feat: read build-backend ignore lists

Signed-off-by: Henry Schreiner <[email protected]>

Author: henryiii

README.md

@@ -99,6 +99,7 @@ git-only = []
 default-ignore = true
 recurse-submodules = true
 mode = "git"
+build-backend = "auto"
 ```
 
 You can add `.gitignore` style lines here, and you can turn off the default
@@ -112,6 +113,12 @@ You can also select `mode = "all"`, which will instead check every file on your
 system. Be prepared to ignore lots of things manually, like `*.pyc` files, if
 you use this.
 
+You can tell check-sdist to look for exclude lists for a specific build backend
+with `build-backend`, or `"none"` to only use it's own exclude list. Build
+backends supported are `"flit_core.buildapi"`, `"hatchling.build"`, and
+`"scikit_build_core.build"`. The default, `"auto"`, will try to detect the build
+backend if `build-system.build-backend` is set to a known value.
+
 ### See also
 
 - [check-manifest](https://github.com/mgedmin/check-manifest): A (currently)

src/check_sdist/__main__.py

@@ -10,6 +10,7 @@
 
 from . import __version__
 from ._compat import tomllib
+from .backends import backend_ignored_patterns
 from .git import git_files
 from .inject import inject_junk_files
 from .resources import resources
@@ -54,6 +55,7 @@ def compare(
 
     installer = select_installer(installer)
 
+    pyproject = {}
     config = {}
     pyproject_toml = source_dir.joinpath("pyproject.toml")
     with contextlib.suppress(FileNotFoundError), pyproject_toml.open("rb") as f:
@@ -65,6 +67,7 @@ def compare(
     default_ignore = config.get("default-ignore", True)
     recurse_submodules = config.get("recurse-submodules", True)
     mode = config.get("mode", "git")
+    backend = config.get("build-backend", "auto")
 
     sdist = sdist_files(source_dir, isolated=isolated, installer=installer) - {
         "PKG-INFO"
@@ -90,6 +93,8 @@ def compare(
     sdist_only = frozenset(p for p in sdist - git if not sdist_spec.match_file(p))
     git_only = frozenset(p for p in git - sdist if not git_spec.match_file(p))
 
+    git_only = backend_ignored_patterns(backend, pyproject, git_only)
+
     if verbose:
         print("SDist contents:")
         print(*(f"  {x}" for x in sorted(sdist)), sep="\n")

src/check_sdist/backends.py

@@ -0,0 +1,64 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+import pathspec
+
+
+def backend_ignored_patterns(
+    backend: str, pyproject: dict[str, Any], files: frozenset[str]
+) -> frozenset[str]:
+    """
+    Return the ignored patterns for the given backend. If the generator is
+    "none", then no patterns are ignored. If the generator is "auto", then the
+    value is read from the pyproject.toml file. Any recognized backend can also
+    be specified directly.
+    """
+
+    if backend == "none":
+        return files
+
+    backend_resolved = (
+        pyproject.get("build-system", {}).get(
+            "build-backend", "setuptools.build_meta.__legacy__"
+        )
+        if backend == "auto"
+        else backend
+    )
+
+    if backend_resolved == "flit_core.buildapi":
+        flit_core_exclude = (
+            pyproject.get("tool", {})
+            .get("flit", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        for pattern in flit_core_exclude:
+            results = {str(p) for p in Path().glob(pattern)}
+            files = files - results
+        return files
+    if backend_resolved == "hatchling.build":
+        hatch_exclude = (
+            pyproject.get("tool", {})
+            .get("hatch", {})
+            .get("build", {})
+            .get("targets", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        sdist_spec = pathspec.GitIgnoreSpec.from_lines(hatch_exclude)
+        return frozenset(p for p in files if not sdist_spec.match_file(p))
+    if backend_resolved == "scikit_build_core.build":
+        hatch_exclude = (
+            pyproject.get("tool", {})
+            .get("scikit-build", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        sdist_spec = pathspec.GitIgnoreSpec.from_lines(hatch_exclude)
