gh-93259: Validate arg to Distribution.from_name.

Syncs with importlib_metadata 4.12.0.

Author: jaraco

Doc/library/importlib.metadata.rst

@@ -17,8 +17,9 @@
 package metadata.  Built in part on Python's import system, this library
 intends to replace similar functionality in the `entry point
 API`_ and `metadata API`_ of ``pkg_resources``.  Along with
-:mod:`importlib.resources` (with new features backported to the
-`importlib_resources`_ package), this can eliminate the need to use the older
+:mod:`importlib.resources` (with new features backported to
+:doc:`importlib_resources <importlib_resources:index>`),
+this package can eliminate the need to use the older
 and less efficient
 ``pkg_resources`` package.
 
@@ -32,6 +33,13 @@ By default, package metadata can live on the file system or in zip archives on
 anywhere.
 
 
+.. seealso::
+
+   https://importlib-metadata.readthedocs.io/
+      The documentation for ``importlib_metadata``, which supplies a
+      backport of ``importlib.metadata``.
+
+
 Overview
 ========
 
@@ -54,9 +62,9 @@ You can get the version string for ``wheel`` by running the following:
     >>> version('wheel')  # doctest: +SKIP
     '0.32.3'
 
-You can also get the set of entry points keyed by group, such as
+You can also get a collection of entry points selectable by properties of the EntryPoint (typically 'group' or 'name'), such as
 ``console_scripts``, ``distutils.commands`` and others.  Each group contains a
-sequence of :ref:`EntryPoint <entry-points>` objects.
+collection of :ref:`EntryPoint <entry-points>` objects.
 
 You can get the :ref:`metadata for a distribution <metadata>`::
 
@@ -91,7 +99,7 @@ Query all entry points::
     >>> eps = entry_points()  # doctest: +SKIP
 
 The ``entry_points()`` function returns an ``EntryPoints`` object,
-a sequence of all ``EntryPoint`` objects with ``names`` and ``groups``
+a collection of all ``EntryPoint`` objects with ``names`` and ``groups``
 attributes for convenience::
 
     >>> sorted(eps.groups)  # doctest: +SKIP
@@ -174,6 +182,13 @@ all the metadata in a JSON-compatible form per :PEP:`566`::
     >>> wheel_metadata.json['requires_python']
     '>=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*'
 
+.. note::
+
+    The actual type of the object returned by ``metadata()`` is an
+    implementation detail and should be accessed only through the interface
+    described by the
+    `PackageMetadata protocol <https://importlib-metadata.readthedocs.io/en/latest/api.html#importlib_metadata.PackageMetadata>`.
+
 .. versionchanged:: 3.10
    The ``Description`` is now included in the metadata when presented
    through the payload. Line continuation characters have been removed.
@@ -295,6 +310,15 @@ The full set of available metadata is not described here.  See :pep:`566`
 for additional details.
 
 
+Distribution Discovery
+======================
+
+By default, this package provides built-in support for discovery of metadata for file system and zip file packages. This metadata finder search defaults to ``sys.path``, but varies slightly in how it interprets those values from how other import machinery does. In particular:
+
+- ``importlib.metadata`` does not honor :class:`bytes` objects on ``sys.path``.
+- ``importlib.metadata`` will incidentally honor :py:class:`pathlib.Path` objects on ``sys.path`` even though such values will be ignored for imports.
+
+
 Extending the search algorithm
 ==============================
 

Lib/importlib/metadata/__init__.py

@@ -543,21 +543,21 @@ def locate_file(self, path):
         """
 
     @classmethod
-    def from_name(cls, name):
+    def from_name(cls, name: str):
         """Return the Distribution for the given package name.
 
         :param name: The name of the distribution package to search for.
         :return: The Distribution instance (or subclass thereof) for the named
             package, if found.
         :raises PackageNotFoundError: When the named package's distribution
             metadata cannot be found.
+        :raises ValueError: When an invalid value is supplied for name.
         """
-        for resolver in cls._discover_resolvers():
-            dists = resolver(DistributionFinder.Context(name=name))
-            dist = next(iter(dists), None)
-            if dist is not None:
-                return dist
-        else:
+        if not name:
+            raise ValueError("A distribution name is required.")
+        try:
+            return next(cls.discover(name=name))
+        except StopIteration:
             raise PackageNotFoundError(name)
 
     @classmethod
@@ -945,13 +945,26 @@ def _normalized_name(self):
         normalized name from the file system path.
         """
         stem = os.path.basename(str(self._path))
-        return self._name_from_stem(stem) or super()._normalized_name
+        return (
+            pass_none(Prepared.normalize)(self._name_from_stem(stem))
+            or super()._normalized_name
+        )
 
