feat: read build-backend ignore lists (#66)

* feat: read build-backend ignore lists Signed-off-by: Henry Schreiner <[email protected]> * feat: add some more backends Signed-off-by: Henry Schreiner <[email protected]> * fix: pull out logic Signed-off-by: Henry Schreiner <[email protected]> * fix: adding maturin test Signed-off-by: Henry Schreiner <[email protected]> * tests: add some extra ignored files Signed-off-by: Henry Schreiner <[email protected]> * tests: test for unknown backend Signed-off-by: Henry Schreiner <[email protected]> * chore: ignore coverage on return __all__ Signed-off-by: Henry Schreiner <[email protected]> --------- Signed-off-by: Henry Schreiner <[email protected]>
henryiii · Oct 14, 2024 · 2b869e1 · 2b869e1
1 parent 08067d2
commit 2b869e1
Show file tree

Hide file tree

Showing 6 changed files with 366 additions and 0 deletions.
diff --git a/README.md b/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 +113,13 @@ 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"`,
+`"scikit_build_core.build"`, `"pdm.backend"`, `"maturin"`, and
+`"poetry.core.masonry.api"`. 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)

diff --git a/pyproject.toml b/pyproject.toml
@@ -84,6 +84,9 @@ testpaths = [
 
 [tool.coverage]
 run.source = ["check_sdist"]
+report.exclude_also = [
+    "return __all__",
+]
 
 
 [tool.mypy]

diff --git a/src/check_sdist/__main__.py b/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")

diff --git a/src/check_sdist/backends.py b/src/check_sdist/backends.py
@@ -0,0 +1,102 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+import pathspec
+
+__all__ = ["backend_ignored_patterns"]
+
+
+def __dir__() -> list[str]:
+    return __all__
+
+
+def glob_filter(patterns: list[str], files: frozenset[str]) -> frozenset[str]:
+    """
+    Filter out files based on glob patterns.
+    """
+    new_files = set(files)
+    for pattern in patterns:
+        results = {str(p.as_posix()) for p in Path().glob(pattern)}
+        new_files -= results
+    return frozenset(new_files)
+
+
+def pathspec_filter(patterns: list[str], files: frozenset[str]) -> frozenset[str]:
+    """
+    Filter out files based on gitignore-style patterns.
+    """
+    spec = pathspec.GitIgnoreSpec.from_lines(patterns)
+    return frozenset(p for p in files if not spec.match_file(p))
+
+
+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":
+        exclude = (
+            pyproject.get("tool", {})
+            .get("flit", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        return glob_filter(exclude, files)
+    if backend_resolved == "hatchling.build":
+        exclude = (
+            pyproject.get("tool", {})
+            .get("hatch", {})
+            .get("build", {})
+            .get("targets", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        return pathspec_filter(exclude, files)
+    if backend_resolved == "scikit_build_core.build":
+        exclude = (
+            pyproject.get("tool", {})
+            .get("scikit-build", {})
+            .get("sdist", {})
+            .get("exclude", [])
+        )
+        return pathspec_filter(exclude, files)
+    if backend_resolved == "pdm.backend":
+        exclude = (
+            pyproject.get("tool", {})
+            .get("pdm", {})
+            .get("build", {})
+            .get("excludes", [])
+        )
+        return glob_filter(exclude, files)
+    if backend_resolved == "poetry.core.masonry.api":
+        exclude = [
+            x if isinstance(x, str) else x["path"]
+            for x in pyproject.get("tool", {}).get("poetry", {}).get("exclude", [])
+            if isinstance(x, str) or "sdist" in x.get("format", ["sdist"])
+        ]
+        return glob_filter(exclude, files)
+    if backend_resolved == "maturin":
+        exclude = pyproject.get("tool", {}).get("maturin", {}).get("exclude", [])
+        return glob_filter(exclude, files)
+    if backend != "auto":
+        msg = f"Unknown backend: {backend} - please add support in check_dist.backends"
+        raise ValueError(msg)
+    return files
diff --git a/src/check_sdist/resources/check-sdist.schema.json b/src/check_sdist/resources/check-sdist.schema.json
@@ -33,6 +33,20 @@
       "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",
+        "pdm.backend.api",
+        "poetry.core.masonry.api",
+        "maturin"
+      ]
     }
   }
 }