+        return frozenset(p for p in files if not sdist_spec.match_file(p))
+    if backend != "auto":
+        msg = f"Unknown backend: {backend} - please add support in check_dist.backends"
+        raise ValueError(msg)
+    return files

src/check_sdist/resources/check-sdist.schema.json

@@ -33,6 +33,17 @@
       "description": "What to use as baseline",
       "default": "git",
       "enum": ["git", "all"]
+    },
+    "build-backend": {
+      "description": "What to expect as build-backend, in order to look for exclude lists",
+      "default": "auto",
+      "enum": [
+        "auto",
+        "none",
+        "flit_core.buildapi",
+        "hatchling.build",
+        "scikit_build_core.build"
+      ]
     }
   }
 }

tests/test_package.py

@@ -1,8 +1,12 @@
 from __future__ import annotations
 
+import inspect
 import os
+import subprocess
 from pathlib import Path
 
+import pytest
+
 from check_sdist.__main__ import compare
 from check_sdist.inject import inject_junk_files
 
@@ -27,3 +31,92 @@ def test_self_dir_injected():
         assert compare(DIR.parent, isolated=True) == 0
     end = get_all_files(DIR.parent)
     assert start == end
+
+
+@pytest.fixture
+def git_dir(tmp_path: Path, monkeypatch: pytest.MonkeyPatch):
+    monkeypatch.chdir(tmp_path)
+    subprocess.run(["git", "init"], check=True)
+    subprocess.run(["git", "config", "user.email", "[email protected]"], check=True)
+    subprocess.run(["git", "config", "user.name", "CI"], check=True)
+    return tmp_path
+
+
+@pytest.mark.usefixtures("git_dir")
+@pytest.mark.parametrize("backend", ["auto", "none"])
+def test_hatchling(backend: str):
+    Path("pyproject.toml").write_text(
+        inspect.cleandoc(f"""
+            [build-system]
+            requires = ["hatchling"]
+            build-backend = "hatchling.build"
+
+            [project]
+            name = "hatchling-test"
+            version = "0.1.0"
+
+            [tool.hatch]
+            build.targets.sdist.exclude = ["ignore*", "some-file", "**/notme.txt"]
+
+            [tool.check-sdist]
+            build-backend = "{backend}"
+        """)
+    )
+    Path(".gitignore").write_text(
+        inspect.cleandoc("""
+        some-ignored-file.txt
+    """)
+    )
+    Path("ignore-me.txt").touch()
+    Path("not-ignored.txt").touch()
+    Path("some-file").touch()
+    Path("some-dir").mkdir()
+    Path("some-dir/notme.txt").touch()
+    subprocess.run(["git", "add", "."], check=True)
+    subprocess.run(["git", "commit", "-m", "Initial commit"], check=True)
+    Path("some-ignored-file.txt").touch()
+    assert compare(Path(), isolated=True, verbose=True) == (
+        0 if backend == "auto" else 2
+    )
+
+
+@pytest.mark.usefixtures("git_dir")
+@pytest.mark.parametrize("backend", ["auto", "none"])
+def test_flit_core(backend: str):
+    Path("pyproject.toml").write_text(
+        inspect.cleandoc(f"""
+            [build-system]
+            requires = ["flit-core"]
+            build-backend = "flit_core.buildapi"
+
+            [project]
+            name = "flit-core-test"
+            version = "0.1.0"
+            description = "A test package"
+
+            [tool.flit.sdist]
+            include = ["not-ignored.txt", ".gitignore"]
+            exclude = ["ignore*", "some-file", "**/notme.txt"]
+
+            [tool.check-sdist]
+            build-backend = "{backend}"
+        """)
+    )
+    Path(".gitignore").write_text(
+        inspect.cleandoc("""
+        some-ignored-file.txt
+    """)
+    )
+    Path("ignore-me.txt").touch()
+    Path("not-ignored.txt").touch()
+    Path("some-file").touch()
+    Path("some-dir").mkdir()
+    Path("some-dir/notme.txt").touch()
+    Path("flit_core_test").mkdir()
+    Path("flit_core_test/__init__.py").touch()
+    subprocess.run(["git", "add", "."], check=True)
+    subprocess.run(["git", "commit", "-m", "Initial commit"], check=True)
+    Path("some-ignored-file.txt").touch()
+    assert compare(Path(), isolated=True, verbose=True) == (
+        0 if backend == "auto" else 2
+    )