-    def _name_from_stem(self, stem):
-        name, ext = os.path.splitext(stem)
+    @staticmethod
+    def _name_from_stem(stem):
+        """
+        >>> PathDistribution._name_from_stem('foo-3.0.egg-info')
+        'foo'
+        >>> PathDistribution._name_from_stem('CherryPy-3.0.dist-info')
+        'CherryPy'
+        >>> PathDistribution._name_from_stem('face.egg-info')
+        'face'
+        >>> PathDistribution._name_from_stem('foo.bar')
+        """
+        filename, ext = os.path.splitext(stem)
         if ext not in ('.dist-info', '.egg-info'):
             return
-        name, sep, rest = stem.partition('-')
+        name, sep, rest = filename.partition('-')
         return name
 
 
@@ -991,6 +1004,15 @@ def version(distribution_name):
     return distribution(distribution_name).version
 
 
+_unique = functools.partial(
+    unique_everseen,
+    key=operator.attrgetter('_normalized_name'),
+)
+"""
+Wrapper for ``distributions`` to return unique distributions by name.
+"""
+
+
 def entry_points(**params) -> Union[EntryPoints, SelectableGroups]:
     """Return EntryPoint objects for all installed packages.
 
@@ -1008,10 +1030,8 @@ def entry_points(**params) -> Union[EntryPoints, SelectableGroups]:
 
     :return: EntryPoints or SelectableGroups for all installed packages.
     """
-    norm_name = operator.attrgetter('_normalized_name')
-    unique = functools.partial(unique_everseen, key=norm_name)
     eps = itertools.chain.from_iterable(
-        dist.entry_points for dist in unique(distributions())
+        dist.entry_points for dist in _unique(distributions())
     )
     return SelectableGroups.load(eps).select(**params)
 

Lib/test/test_importlib/fixtures.py

@@ -5,6 +5,7 @@
 import pathlib
 import tempfile
 import textwrap
+import functools
 import contextlib
 
 from test.support.os_helper import FS_NONASCII
@@ -296,3 +297,18 @@ def setUp(self):
         # Add self.zip_name to the front of sys.path.
         self.resources = contextlib.ExitStack()
         self.addCleanup(self.resources.close)
+
+
+def parameterize(*args_set):
+    """Run test method with a series of parameters."""
+
+    def wrapper(func):
+        @functools.wraps(func)
+        def _inner(self):
+            for args in args_set:
+                with self.subTest(**args):
+                    func(self, **args)
+
+        return _inner
+
+    return wrapper

Lib/test/test_importlib/test_main.py

@@ -1,7 +1,6 @@
 import re
 import json
 import pickle
-import textwrap
 import unittest
 import warnings
 import importlib.metadata
@@ -16,6 +15,7 @@
     Distribution,
     EntryPoint,
     PackageNotFoundError,
+    _unique,
     distributions,
     entry_points,
     metadata,
@@ -51,6 +51,14 @@ def test_package_not_found_mentions_metadata(self):
     def test_new_style_classes(self):
         self.assertIsInstance(Distribution, type)
 
+    @fixtures.parameterize(
+        dict(name=None),
+        dict(name=''),
+    )
+    def test_invalid_inputs_to_from_name(self, name):
+        with self.assertRaises(Exception):
+            Distribution.from_name(name)
+
 
 class ImportTests(fixtures.DistInfoPkg, unittest.TestCase):
     def test_import_nonexistent_module(self):
@@ -78,48 +86,50 @@ def test_resolve_without_attr(self):
 
 class NameNormalizationTests(fixtures.OnSysPath, fixtures.SiteDir, unittest.TestCase):
     @staticmethod
-    def pkg_with_dashes(site_dir):
+    def make_pkg(name):
         """
-        Create minimal metadata for a package with dashes
-        in the name (and thus underscores in the filename).
+        Create minimal metadata for a dist-info package with
+        the indicated name on the file system.
         """
-        metadata_dir = site_dir / 'my_pkg.dist-info'
-        metadata_dir.mkdir()
-        metadata = metadata_dir / 'METADATA'
-        with metadata.open('w', encoding='utf-8') as strm:
-            strm.write('Version: 1.0\n')
-        return 'my-pkg'
+        return {
+            f'{name}.dist-info': {
+                'METADATA': 'VERSION: 1.0\n',
+            },
+        }
 
