GH-113225: Speed up pathlib.Path.iterdir() and glob()

barneygale · barneygale · commit 461b73decf90 · 2023-12-17T04:51:48.000Z
Use `os.DirEntry.path` as the string representation of child paths, unless
the parent path is empty, in which case we use the entry `name`.
diff --git a/Lib/pathlib/__init__.py b/Lib/pathlib/__init__.py
@@ -272,11 +272,18 @@ def iterdir(self):
         The children are yielded in arbitrary order, and the
         special entries '.' and '..' are not included.
         """
-        return (self._make_child_relpath(name) for name in os.listdir(self))
+        with os.scandir(self) as entries:
+            return iter([self._make_child_entry(entry) for entry in entries])
 
     def _scandir(self):
         return os.scandir(self)
 
+    def _make_child_entry(self, entry):
+        # Transform an entry yielded from _scandir() into a path object.
+        path = self.with_segments(entry.path)
+        path._str = entry.name if str(self) == '.' else entry.path
+        return path
+
     def absolute(self):
         """Return an absolute version of this path
         No normalization or symlink resolution is performed.
diff --git a/Lib/pathlib/_abc.py b/Lib/pathlib/_abc.py
@@ -89,9 +89,8 @@ def _select_children(parent_paths, dir_only, follow_symlinks, match):
                             continue
                     except OSError:
                         continue
-                name = entry.name
-                if match(name):
-                    yield parent_path._make_child_relpath(name)
+                if match(entry.name):
+                    yield parent_path._make_child_entry(entry)
 
 
 def _select_recursive(parent_paths, dir_only, follow_symlinks):
@@ -114,12 +113,12 @@ def _select_recursive(parent_paths, dir_only, follow_symlinks):
                 for entry in entries:
                     try:
                         if entry.is_dir(follow_symlinks=follow_symlinks):
-                            paths.append(path._make_child_relpath(entry.name))
+                            paths.append(path._make_child_entry(entry))
                             continue
                     except OSError:
                         pass
                     if not dir_only:
-                        yield path._make_child_relpath(entry.name)
+                        yield path._make_child_entry(entry)
 
 
 def _select_unique(paths):
@@ -797,6 +796,10 @@ def _scandir(self):
         from contextlib import nullcontext
         return nullcontext(self.iterdir())
 
+    def _make_child_entry(self, entry):
+        # Transform an entry yielded from _scandir() into a path object.
+        return entry
+
     def _make_child_relpath(self, name):
         path_str = str(self)
         tail = self._tail
diff --git a/Misc/NEWS.d/next/Library/2023-12-17-04-43-57.gh-issue-113225.dhxhiZ.rst b/Misc/NEWS.d/next/Library/2023-12-17-04-43-57.gh-issue-113225.dhxhiZ.rst
@@ -0,0 +1,2 @@
+Speed up :meth:`pathlib.Path.iterdir` and :meth:`~pathlib.Path.glob` by
+using :attr:`os.DirEntry.path` where possible.

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	+Speed up :meth:`pathlib.Path.iterdir` and :meth:`~pathlib.Path.glob` by
	`2`	+using :attr:`os.DirEntry.path` where possible.