add fuzzy matching to package_data sanity (#85103) (#85135)

* add fuzzy matching to package_data sanity

* relaxes exact directory matches for license files to allow setuptools > 72 to pass

* sanity

(cherry picked from commit 7e00053a30)

pyproject.toml

@@ -1,3 +1,3 @@
 [build-system]
-requires = ["setuptools >= 66.1.0, <= 76.0.0", "wheel == 0.45.1"]  # lower bound to support controller Python versions, upper bound for latest version tested at release
+requires = ["setuptools >= 66.1.0, <= 80.3.1", "wheel == 0.45.1"]  # lower bound to support controller Python versions, upper bound for latest version tested at release
 build-backend = "setuptools.build_meta"

test/sanity/code-smell/package-data.py

@@ -15,6 +15,9 @@ import zipfile
 
 from ansible.release import __version__
 
+fuzzy_match_basenames: set[str] = {'COPYING'}
+"""Account for PEP639 differences in the placement of some files."""
+
 
 def collect_sdist_files(complete_file_list: list[str]) -> list[str]:
     """Return a list of files which should be present in the sdist."""
@@ -135,14 +138,32 @@ def list_wheel(path: pathlib.Path) -> list[str]:
     return paths
 
 
+def filter_fuzzy_matches(missing: set[str], extra: set[str]) -> None:
+    """
+    Removes entries from `missing` and `extra` that share a common basename that also appears in `fuzzy_match_basenames`.
+    Accounts for variable placement of non-runtime files by different versions of setuptools.
+    """
+    if not (missing or extra):
+        return
+
+    # calculate a set of basenames that appear in both missing and extra that are also marked as possibly needing fuzzy matching
+    corresponding_fuzzy_basenames = {os.path.basename(p) for p in missing}.intersection(os.path.basename(p) for p in extra).intersection(fuzzy_match_basenames)
+
+    # filter successfully fuzzy-matched entries from missing and extra
+    missing.difference_update({p for p in missing if os.path.basename(p) in corresponding_fuzzy_basenames})
+    extra.difference_update({p for p in extra if os.path.basename(p) in corresponding_fuzzy_basenames})
+
+
 def check_files(source: str, expected: list[str], actual: list[str]) -> list[str]:
     """Verify the expected files exist and no extra files exist."""
-    missing = sorted(set(expected) - set(actual))
+    missing = set(expected) - set(actual)
-    extra = sorted(set(actual) - set(expected))
+    extra = set(actual) - set(expected)
+
+    filter_fuzzy_matches(missing, extra)
 
     errors = (
-        [f'{path}: missing from {source}' for path in missing] +
+        [f'{path}: missing from {source}' for path in sorted(missing)] +
-        [f'{path}: unexpected in {source}' for path in extra]
+        [f'{path}: unexpected in {source}' for path in sorted(extra)]
     )
 
     return errors