     def test_dashes_in_dist_name_found_as_underscores(self):
         """
         For a package with a dash in the name, the dist-info metadata
         uses underscores in the name. Ensure the metadata loads.
         """
-        pkg_name = self.pkg_with_dashes(self.site_dir)
-        assert version(pkg_name) == '1.0'
-
-    @staticmethod
-    def pkg_with_mixed_case(site_dir):
-        """
-        Create minimal metadata for a package with mixed case
-        in the name.
-        """
-        metadata_dir = site_dir / 'CherryPy.dist-info'
-        metadata_dir.mkdir()
-        metadata = metadata_dir / 'METADATA'
-        with metadata.open('w', encoding='utf-8') as strm:
-            strm.write('Version: 1.0\n')
-        return 'CherryPy'
+        fixtures.build_files(self.make_pkg('my_pkg'), self.site_dir)
+        assert version('my-pkg') == '1.0'
 
     def test_dist_name_found_as_any_case(self):
         """
         Ensure the metadata loads when queried with any case.
         """
-        pkg_name = self.pkg_with_mixed_case(self.site_dir)
+        pkg_name = 'CherryPy'
+        fixtures.build_files(self.make_pkg(pkg_name), self.site_dir)
         assert version(pkg_name) == '1.0'
         assert version(pkg_name.lower()) == '1.0'
         assert version(pkg_name.upper()) == '1.0'
 
+    def test_unique_distributions(self):
+        """
+        Two distributions varying only by non-normalized name on
+        the file system should resolve as the same.
+        """
+        fixtures.build_files(self.make_pkg('abc'), self.site_dir)
+        before = list(_unique(distributions()))
+
+        alt_site_dir = self.fixtures.enter_context(fixtures.tempdir())
+        self.fixtures.enter_context(self.add_sys_path(alt_site_dir))
+        fixtures.build_files(self.make_pkg('ABC'), alt_site_dir)
+        after = list(_unique(distributions()))
+
+        assert len(after) == len(before)
+
 
 class NonASCIITests(fixtures.OnSysPath, fixtures.SiteDir, unittest.TestCase):
     @staticmethod
@@ -128,11 +138,12 @@ def pkg_with_non_ascii_description(site_dir):
         Create minimal metadata for a package with non-ASCII in
         the description.
         """
-        metadata_dir = site_dir / 'portend.dist-info'
-        metadata_dir.mkdir()
-        metadata = metadata_dir / 'METADATA'
-        with metadata.open('w', encoding='utf-8') as fp:
-            fp.write('Description: pôrˈtend')
+        contents = {
+            'portend.dist-info': {
+                'METADATA': 'Description: pôrˈtend',
+            },
+        }
+        fixtures.build_files(contents, site_dir)
         return 'portend'
 
     @staticmethod
@@ -141,19 +152,15 @@ def pkg_with_non_ascii_description_egg_info(site_dir):
         Create minimal metadata for an egg-info package with
         non-ASCII in the description.
         """
-        metadata_dir = site_dir / 'portend.dist-info'
-        metadata_dir.mkdir()
-        metadata = metadata_dir / 'METADATA'
-        with metadata.open('w', encoding='utf-8') as fp:
-            fp.write(
-                textwrap.dedent(
-                    """
+        contents = {
+            'portend.dist-info': {
+                'METADATA': """
                 Name: portend
 
-                pôrˈtend
-                """
-                ).strip()
-            )
+                pôrˈtend""",
+            },
+        }
+        fixtures.build_files(contents, site_dir)
         return 'portend'
 
     def test_metadata_loads(self):

Lib/test/test_importlib/test_metadata_api.py

@@ -89,15 +89,15 @@ def test_entry_points_distribution(self):
             self.assertIn(ep.dist.name, ('distinfo-pkg', 'egginfo-pkg'))
             self.assertEqual(ep.dist.version, "1.0.0")
 
-    def test_entry_points_unique_packages(self):
+    def test_entry_points_unique_packages_normalized(self):
         """
         Entry points should only be exposed for the first package
-        on sys.path with a given name.
+        on sys.path with a given name (even when normalized).
         """
         alt_site_dir = self.fixtures.enter_context(fixtures.tempdir())
         self.fixtures.enter_context(self.add_sys_path(alt_site_dir))
         alt_pkg = {
-            "distinfo_pkg-1.1.0.dist-info": {
+            "DistInfo_pkg-1.1.0.dist-info": {
                 "METADATA": """
                 Name: distinfo-pkg
                 Version: 1.1.0

Misc/NEWS.d/next/Library/2022-06-25-13-38-53.gh-issue-93259.FAGw-2.rst

@@ -0,0 +1,2 @@
+Now raise ``ValueError`` when ``None`` or an empty string are passed to
+``Distribution.from_name`` (and other callers).