Skip to content

Commit

Permalink
GH-77609: Add recurse_symlinks argument to pathlib.Path.glob() (#11…
Browse files Browse the repository at this point in the history 
…7311)

Replace tri-state `follow_symlinks` with boolean `recurse_symlinks` argument. The new argument controls whether symlinks are followed when expanding recursive `**` wildcards. The possible argument values correspond as follows:

    follow_symlinks  recurse_symlinks
    ===============  ================
    False            N/A
    None             False
    True             True

We therefore drop support for not following symlinks when expanding non-recursive pattern parts; it wasn't requested in the original issue, and it's a feature not found in any shells.

This makes the API a easier to grok by eliminating `None` as an option.

No news blurb as `follow_symlinks` was new in 3.13.
  • Loading branch information
@barneygale
barneygale authored Apr 5, 2024
1 parent 4d4a6f1 commit 6150bb2
Show file tree
Hide file tree
Showing 5 changed files with 34 additions and 95 deletions.
21 changes: 8 additions & 13 deletions Doc/library/pathlib.rst
View file
Original file line number Diff line number Diff line change
Expand Up @@ -985,7 +985,7 @@ call fails (for example because the path doesn't exist).
.. versionadded:: 3.5


.. method:: Path.glob(pattern, *, case_sensitive=None, follow_symlinks=None)
.. method:: Path.glob(pattern, *, case_sensitive=None, recurse_symlinks=False)

Glob the given relative *pattern* in the directory represented by this path,
yielding all matching files (of any kind)::
Expand Down Expand Up @@ -1013,28 +1013,23 @@ call fails (for example because the path doesn't exist).
typically, case-sensitive on POSIX, and case-insensitive on Windows.
Set *case_sensitive* to ``True`` or ``False`` to override this behaviour.

By default, or when the *follow_symlinks* keyword-only argument is set to
``None``, this method follows symlinks except when expanding "``**``"
wildcards. Set *follow_symlinks* to ``True`` to always follow symlinks, or
``False`` to treat all symlinks as files.

.. tip::
Set *follow_symlinks* to ``True`` or ``False`` to improve performance
of recursive globbing.
By default, or when the *recurse_symlinks* keyword-only argument is set to
``False``, this method follows symlinks except when expanding "``**``"
wildcards. Set *recurse_symlinks* to ``True`` to always follow symlinks.

.. audit-event:: pathlib.Path.glob self,pattern pathlib.Path.glob

.. versionchanged:: 3.12
The *case_sensitive* parameter was added.

.. versionchanged:: 3.13
The *follow_symlinks* parameter was added.
The *recurse_symlinks* parameter was added.

.. versionchanged:: 3.13
The *pattern* parameter accepts a :term:`path-like object`.


.. method:: Path.rglob(pattern, *, case_sensitive=None, follow_symlinks=None)
.. method:: Path.rglob(pattern, *, case_sensitive=None, recurse_symlinks=False)

Glob the given relative *pattern* recursively. This is like calling
:func:`Path.glob` with "``**/``" added in front of the *pattern*.
Expand All @@ -1048,7 +1043,7 @@ call fails (for example because the path doesn't exist).
The *case_sensitive* parameter was added.

.. versionchanged:: 3.13
The *follow_symlinks* parameter was added.
The *recurse_symlinks* parameter was added.

.. versionchanged:: 3.13
The *pattern* parameter accepts a :term:`path-like object`.
Expand Down Expand Up @@ -1675,7 +1670,7 @@ The patterns accepted and results generated by :meth:`Path.glob` and
passing ``recursive=True`` to :func:`glob.glob`.
3. "``**``" pattern components do not follow symlinks by default in pathlib.
This behaviour has no equivalent in :func:`glob.glob`, but you can pass
``follow_symlinks=True`` to :meth:`Path.glob` for compatible behaviour.
``recurse_symlinks=True`` to :meth:`Path.glob` for compatible behaviour.
4. Like all :class:`PurePath` and :class:`Path` objects, the values returned
from :meth:`Path.glob` and :meth:`Path.rglob` don't include trailing
slashes.
Expand Down
11 changes: 7 additions & 4 deletions Doc/whatsnew/3.13.rst
View file
Original file line number Diff line number Diff line change
Expand Up @@ -559,12 +559,15 @@ pathlib
implementation of :mod:`os.path` used for low-level path parsing and
joining: either ``posixpath`` or ``ntpath``.

