GH-119169: Implement pathlib.Path.walk() using os.walk()

For silly reasons, pathlib's generic implementation of `walk()` currently resides in `glob._Globber`. This commit moves it into `pathlib._abc.PathBase.walk()` where it really belongs, and makes `pathlib.Path.walk()` call though to `os.walk()`. Symlink handling is a little different between the two `walk()` implementations when `followlinks=False`. In `pathlib` it means never following symlinks, not even for distinguishing between files and directories. In `os` it means never *walking* into symlinks, including any symlinks created by the user between iterations. We smooth over these differences with a private sentinel - `os._walk_symlinks_as_files` - that enables the pathlib behaviour.
barneygale · May 29, 2024 · May 26, 2024 · May 29, 2024 · May 26, 2024 · 1f7d5adcf5b2124b3a896488c0dec38a8224fe40
commit 1f7d5adcf5b2124b3a896488c0dec38a8224fe40
diff --git a/Lib/glob.py b/Lib/glob.py
@@ -519,43 +519,6 @@ def select_exists(self, path, exists=False):
         elif self.lexists(path):
             yield path
 
-    @classmethod
-    def walk(cls, root, top_down, on_error, follow_symlinks):
-        """Walk the directory tree from the given root, similar to os.walk().
-        """
-        paths = [root]
-        while paths:
-            path = paths.pop()
-            if isinstance(path, tuple):
-                yield path
-                continue
-            try:
-                with cls.scandir(path) as scandir_it:
-                    dirnames = []
-                    filenames = []
-                    if not top_down:
-                        paths.append((path, dirnames, filenames))
-                    for entry in scandir_it:
-                        name = entry.name
-                        try:
-                            if entry.is_dir(follow_symlinks=follow_symlinks):
-                                if not top_down:
-                                    paths.append(cls.parse_entry(entry))
-                                dirnames.append(name)
-                            else:
-                                filenames.append(name)
-                        except OSError:
-                            filenames.append(name)
-            except OSError as error:
-                if on_error is not None:
-                    on_error(error)
-            else:
-                if top_down:
-                    yield path, dirnames, filenames
-                    if dirnames:
-                        prefix = cls.add_slash(path)
-                        paths += [cls.concat_path(prefix, d) for d in reversed(dirnames)]
-
 
 class _StringGlobber(_Globber):
     lexists = staticmethod(os.path.lexists)

diff --git a/Lib/os.py b/Lib/os.py
@@ -281,6 +281,11 @@ def renames(old, new):
 
 __all__.extend(["makedirs", "removedirs", "renames"])
 
+# Private sentinel that can be passed to os.walk() to classify all symlinks as
+# files, and walk into every path classified as a directory (potentially after
+# user modification in topdown mode). Used by pathlib.Path.walk().
+_walk_symlinks_as_files = object()
+
 def walk(top, topdown=True, onerror=None, followlinks=False):
     """Directory tree generator.
 
@@ -382,7 +387,7 @@ def walk(top, topdown=True, onerror=None, followlinks=False):
                     break
 
                 try:
-                    is_dir = entry.is_dir()
+                    is_dir = entry.is_dir(follow_symlinks=followlinks is not _walk_symlinks_as_files)
                 except OSError:
                     # If is_dir() raises an OSError, consider the entry not to
                     # be a directory, same behaviour as os.path.isdir().

@@ -623,7 +623,37 @@ def rglob(self, pattern, *, case_sensitive=None, recurse_symlinks=True):
 
     def walk(self, top_down=True, on_error=None, follow_symlinks=False):
         """Walk the directory tree from this directory, similar to os.walk()."""
-        return self._globber.walk(self, top_down, on_error, follow_symlinks)
+        paths = [self]
+        while paths:
+            path = paths.pop()
+            if isinstance(path, tuple):
+                yield path
+                continue
+            dirnames = []
+            filenames = []
+            if not top_down:
+                paths.append((path, dirnames, filenames))
+            try:
+                for child in path.iterdir():
+                    try:
+                        if child.is_dir(follow_symlinks=follow_symlinks):
+                            if not top_down:
+                                paths.append(child)
+                            dirnames.append(child.name)
+                        else:
+                            filenames.append(child.name)
+                    except OSError:
+                        filenames.append(child.name)
+            except OSError as error:
+                if on_error is not None:
+                    on_error(error)
+                if not top_down:
+                    while not isinstance(paths.pop(), tuple):
+                        pass
+                continue
+            if top_down:
+                yield path, dirnames, filenames
+                paths += [path.joinpath(d) for d in reversed(dirnames)]
 
     def absolute(self):
         """Return an absolute version of this path

@@ -638,7 +638,9 @@ def walk(self, top_down=True, on_error=None, follow_symlinks=False):
         """Walk the directory tree from this directory, similar to os.walk()."""
         sys.audit("pathlib.Path.walk", self, on_error, follow_symlinks)
         root_dir = str(self)
-        results = self._globber.walk(root_dir, top_down, on_error, follow_symlinks)
+        if not follow_symlinks:
+            follow_symlinks = os._walk_symlinks_as_files
+        results = os.walk(root_dir, top_down, on_error, follow_symlinks)
         for path_str, dirnames, filenames in results:
             if root_dir == '.':
                 path_str = path_str[2:]