timotheuslin · April 5, 2020 05:07 · Apr 5, 2020 · Apr 5, 2020
diff --git a/scandir.py b/scandir.py
@@ -44,18 +44,18 @@ def _nt(fx):
 
     def _scandir():
         dirX = [pathlib.Path(basedir)]
-        fileX = []
-        while dirX or fileX:
-            if fileX:
-                yield fileX.pop()
+        entX = []
+        while dirX or entX:
+            if entX:
+                yield entX.pop()
                 continue
             try:
                 fx0 = [f for f in sorted(dirX.pop().glob('*'), reverse=True)
                     if not _nt(f) and not _dot(f) and not _match(f)]
             except StopIteration:
                 continue
             dirX += [d for d in fx0 if d.is_dir()]
-            fileX += [f for f in fx0 if f.is_file()]
+            entX += fx0
 
     for f in _scandir():
         yield f

diff --git a/scandir.py b/scandir.py
@@ -0,0 +1,65 @@
+""" A customized scandir() which honors: (1)the non-traverse anchor files, (2) the
+excluded file pattern. 
+
+Timothy Lin (c) 4/4/2020
+"""
+
+__all__ = ['scandir', 'NO_TRAVERSE_ANCHOR']
+
+import os
+import sys
+import pathlib
+
+# an anchor file to instruct the tool to ignore this folder altogether.
+NO_TRAVERSE_ANCHOR = {'.scandir.ignore'}
+
+def scandir(basedir, exclude=None, nontraverse=None):
+    """ a non-recursive pathlib-based scandir() with breadth-first search.
+
+    basedir - the base directory to start traversing.
+    exclude - a set() of the pattern to match
+    nontraverse - a set() of anchor-file names to indicate folders to be ignored when they
+        contains any one of these files.
+
+    scandir() yields a pathlib.Path object.
+    
+    Notes:
+        1. A file or folder with a leading '.' in its name is ignored. This follows the Un*x
+            convention.
+        2. If a folder contains an anchor file, all its subordinate directories and files
+            are ignored."""
+
+    exclude = {} if (exclude is None) else {os.fspath(pathlib.Path(e)) for e in exclude}
+    if nontraverse is None:
+        nontraverse = NO_TRAVERSE_ANCHOR
+
+    def _dot(fx):
+        return fx.name[0] == '.'
+
+    def _match(fx):
+        return bool({ex for ex in exclude if fx.match(ex)})
+
+    def _nt(fx):
+        return next(iter('')) if (fx.name in nontraverse) else False
+
+    def _scandir():
+        dirX = [pathlib.Path(basedir)]
+        fileX = []
+        while dirX or fileX:
+            if fileX:
+                yield fileX.pop()
+                continue
+            try:
+                fx0 = [f for f in sorted(dirX.pop().glob('*'), reverse=True)
+                    if not _nt(f) and not _dot(f) and not _match(f)]
+            except StopIteration:
+                continue
+            dirX += [d for d in fx0 if d.is_dir()]
+            fileX += [f for f in fx0 if f.is_file()]
+
+    for f in _scandir():
+        yield f
+
+if __name__ == '__main__':
+    for n in scandir(sys.argv[1]):
+        print(f'{os.fspath(n)}')
No results found