* Add *follow_symlinks* keyword-only argument to :meth:`pathlib.Path.glob`,
:meth:`~pathlib.Path.rglob`, :meth:`~pathlib.Path.is_file`,
* Add *recurse_symlinks* keyword-only argument to :meth:`pathlib.Path.glob`
and :meth:`~pathlib.Path.rglob`.
(Contributed by Barney Gale in :gh:`77609`).

* Add *follow_symlinks* keyword-only argument to :meth:`~pathlib.Path.is_file`,
:meth:`~pathlib.Path.is_dir`, :meth:`~pathlib.Path.owner`,
:meth:`~pathlib.Path.group`.
(Contributed by Barney Gale in :gh:`77609` and :gh:`105793`, and
Kamil Turek in :gh:`107962`).
(Contributed by Barney Gale in :gh:`105793`, and Kamil Turek in
:gh:`107962`).

* Return files and directories from :meth:`pathlib.Path.glob` and
:meth:`~pathlib.Path.rglob` when given a pattern that ends with "``**``". In
Expand Down
8 changes: 4 additions & 4 deletions Lib/pathlib/__init__.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -619,17 +619,17 @@ def _make_child_relpath(self, name):
path._tail_cached = tail + [name]
return path

def glob(self, pattern, *, case_sensitive=None, follow_symlinks=None):
def glob(self, pattern, *, case_sensitive=None, recurse_symlinks=False):
"""Iterate over this subtree and yield all existing files (of any
kind, including directories) matching the given relative pattern.
"""
sys.audit("pathlib.Path.glob", self, pattern)
if not isinstance(pattern, PurePath):
pattern = self.with_segments(pattern)
return _abc.PathBase.glob(
self, pattern, case_sensitive=case_sensitive, follow_symlinks=follow_symlinks)
self, pattern, case_sensitive=case_sensitive, recurse_symlinks=recurse_symlinks)

def rglob(self, pattern, *, case_sensitive=None, follow_symlinks=None):
def rglob(self, pattern, *, case_sensitive=None, recurse_symlinks=False):
"""Recursively yield all existing files (of any kind, including
directories) matching the given relative pattern, anywhere in
this subtree.
Expand All @@ -639,7 +639,7 @@ def rglob(self, pattern, *, case_sensitive=None, follow_symlinks=None):
pattern = self.with_segments(pattern)
pattern = '**' / pattern
return _abc.PathBase.glob(
self, pattern, case_sensitive=case_sensitive, follow_symlinks=follow_symlinks)
self, pattern, case_sensitive=case_sensitive, recurse_symlinks=recurse_symlinks)

def walk(self, top_down=True, on_error=None, follow_symlinks=False):
"""Walk the directory tree from this directory, similar to os.walk()."""
Expand Down
20 changes: 8 additions & 12 deletions Lib/pathlib/_abc.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -66,10 +66,8 @@ def _select_special(paths, part):
yield path._make_child_relpath(part)


def _select_children(parent_paths, dir_only, follow_symlinks, match):
def _select_children(parent_paths, dir_only, match):
"""Yield direct children of given paths, filtering by name and type."""
if follow_symlinks is None:
follow_symlinks = True
for parent_path in parent_paths:
try:
# We must close the scandir() object before proceeding to
Expand All @@ -82,7 +80,7 @@ def _select_children(parent_paths, dir_only, follow_symlinks, match):
for entry in entries:
if dir_only:
try:
if not entry.is_dir(follow_symlinks=follow_symlinks):
if not entry.is_dir():
continue
except OSError:
continue
Expand All @@ -96,8 +94,6 @@ def _select_recursive(parent_paths, dir_only, follow_symlinks, match):
"""Yield given paths and all their children, recursively, filtering by
string and type.
"""
if follow_symlinks is None:
follow_symlinks = False
for parent_path in parent_paths:
if match is not None:
# If we're filtering paths through a regex, record the length of
Expand Down Expand Up @@ -789,7 +785,7 @@ def _make_child_direntry(self, entry):
def _make_child_relpath(self, name):
return self.joinpath(name)

