Merge branch 'auto-discovery' into experimental/support-pyproject (#2894)

Adds automatic discovery of `packages`, `py_modules` and `name` for
`flat-` and `src-` layouts.

Author: abravalheri

changelog.d/2887.change.1.rst

@@ -0,0 +1,20 @@
+Added automatic discovery for ``py_modules`` and ``packages``
+-- by :user:`abravalheri`.
+
+Setuptools will try to find these values assuming that the package uses either
+the *src-layout* (a ``src`` directory containing all the packages or modules),
+the *flat-layout* (package directories directly under the project root),
+or the *single-module* approach (isolated Python files, directly under
+the project root).
+
+The automatic discovery will also respect layouts that are explicitly
+configured using the ``package_dir`` option.
+
+For backward-compatibility, this behavior will be observed **only if both**
+``py_modules`` **and** ``packages`` **are not set**.
+
+If setuptools detects modules or packages that are not supposed to be in the
+distribution, please manually set ``py_modules`` and ``packages`` in your
+``setup.cfg`` or ``setup.py`` file.
+If you are using a *flat-layout*, you can also consider switching to
+*src-layout*.

changelog.d/2887.change.2.rst

@@ -0,0 +1,9 @@
+Added automatic configuration for the ``name`` metadata
+-- by :user:`abravalheri`.
+
+Setuptools will adopt the name of the top-level package (or module in the case
+of single-module distributions), **only when** ``name`` **is not explicitly
+provided**.
+
+Please note that it is not possible to automatically derive a single name when
+the distribution consists of multiple top-level packages or modules.

changelog.d/2894.breaking.rst

@@ -0,0 +1,10 @@
+If you purposefully want to create an *"empty distribution"*, please be aware
+that some Python files (or general folders) might be automatically detected and
+included.
+
+Projects that currently don't specify both ``packages`` and ``py_modules`` in their
+configuration and have extra Python files and folders (not meant for distribution),
+might see these files being included in the wheel archive.
+
+You can check details about the automatic discovery behaviour (and
+how to configure a different one) in :doc:`/userguide/package_discovery`.

docs/userguide/package_discovery.rst

@@ -38,8 +38,142 @@ included manually in the following manner:
             packages=['mypkg1', 'mypkg2']
         )
 
