Include tests in source distributions

dev-requirements.txt

@@ -5,6 +5,8 @@ pre-commit
 pytest
 pytest-cov
 psutil
+# Build distributions in the packaging regression tests
+flit_core >=3.2,<5
 # To be able to test tornado coroutines
 tornado
 # To be able to test behavior in jupyter-notebooks

pyproject.toml

@@ -34,6 +34,15 @@ classifiers = [
 [project.urls]
 Homepage = "https://github.com/cloudpipe/cloudpickle"
 
+[tool.flit.sdist]
+include = ["tests/", "dev-requirements.txt", "tox.ini"]
+exclude = [
+    "**/__pycache__",
+    "**/*.py[co]",
+    "tests/cloudpickle_testpkg/*.egg-info",
+    "tests/cloudpickle_testpkg/build",
+]
+
 [tool.black]
 line-length = 88
 target_version = ['py38', 'py39', 'py310', 'py311', 'py312', 'py313']

tests/test_packaging.py

@@ -0,0 +1,91 @@
+"""Check the distributions used by downstream package maintainers."""
+
+import shutil
+import tarfile
+import zipfile
+from pathlib import Path
+
+import pytest
+from flit_core import buildapi
+
+
+@pytest.fixture
+def distributions(tmp_path, monkeypatch):
+    project = Path(__file__).resolve().parents[1]
+    source = tmp_path / "source"
+    source.mkdir()
+    # Copy build inputs only, not virtual environments such as .tox or .venv.
+    for name in (
+        "pyproject.toml",
+        "README.md",
+        "LICENSE",
+        "dev-requirements.txt",
+        "tox.ini",
+    ):
+        shutil.copy2(project / name, source / name)
+    for name in ("cloudpickle", "tests"):
+        shutil.copytree(
+            project / name,
+            source / name,
+            ignore=shutil.ignore_patterns(
+                ".pytest_cache", "__pycache__", "*.py[co]", "*.egg-info", "build"
+            ),
+        )
+    # Installing the test helper and running tests must not pollute an sdist.
+    generated = [
+        "tests/__pycache__/cloudpickle_test.cpython-312.pyc",
+        "tests/testutils.pyc",
+        "tests/cloudpickle_testpkg/cloudpickle_testpkg.egg-info/PKG-INFO",
+        "tests/cloudpickle_testpkg/build/lib/_cloudpickle_testpkg/__init__.py",
+    ]
+    for name in generated:
+        path = source / name
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_bytes(b"generated file")
+
+    dist = tmp_path / "dist"
+    dist.mkdir()
+    monkeypatch.chdir(source)
+    sdist = dist / buildapi.build_sdist(str(dist))
+    with tarfile.open(sdist) as archive:
+        names = {name.split("/", 1)[1] for name in archive.getnames()}
+        # This archive was built locally from the test's own source snapshot.
+        options = {"filter": "data"} if hasattr(tarfile, "data_filter") else {}
+        archive.extractall(tmp_path / "unpacked", **options)
+    unpacked = next((tmp_path / "unpacked").iterdir())
+    monkeypatch.chdir(unpacked)
+    wheel = dist / buildapi.build_wheel(str(dist))
+    return source, names, generated, wheel
+
+
+def test_sdist_contains_test_suite(distributions):
+    _, names, generated, _ = distributions
+    required = {
+        "dev-requirements.txt",
+        "tox.ini",
+        "tests/__init__.py",
+        "tests/cloudpickle_test.py",
+        "tests/cloudpickle_file_test.py",
+        "tests/cloudpickle_ipykernel_test.py",
+        "tests/test_backward_compat.py",
+        "tests/test_packaging.py",
+        "tests/testutils.py",
+        "tests/mock_local_folder/subfolder/submod.py",
+        "tests/cloudpickle_testpkg/setup.py",
+        "tests/cloudpickle_testpkg/_cloudpickle_testpkg/mod.py",
+        "tests/old_pickles/cpython_311/simple_func.pkl",
+    }
+    assert required <= names, required - names
+    assert not set(generated) & names
+
+
+def test_wheel_from_sdist_contains_only_runtime_package(distributions):
+    source, _, _, wheel = distributions
+    with zipfile.ZipFile(wheel) as archive:
+        names = archive.namelist()
+        assert all(
+            name.startswith("cloudpickle/") or ".dist-info/" in name for name in names
+        )
+        for name in names:
+            if name.startswith("cloudpickle/"):
+                assert archive.read(name) == (source / name).read_bytes()