def glob(self, pattern, *, case_sensitive=None, follow_symlinks=True):
def glob(self, pattern, *, case_sensitive=None, recurse_symlinks=True):
"""Iterate over this subtree and yield all existing files (of any
kind, including directories) matching the given relative pattern.
"""
Expand Down Expand Up @@ -818,7 +814,7 @@ def glob(self, pattern, *, case_sensitive=None, follow_symlinks=True):
# Consume following non-special components, provided we're
# treating symlinks consistently. Each component is joined
# onto 'part', which is used to generate an re.Pattern object.
if follow_symlinks is not None:
if recurse_symlinks:
while stack and stack[-1] not in specials:
part += sep + stack.pop()

Expand All @@ -827,7 +823,7 @@ def glob(self, pattern, *, case_sensitive=None, follow_symlinks=True):
match = _compile_pattern(part, sep, case_sensitive) if part != '**' else None

# Recursively walk directories, filtering by type and regex.
paths = _select_recursive(paths, bool(stack), follow_symlinks, match)
paths = _select_recursive(paths, bool(stack), recurse_symlinks, match)

# De-duplicate if we've already seen a '**' component.
if deduplicate_paths:
Expand All @@ -843,18 +839,18 @@ def glob(self, pattern, *, case_sensitive=None, follow_symlinks=True):
match = _compile_pattern(part, sep, case_sensitive) if part != '*' else None

# Iterate over directories' children filtering by type and regex.
paths = _select_children(paths, bool(stack), follow_symlinks, match)
paths = _select_children(paths, bool(stack), match)
return paths

def rglob(self, pattern, *, case_sensitive=None, follow_symlinks=True):
def rglob(self, pattern, *, case_sensitive=None, recurse_symlinks=True):
"""Recursively yield all existing files (of any kind, including
directories) matching the given relative pattern, anywhere in
this subtree.
"""
if not isinstance(pattern, PurePathBase):
pattern = self.with_segments(pattern)
pattern = '**' / pattern
return self.glob(pattern, case_sensitive=case_sensitive, follow_symlinks=follow_symlinks)
return self.glob(pattern, case_sensitive=case_sensitive, recurse_symlinks=recurse_symlinks)

def walk(self, top_down=True, on_error=None, follow_symlinks=False):
"""Walk the directory tree from this directory, similar to os.walk()."""
Expand Down
69 changes: 7 additions & 62 deletions Lib/test/test_pathlib/test_pathlib_abc.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1776,9 +1776,9 @@ def _check(path, pattern, case_sensitive, expected):
_check(path, "dirb/file*", False, ["dirB/fileB"])

@needs_symlinks
def test_glob_follow_symlinks_common(self):
def test_glob_recurse_symlinks_common(self):
def _check(path, glob, expected):
actual = {path for path in path.glob(glob, follow_symlinks=True)
actual = {path for path in path.glob(glob, recurse_symlinks=True)
if path.parts.count("linkD") <= 1} # exclude symlink loop.
self.assertEqual(actual, { P(self.base, q) for q in expected })
P = self.cls
Expand Down Expand Up @@ -1812,39 +1812,9 @@ def _check(path, glob, expected):
_check(p, "*/dirD/**", ["dirC/dirD/", "dirC/dirD/fileD"])
_check(p, "*/dirD/**/", ["dirC/dirD/"])

