Compare update_lib contents and benchmark CPython parsing consistently (#9393)

Assisted-by: Codex:GPT-6

scripts/update_lib/file_utils.py

@@ -222,11 +222,29 @@ def get_cpython_dir(src_path: pathlib.Path) -> pathlib.Path:
 # === Comparison utilities ===
 
 
+def _files_are_same(left: pathlib.Path, right: pathlib.Path) -> bool:
+    # Recheck contents even when size and timestamps match, including after a
+    # previous comparison. filecmp caches results by those metadata alone.
+    with left.open("rb") as left_file, right.open("rb") as right_file:
+        while True:
+            chunk = left_file.read(65536)
+            if chunk != right_file.read(65536):
+                return False
+            if not chunk:
+                return True
+
+
 def _dircmp_is_same(dcmp: filecmp.dircmp) -> bool:
     """Recursively check if two directories are identical."""
-    if dcmp.diff_files or dcmp.left_only or dcmp.right_only:
+    if dcmp.left_only or dcmp.right_only or dcmp.common_funny:
         return False
 
+    for name in dcmp.common_files:
+        if not _files_are_same(
+            pathlib.Path(dcmp.left) / name, pathlib.Path(dcmp.right) / name
+        ):
+            return False
+
     for subdir in dcmp.subdirs.values():
         if not _dircmp_is_same(subdir):
             return False
@@ -236,14 +254,17 @@ def _dircmp_is_same(dcmp: filecmp.dircmp) -> bool:
 
 def compare_paths(cpython_path: pathlib.Path, local_path: pathlib.Path) -> bool:
     """Compare a CPython path with a local path (file or directory)."""
-    if not local_path.exists():
+    if not cpython_path.exists() or not local_path.exists():
         return False
 
-    if cpython_path.is_file():
-        return filecmp.cmp(cpython_path, local_path, shallow=False)
-
-    dcmp = filecmp.dircmp(cpython_path, local_path)
-    return _dircmp_is_same(dcmp)
+    try:
+        if cpython_path.is_file():
+            return local_path.is_file() and _files_are_same(cpython_path, local_path)
+        if not cpython_path.is_dir() or not local_path.is_dir():
+            return False
+        return _dircmp_is_same(filecmp.dircmp(cpython_path, local_path))
+    except OSError:
+        return False
 
 
 def compare_file_contents(

scripts/update_lib/tests/test_deps.py

@@ -325,6 +325,26 @@ import nonexistent
 class TestDircmpIsSame(unittest.TestCase):
     """Tests for _dircmp_is_same function."""
 
+    def test_matching_metadata(self):
+        import os
+
+        from update_lib.file_utils import compare_paths
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            left, right = (pathlib.Path(tmpdir) / name for name in ("left", "right"))
+            for directory in (left, right):
+                (directory / "nested").mkdir(parents=True)
+                (directory / "nested/module.py").write_bytes(b"x = 1\n")
+            source, destination = (p / "nested/module.py" for p in (left, right))
+            timestamp = source.stat().st_mtime_ns
+            os.utime(destination, ns=(timestamp, timestamp))
+            self.assertTrue(compare_paths(source, destination))
+            self.assertTrue(compare_paths(left, right))
+            destination.write_bytes(b"x = 2\n")
+            os.utime(destination, ns=(timestamp, timestamp))
+            self.assertFalse(compare_paths(source, destination))
+            self.assertFalse(compare_paths(left, right))
+
     def test_identical_directories(self):
         """Test that identical directories return True."""
         import filecmp