``location``, a string value is not necessarily a
         filesystem path. ``None`` means the distribution is created in-memory.
@@ -112,6 +147,65 @@ def info_directory(self) -> Optional[str]:
         """
         raise NotImplementedError()
 
+    @property
+    def installed_by_distutils(self) -> bool:
+        """Whether this distribution is installed with legacy distutils format.
+
+        A distribution installed with "raw" distutils not patched by setuptools
+        uses one single file at ``info_location`` to store metadata. We need to
+        treat this specially on uninstallation.
+        """
+        info_location = self.info_location
+        if not info_location:
+            return False
+        return pathlib.Path(info_location).is_file()
+
+    @property
+    def installed_as_egg(self) -> bool:
+        """Whether this distribution is installed as an egg.
+
+        This usually indicates the distribution was installed by (older versions
+        of) easy_install.
+        """
+        location = self.location
+        if not location:
+            return False
+        return location.endswith(".egg")
+
+    @property
+    def installed_with_setuptools_egg_info(self) -> bool:
+        """Whether this distribution is installed with the ``.egg-info`` format.
+
+        This usually indicates the distribution was installed with setuptools
+        with an old pip version or with ``single-version-externally-managed``.
+
+        Note that this ensure the metadata store is a directory. distutils can
+        also installs an ``.egg-info``, but as a file, not a directory. This
+        property is *False* for that case. Also see ``installed_by_distutils``.
+        """
+        info_location = self.info_location
+        if not info_location:
+            return False
+        if not info_location.endswith(".egg-info"):
+            return False
+        return pathlib.Path(info_location).is_dir()
+
+    @property
+    def installed_with_dist_info(self) -> bool:
+        """Whether this distribution is installed with the "modern format".
+
+        This indicates a "modern" installation, e.g. storing metadata in the
+        ``.dist-info`` directory. This applies to installations made by
+        setuptools (but through pip, not directly), or anything using the
+        standardized build backend interface (PEP 517).
+        """
+        info_location = self.info_location
+        if not info_location:
+            return False
+        if not info_location.endswith(".dist-info"):
+            return False
+        return pathlib.Path(info_location).is_dir()
+
     @property
     def canonical_name(self) -> NormalizedName:
         raise NotImplementedError()
@@ -120,6 +214,14 @@ def canonical_name(self) -> NormalizedName:
     def version(self) -> DistributionVersion:
         raise NotImplementedError()
 
+    @property
+    def setuptools_filename(self) -> str:
+        """Convert a project name to its setuptools-compatible filename.
+
+        This is a copy of ``pkg_resources.to_filename()`` for compatibility.
+        """
+        return self.raw_name.replace("-", "_")
+
     @property
     def direct_url(self) -> Optional[DirectUrl]:
         """Obtain a DirectUrl from this distribution.
@@ -166,11 +268,24 @@ def in_usersite(self) -> bool:
     def in_site_packages(self) -> bool:
         raise NotImplementedError()
 
-    def read_text(self, name: str) -> str:
-        """Read a file in the .dist-info (or .egg-info) directory.
+    def is_file(self, path: InfoPath) -> bool:
+        """Check whether an entry in the info directory is a file."""
+        raise NotImplementedError()
+
+    def iterdir(self, path: InfoPath) -> Iterator[pathlib.PurePosixPath]:
+        """Iterate through a directory in the info directory.
+
+        Each item yielded would be a path relative to the info directory.
 
-        Should raise ``FileNotFoundError`` if ``name`` does not exist in the
-        metadata directory.
+        :raise FileNotFoundError: If ``name`` does not exist in the directory.
+        :raise NotADirectoryError: If ``name`` does not point to a directory.
+        """
+        raise NotImplementedError()
+
+    def read_text(self, path: InfoPath) -> str:
+        """Read a file in the info directory.
+
+        :raise FileNotFoundError: If ``name`` does not exist in the directory.
         """
         raise NotImplementedError()
 
@@ -229,6 +344,51 @@ def iter_provided_extras(self) -> Iterable[str]:
         """
         raise NotImplementedError()
 
+    def _iter_declared_entries_from_record(self) -> Optional[Iterator[str]]:
+        try:
+            text = self.read_text("RECORD")
+        except FileNotFoundError:
+            return None
+        # This extra Path-str cast normalizes entries.
+        return (str(pathlib.Path(row[0])) for row in csv.reader(text.splitlines()))
+
+    def _iter_declared_entries_from_legacy(self) -> Optional[Iterator[str]]:
+        try:
+            text = self.read_text("installed-files.txt")
+        except FileNotFoundError:
+            return None
+        paths = (p for p in text.splitlines(keepends=False) if p)
+        root = self.location
+        info = self.info_location
+        if root is None or info is None:
+            return paths
+        try:
+            info_rel = pathlib.Path(info).relative_to(root)
+        except ValueError:  # info is not relative to root.
+            return paths
+        if not info_rel.parts:  # info *is* root.
+            return paths
+        return (
+            _convert_installed_files_path(pathlib.Path(p).parts, info_rel.parts)
+            for p in paths
+        )
+
+    def iter_declared_entries(self) -> Optional[Iterator[str]]:
+        """Iterate through file entires declared in this distribution.
+
+        For modern .dist-info distributions, this is the files listed in the
+        ``RECORD`` metadata file. For legacy setuptools distributions, this
+        comes from ``installed-files.txt``, with entries normalized to be
+        compatible with the format used by ``RECORD``.
+
+        :return: An iterator for listed entries, or None if the distribution
+            contains neither ``RECORD`` nor ``installed-files.txt``.
+        """
+        return (
+            self._iter_declared_entries_from_record()
+            or self._iter_declared_entries_from_legacy()
+        )
+
 
 class BaseEnvironment:
     """An environment containing distributions to introspect."""