@needs_symlinks
def test_glob_no_follow_symlinks_common(self):
def _check(path, glob, expected):
actual = {path for path in path.glob(glob, follow_symlinks=False)}
self.assertEqual(actual, { P(self.base, q) for q in expected })
P = self.cls
p = P(self.base)
_check(p, "fileB", [])
_check(p, "dir*/file*", ["dirB/fileB", "dirC/fileC"])
_check(p, "*A", ["dirA", "fileA", "linkA"])
_check(p, "*B/*", ["dirB/fileB", "dirB/linkD"])
_check(p, "*/fileB", ["dirB/fileB"])
_check(p, "*/", ["dirA/", "dirB/", "dirC/", "dirE/"])
_check(p, "dir*/*/..", ["dirC/dirD/.."])
_check(p, "dir*/**", [
"dirA/", "dirA/linkC",
"dirB/", "dirB/fileB", "dirB/linkD",
"dirC/", "dirC/fileC", "dirC/dirD", "dirC/dirD/fileD", "dirC/novel.txt",
"dirE/"])
_check(p, "dir*/**/", ["dirA/", "dirB/", "dirC/", "dirC/dirD/", "dirE/"])
_check(p, "dir*/**/..", ["dirA/..", "dirB/..", "dirC/..", "dirC/dirD/..", "dirE/.."])
_check(p, "dir*/*/**", ["dirC/dirD/", "dirC/dirD/fileD"])
_check(p, "dir*/*/**/", ["dirC/dirD/"])
_check(p, "dir*/*/**/..", ["dirC/dirD/.."])
_check(p, "dir*/**/fileC", ["dirC/fileC"])
_check(p, "dir*/*/../dirD/**", ["dirC/dirD/../dirD/", "dirC/dirD/../dirD/fileD"])
_check(p, "dir*/*/../dirD/**/", ["dirC/dirD/../dirD/"])
_check(p, "*/dirD/**", ["dirC/dirD/", "dirC/dirD/fileD"])
_check(p, "*/dirD/**/", ["dirC/dirD/"])

def test_rglob_follow_symlinks_none(self):
def test_rglob_recurse_symlinks_false(self):
def _check(path, glob, expected):
actual = set(path.rglob(glob, follow_symlinks=None))
actual = set(path.rglob(glob, recurse_symlinks=False))
self.assertEqual(actual, { P(self.base, q) for q in expected })
P = self.cls
p = P(self.base)
Expand Down Expand Up @@ -1901,9 +1871,9 @@ def test_rglob_windows(self):
self.assertEqual(set(map(str, p.rglob("FILEd"))), {f"{p}\\dirD\\fileD"})

@needs_symlinks
def test_rglob_follow_symlinks_common(self):
def test_rglob_recurse_symlinks_common(self):
def _check(path, glob, expected):
actual = {path for path in path.rglob(glob, follow_symlinks=True)
actual = {path for path in path.rglob(glob, recurse_symlinks=True)
if path.parts.count("linkD") <= 1} # exclude symlink loop.
self.assertEqual(actual, { P(self.base, q) for q in expected })
P = self.cls
Expand Down Expand Up @@ -1932,37 +1902,12 @@ def _check(path, glob, expected):
_check(p, "*.txt", ["dirC/novel.txt"])
_check(p, "*.*", ["dirC/novel.txt"])

@needs_symlinks
def test_rglob_no_follow_symlinks_common(self):
def _check(path, glob, expected):
actual = {path for path in path.rglob(glob, follow_symlinks=False)}
self.assertEqual(actual, { P(self.base, q) for q in expected })
P = self.cls
p = P(self.base)
_check(p, "fileB", ["dirB/fileB"])
_check(p, "*/fileA", [])
_check(p, "*/fileB", ["dirB/fileB"])
_check(p, "file*", ["fileA", "dirB/fileB", "dirC/fileC", "dirC/dirD/fileD", ])
_check(p, "*/", ["dirA/", "dirB/", "dirC/", "dirC/dirD/", "dirE/"])
_check(p, "", ["", "dirA/", "dirB/", "dirC/", "dirE/", "dirC/dirD/"])

p = P(self.base, "dirC")
_check(p, "*", ["dirC/fileC", "dirC/novel.txt",
"dirC/dirD", "dirC/dirD/fileD"])
_check(p, "file*", ["dirC/fileC", "dirC/dirD/fileD"])
_check(p, "*/*", ["dirC/dirD/fileD"])
_check(p, "*/", ["dirC/dirD/"])
_check(p, "", ["dirC/", "dirC/dirD/"])
# gh-91616, a re module regression
_check(p, "*.txt", ["dirC/novel.txt"])
_check(p, "*.*", ["dirC/novel.txt"])

@needs_symlinks
def test_rglob_symlink_loop(self):
# Don't get fooled by symlink loops (Issue #26012).
P = self.cls
p = P(self.base)
given = set(p.rglob('*', follow_symlinks=None))
given = set(p.rglob('*', recurse_symlinks=False))
expect = {'brokenLink',
'dirA', 'dirA/linkC',
'dirB', 'dirB/fileB', 'dirB/linkD',
Expand Down

0 comments on commit 6150bb2

Please sign in to comment.