-This can get tiresome really quickly. To speed things up, we introduce two
-functions provided by setuptools:
+This can get tiresome really quickly. To speed things up, you can rely on
+setuptools automatic discovery, or use the provided tools, as explained in
+the following sections.
+
+
+Automatic discovery
+===================
+
+By default setuptools will consider 2 popular project layouts, each one with
+its own set of advantages and disadvantages [#layout1]_ [#layout2]_.
+
+src-layout:
+    The project should contain a ``src`` directory under the project root and
+    all modules and packages meant for distribution are placed inside this
+    directory::
+
+        project_root_directory
+        ├── pyproject.toml
+        ├── setup.cfg  # or setup.py
+        ├── ...
+        └── src/
+            └── mypkg/
+                ├── __init__.py
+                ├── ...
+                └── mymodule.py
+
+    This layout is very handy when you wish to use automatic discovery,
+    since you don't have to worry about other Python files or folders in your
+    project root being distributed by mistake. In some circumstances it can be
+    also less error-prone for testing or when using :pep:`420`-style packages.
+    On the other hand you cannot rely on the implicit ``PYTHONPATH=.`` to fire
+    up the Python REPL and play with your package (you will need an
+    `editable install`_ to be able to do that).
+
+flat-layout (also known as "adhoc"):
+    The package folder(s) are placed directly under the project root::
+
+        project_root_directory
+        ├── pyproject.toml
+        ├── setup.cfg  # or setup.py
+        ├── ...
+        └── mypkg/
+            ├── __init__.py
+            ├── ...
+            └── mymodule.py
+
+    This layout is very practical for using the REPL, but in some situations
+    it can be can be more error-prone (e.g. during tests or if you have a bunch
+    of folders or Python files hanging around your project root)
+
+There is also a handy variation of the *flat-layout* for utilities/libraries
+that can be implemented with a single Python file:
+
+single-module approach (or "few top-level modules"):
+    Standalone modules are placed directly under the project root, instead of
+    inside a package folder::
+
+        project_root_directory
+        ├── pyproject.toml
+        ├── setup.cfg  # or setup.py
+        ├── ...
+        └── single_file_lib.py
+
+Setuptools will automatically scan your project directory looking for these
+layouts and try to guess the correct values for the :ref:`packages <declarative
+config>` and :doc:`py_modules </references/keywords>` configuration.
+
+To avoid confusion, file and folder names that are used by popular tools (or
+that correspond to well-known conventions, such as distributing documentation
+alongside the project code) are automatically filtered out in the case of
+*flat-layouts*:
+
+.. autoattribute:: setuptools.discovery.FlatLayoutPackageFinder.DEFAULT_EXCLUDE
+
+.. autoattribute:: setuptools.discovery.FlatLayoutModuleFinder.DEFAULT_EXCLUDE
+
+Also note that you can customise your project layout by explicitly setting
+``package_dir``:
+
+.. tab:: setup.cfg
+
+    .. code-block:: ini
+
+        [options]
+        # ...
+        package_dir =
+            = lib
+            # similar to "src-layout" but using the "lib" folder
+            # pkg.mod corresponds to lib/pkg/mod.py
+        # OR
+        package_dir =
+            pkg1 = lib1
+            # pkg1.mod corresponds to lib1/mod.py
+            # pkg1.subpkg.mod corresponds to lib1/subpkg/mod.py
+            pkg2 = lib2
+            # pkg2.mod corresponds to lib2/mod.py
+            pkg2.subpkg = lib3
+            # pkg2.subpkg.mod corresponds to lib3/mod.py
+
+.. tab:: setup.py
+
+    .. code-block:: python
+
+        setup(
+            # ...
+            package_dir = {"": "lib"}
+            # similar to "src-layout" but using the "lib" folder
+            # pkg.mod corresponds to lib/pkg/mod.py
+        )
+
+        # OR
+
+        setup(
+            # ...
+            package_dir = {
+                "pkg1": "lib1",  # pkg1.mod corresponds to lib1/mod.py
+                                 # pkg1.subpkg.mod corresponds to lib1/subpkg/mod.py
+                "pkg2": "lib2",   # pkg2.mod corresponds to lib2/mod.py
+                "pkg2.subpkg": "lib3"  # pkg2.subpkg.mod corresponds to lib3/mod.py
+                # ...
+        )
+
+.. important:: Automatic discovery will **only** be enabled if you don't
+   provide any configuration for both ``packages`` and ``py_modules``.
+   If at least one of them is explicitly set, automatic discovery will not take
+   place.
+
+
+Custom discovery
+================
+
+If the automatic discovery does not work for you
+(e.g., you want to *include* in the distribution top-level packages with
+reserved names such as ``tasks``, ``example`` or ``docs``, or you want to
+*exclude* nested packages that would be otherwise included), you can use
+the provided tools for package discovery:
 
 .. tab:: setup.cfg
 
@@ -61,7 +195,7 @@ functions provided by setuptools:
 
 
 Using ``find:`` or ``find_packages``
-====================================
+------------------------------------
 Let's start with the first tool. ``find:`` (``find_packages``) takes a source
 directory and two lists of package name patterns to exclude and include, and
 then return a list of ``str`` representing the packages it could find. To use
@@ -113,7 +247,7 @@ in ``src`` that starts with the name ``pkg`` and not ``additional``:
 .. _Namespace Packages:
 
 Using ``find_namespace:`` or ``find_namespace_packages``
-========================================================
+--------------------------------------------------------
 ``setuptools``  provides the ``find_namespace:`` (``find_namespace_packages``)
 which behaves similarly to ``find:`` but works with namespace package. Before
 diving in, it is important to have a good understanding of what namespace
@@ -249,3 +383,9 @@ file contains the following:
     __path__ = __import__('pkgutil').extend_path(__path__, __name__)
 
