Switch custom rglob to glob library

- Use built-in method for ClouthPath's since `glob.glob` does not properly recurse through
sub-directories. If it is a local path, just fall back to `glob.glob`. Alternatively, could
just check for Python version and only run this if <py3.13.

Author: kaitj

bids2table/_indexing.py

@@ -10,9 +10,11 @@
 import re
 from concurrent.futures import Executor, ProcessPoolExecutor
 from functools import lru_cache, partial
+from glob import glob
 from typing import Any, Callable, Generator, Iterable, Sequence
 
 import pyarrow as pa
+from cloudpathlib import CloudPath
 from tqdm import tqdm
 
 from ._entities import (
@@ -21,7 +23,7 @@
     validate_bids_entities,
 )
 from ._logging import setup_logger
-from ._pathlib import PathT, as_path, rglob
+from ._pathlib import PathT, as_path
 
 _BIDS_SUBJECT_DIR_PATTERN = re.compile(r"sub-[a-zA-Z0-9]+")
 
@@ -353,7 +355,7 @@ def _is_bids_dataset(path: PathT) -> bool:
 def _contains_bids_subject_dirs(root: PathT) -> bool:
     """Check if a path contains one or more BIDS subject dirs."""
     # Nb, this will return on the first matching path thanks to the generator.
-    return any(_is_bids_subject_dir(path) for path in root.glob(pattern="sub-*"))
+    return any(_is_bids_subject_dir(path) for path in root.glob("sub-*"))
 
 
 def _find_bids_subject_dirs(
@@ -365,7 +367,7 @@ def _find_bids_subject_dirs(
     Note, only looks one level down. Does not find nested subject directories, e.g. in
     derivatives datasets.
     """
-    paths = [path for path in root.glob(pattern="sub-*") if _is_bids_subject_dir(path)]
+    paths = [path for path in root.glob("sub-*") if _is_bids_subject_dir(path)]
 
     if include_subjects:
         filtered_names = _filter_include(
@@ -400,7 +402,15 @@ def _index_bids_subject_dir(
     _, subject = path.name.split("-", maxsplit=1)
 
     records = []
-    for p in rglob(path=path, pattern="sub-*"):
+    if isinstance(path, CloudPath):
+        # glob fails to properly recurse CloudPaths, so use built-in method
+        paths = path.rglob("sub-*")
+    else:
+        # Fall back to glob for any version of Python
+        paths = glob(f"{path}/**/sub-*", recursive=True)
+
+    for p in paths:
+        p = as_path(p)
         if _is_bids_file(p):
             entities = _cache_parse_bids_entities(p)
             valid_entities, extra_entities = validate_bids_entities(entities)

bids2table/_pathlib.py

@@ -1,6 +1,3 @@
-import fnmatch
-import os
-from collections.abc import Iterator
 from pathlib import Path
 
 try:
@@ -31,32 +28,3 @@ def as_path(path: str | PathT) -> PathT:
 def cloudpathlib_is_available() -> bool:
     """Check if cloudpathlib is available."""
     return _CLOUDPATHLIB_AVAILABLE
-
-
-def rglob(path: PathT, pattern: str, follow_symlinks: bool = True) -> Iterator[PathT]:
-    """Safely glob paths, recursing symlinks.
-
-    NOTE: Only needed to support recursive globbing in <py3.13
-    """
-    # Not expecting symlinks for cloudpaths so just use their glob method.
-    if isinstance(path, CloudPath):
-        yield from path.rglob(f"{pattern}")
-        return
-
-    visited = set()
-    for dirpath, dirnames, filenames in os.walk(path, followlinks=follow_symlinks):
-        dirpath = as_path(dirpath)
-        try:
-            stat = os.stat(dirpath, follow_symlinks=False)
-        except FileNotFoundError:
-            continue  # Broken symlink or race condition
-
-        inode = (stat.st_dev, stat.st_ino)
-        if inode in visited:
-            continue
-        visited.add(inode)
-
-        all_names = [name for name in dirnames] + [name for name in filenames]
-        for name in all_names:
-            if fnmatch.fnmatch(name, pattern):
-                yield dirpath / name