oops I branched from the wrong one

MANIFEST.in

@@ -0,0 +1,11 @@
+# Exclude entire directories
+prune .github/
+prune dev_scripts/
+prune docs/
+prune examples/
+prune tests/
+
+# Exclude individual files
+exclude .* \
+    ADMIN.md CONTRIBUTING.md SECURITY.md \
+    CITATION.cff pdm.lock tasks.py

pyproject.toml

@@ -5,6 +5,7 @@ requires = [
     # https://numpy.org/devdocs/dev/depending_on_numpy.html#build-time-dependency
     "numpy>=2.1.0",
     "setuptools>=65.0.0",
+    "setuptools-scm>=8",  # include all Git tracked files as package data file
 ]
 build-backend = "setuptools.build_meta"
 
@@ -122,37 +123,10 @@ feff_plot_cross_section = "pymatgen.cli.feff_plot_cross_section:main"
 feff_plot_dos = "pymatgen.cli.feff_plot_dos:main"
 get_environment = "pymatgen.cli.get_environment:main"
 
-[tool.setuptools]
-include-package-data = false
-
 [tool.setuptools.packages.find]
 where = ["src"]
 include = ["pymatgen", "pymatgen.*"]
 
-[tool.setuptools.package-data]
-"pymatgen.analysis" = ["*.csv", "*.json", "*.yaml"]
-"pymatgen.analysis.chemenv" = [
-    "coordination_environments/coordination_geometries_files/*.json",
-    "coordination_environments/coordination_geometries_files/*.txt",
-    "coordination_environments/strategy_files/ImprovedConfidenceCutoffDefaultParameters.json",
-]
-"pymatgen.analysis.structure_prediction" = ["*.yaml", "data/*.json"]
-"pymatgen.analysis.diffraction" = ["*.json"]
-"pymatgen.analysis.magnetism" = ["default_magmoms.yaml"]
-"pymatgen.analysis.solar" = ["am1.5G.dat"]
-"pymatgen.entries" = ["*.json.gz", "*.yaml", "data/*.json"]
-"pymatgen.core" = ["*.json"]
-"pymatgen" = ["py.typed"]
-"pymatgen.io.vasp" = ["*.json", "*.json.bz2", "*.json.gz", "*.yaml"]
-"pymatgen.io.feff" = ["*.yaml"]
-"pymatgen.io.cp2k" = ["*.yaml"]
-"pymatgen.io.lobster" = ["lobster_basis/*.yaml"]
-"pymatgen.command_line" = ["*"]
-"pymatgen.util" = ["*.json", "structures/*.json"]
-"pymatgen.vis" = ["*.yaml"]
-"pymatgen.io.lammps" = ["CoeffsDataType.yaml", "templates/*.template"]
-"pymatgen.symmetry" = ["*.json", "*.sqlite", "*.yaml"]
-
 [tool.pdm.dev-dependencies]
 lint = ["mypy>=1.10.0", "pre-commit>=3.7.1", "ruff>=0.4.9"]
 test = ["pytest-cov>=5.0.0", "pytest-split>=0.9.0", "pytest>=8.2.2"]

tests/test_pkg.py

@@ -1,35 +1,78 @@
+"""Check the content of source code and source distribution.
+
+Binary distribution (wheel) is checked in test workflow.
+"""
+
 from __future__ import annotations
 
 import os
+import subprocess
+import tarfile
 from glob import glob
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+from monty.tempfile import ScratchDir
+
+if TYPE_CHECKING:
+    from pymatgen.util.typing import PathLike
+
+SRC_DIR = Path(__file__).parent.parent
+
+
+class TestCheckDistribution:
+    def test_source_code(self):
+        """Directly check the source code in the working directory."""
+        src_txt_path = SRC_DIR / "src/pymatgen.egg-info/SOURCES.txt"
 
-import pytest
+        _check_src_txt_is_complete(SRC_DIR, src_txt_path)
 
-SRC_TXT_PATH = "src/pymatgen.egg-info/SOURCES.txt"
+    def test_source_distribution(self):
+        """Build the source distribution and verify its contents."""
 
+        with ScratchDir("."):
+            # Build the source distribution
+            subprocess.run(["python", "-m", "build", "--sdist", SRC_DIR, "--outdir", ".", "-C--quiet"], check=True)
 
-@pytest.mark.skipif(
-    not os.path.isfile(SRC_TXT_PATH),
-    reason=f"{SRC_TXT_PATH=} not found. Run `pip install .` to create",
-)
-def test_egg_sources_txt_is_complete():
-    """Check that all source and data files in pymatgen/ are listed in pymatgen.egg-info/SOURCES.txt."""
+            # Decompress sdist
+            sdist_file = next(Path(".").glob("*.tar.gz"))
+            sdist_dir = sdist_file.name.removesuffix(".tar.gz")
+            with tarfile.open(sdist_file, "r:gz") as tar:
+                # TODO: remove attr check after only 3.12+
+                if hasattr(tarfile, "data_filter"):
+                    tar.extractall("", filter="data")
+                else:
+                    tar.extractall("")  # noqa: S202
 
-    with open(SRC_TXT_PATH, encoding="utf-8") as file:
+            # Verify source distribution contents
+            src_txt_path = f"{sdist_dir}/src/pymatgen.egg-info/SOURCES.txt"
+            _check_src_txt_is_complete(src_dir=sdist_dir, src_txt_path=src_txt_path)
+
+
+def _check_src_txt_is_complete(src_dir: PathLike, src_txt_path: PathLike) -> None:
+    """Check that all source code and data files are listed in given SOURCES.txt.
+
+    Args:
+        src_dir (PathLike): Path to the source code directory.
+        src_txt_path (PathLike): Path to the "SOURCES.txt" file.
+    """
+    src_dir = Path(src_dir)
+    src_txt_path = Path(src_txt_path)
+
+    assert src_dir.is_dir(), f"{src_dir} is not a directory"
+    assert src_txt_path.is_file(), f"{src_txt_path} doesn't exist"
+
+    with open(src_txt_path, encoding="utf-8") as file:
         sources = file.read()
 
-    # check that all files listed in SOURCES.txt exist
+    # Check that all files listed in "SOURCES.txt" exist
     for src_file in sources.splitlines():
-        assert os.path.isfile(src_file), f"{src_file!r} does not exist!"
-
-    # check that all files in pymatgen/ are listed in SOURCES.txt
-    for ext in ("py", "json*", "yaml", "csv"):
-        for filepath in glob(f"pymatgen/**/*.{ext}", recursive=True):
-            unix_path = filepath.replace("\\", "/")
-            if unix_path.endswith("dao.py"):
-                continue
+        assert os.path.isfile(src_dir / src_file), f"{src_file!r} does not exist!"
+
+    # Check that all files in src/pymatgen/ are listed in SOURCES.txt
+    for ext in ("py", "json", "json.*", "yaml", "csv"):
+        for filepath in glob(f"{src_dir}/src/pymatgen/**/*.{ext}", recursive=True):
+            unix_path = os.path.relpath(filepath.replace("\\", "/"), start=src_dir)
+
             if unix_path not in sources:
-                raise ValueError(
-                    f"{unix_path} not found in {SRC_TXT_PATH}. check setup.py package_data for "
-                    "outdated inclusion rules."
-                )
+                raise ValueError(f"{unix_path} not found in {src_txt_path}, check package data config")