 The project layout remains the same and ``setup.cfg`` remains the same.
+
+
+.. [#layout1] https://blog.ionelmc.ro/2014/05/25/python-packaging/#the-structure
+.. [#layout2] https://blog.ionelmc.ro/2017/09/25/rehashing-the-src-layout/
+
+.. _editable install: https://pip.pypa.io/en/stable/cli/pip_install/#editable-installs

setup.cfg

@@ -80,7 +80,6 @@ testing-integration =
 	build[virtualenv]
 	filelock>=3.4.0
 
-
 docs =
 	# upstream
 	sphinx

setuptools/__init__.py

@@ -1,6 +1,5 @@
 """Extensions to the 'distutils' for large or complex distributions"""
 
-from fnmatch import fnmatchcase
 import functools
 import os
 import re
@@ -9,14 +8,14 @@
 
 import distutils.core
 from distutils.errors import DistutilsOptionError
-from distutils.util import convert_path
 
 from ._deprecation_warning import SetuptoolsDeprecationWarning
 
 import setuptools.version
 from setuptools.extension import Extension
 from setuptools.dist import Distribution
 from setuptools.depends import Require
+from setuptools.discovery import PackageFinder, PEP420PackageFinder
 from . import monkey
 from . import logging
 
@@ -37,85 +36,6 @@
 bootstrap_install_from = None
 
 
-class PackageFinder:
-    """
-    Generate a list of all Python packages found within a directory
-    """
-
-    @classmethod
-    def find(cls, where='.', exclude=(), include=('*',)):
-        """Return a list all Python packages found within directory 'where'
-
-        'where' is the root directory which will be searched for packages.  It
-        should be supplied as a "cross-platform" (i.e. URL-style) path; it will
-        be converted to the appropriate local path syntax.
-
-        'exclude' is a sequence of package names to exclude; '*' can be used
-        as a wildcard in the names, such that 'foo.*' will exclude all
-        subpackages of 'foo' (but not 'foo' itself).
-
-        'include' is a sequence of package names to include.  If it's
-        specified, only the named packages will be included.  If it's not
-        specified, all found packages will be included.  'include' can contain
-        shell style wildcard patterns just like 'exclude'.
-        """
-
-        return list(
-            cls._find_packages_iter(
-                convert_path(where),
-                cls._build_filter('ez_setup', '*__pycache__', *exclude),
-                cls._build_filter(*include),
-            )
-        )
-
-    @classmethod
-    def _find_packages_iter(cls, where, exclude, include):
-        """
-        All the packages found in 'where' that pass the 'include' filter, but
-        not the 'exclude' filter.
-        """
-        for root, dirs, files in os.walk(where, followlinks=True):
-            # Copy dirs to iterate over it, then empty dirs.
-            all_dirs = dirs[:]
-            dirs[:] = []
-
-            for dir in all_dirs:
-                full_path = os.path.join(root, dir)
-                rel_path = os.path.relpath(full_path, where)
-                package = rel_path.replace(os.path.sep, '.')
-
-                # Skip directory trees that are not valid packages
-                if '.' in dir or not cls._looks_like_package(full_path):
-                    continue
-
-                # Should this package be included?
-                if include(package) and not exclude(package):
-                    yield package
-
-                # Keep searching subdirectories, as there may be more packages
-                # down there, even if the parent was excluded.
-                dirs.append(dir)
-
-    @staticmethod
-    def _looks_like_package(path):
-        """Does a directory look like a package?"""
-        return os.path.isfile(os.path.join(path, '__init__.py'))
-
-    @staticmethod
-    def _build_filter(*patterns):
-        """
-        Given a list of patterns, return a callable that will be true only if
-        the input matches at least one of the patterns.
-        """
-        return lambda name: any(fnmatchcase(name, pat=pat) for pat in patterns)
-
-
-class PEP420PackageFinder(PackageFinder):
-    @staticmethod
-    def _looks_like_package(path):
-        return True
-
-
 find_packages = PackageFinder.find
 find_namespace_packages = PEP420PackageFinder.find