|
1
|
|
|
# SPDX-FileCopyrightText: 2020 Peter Bittner <[email protected]> |
|
2
|
|
|
# |
|
3
|
|
|
# SPDX-License-Identifier: GPL-3.0-or-later |
|
4
|
|
|
|
|
5
|
|
|
"""Directory traversal and ignore pattern matching.""" |
|
6
|
|
|
|
|
7
|
|
|
from __future__ import annotations |
|
8
|
|
|
|
|
9
|
|
|
import logging |
|
10
|
|
|
import os |
|
11
|
|
|
from pathlib import Path |
|
12
|
|
|
|
|
13
|
|
|
from .runner import Runner |
|
14
|
|
|
|
|
15
|
|
|
log = logging.getLogger(__name__) |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
def normalize(path_pattern: str) -> str: |
|
19
|
|
|
""" |
|
20
|
|
|
Normalize path separators in a pattern for cross-platform support. |
|
21
|
|
|
|
|
22
|
|
|
On Windows, both forward slash and backslash are valid path separators. |
|
23
|
|
|
On Unix/Posix, only forward slash is valid (backslash can be part of filename). |
|
24
|
|
|
""" |
|
25
|
|
|
return path_pattern.replace(os.sep, os.altsep or os.sep) |
|
26
|
|
|
|
|
27
|
|
|
|
|
28
|
|
|
def should_ignore(pathname: str, ignore_patterns: list[str] | None) -> bool: |
|
29
|
|
|
""" |
|
30
|
|
|
Check if a path should be ignored based on ignore patterns. |
|
31
|
|
|
|
|
32
|
|
|
Patterns can be: |
|
33
|
|
|
- Simple names like 'bar': matches any directory with that name |
|
34
|
|
|
- Paths like 'foo/bar': matches 'bar' directory inside 'foo' directory |
|
35
|
|
|
and also ignores everything inside that directory |
|
36
|
|
|
""" |
|
37
|
|
|
if not ignore_patterns: |
|
38
|
|
|
return False |
|
39
|
|
|
|
|
40
|
|
|
path = Path(pathname) |
|
41
|
|
|
|
|
42
|
|
|
for pattern in ignore_patterns: |
|
43
|
|
|
pattern_parts = Path(normalize(pattern)).parts |
|
44
|
|
|
if len(pattern_parts) > 1: |
|
45
|
|
|
if len(path.parts) < len(pattern_parts): |
|
46
|
|
|
continue |
|
47
|
|
|
for i in range(len(path.parts) - len(pattern_parts) + 1): |
|
48
|
|
|
path_slice = path.parts[i : i + len(pattern_parts)] |
|
49
|
|
|
if path_slice == pattern_parts: |
|
50
|
|
|
return True |
|
51
|
|
|
elif path.name == pattern: |
|
52
|
|
|
return True |
|
53
|
|
|
return False |
|
54
|
|
|
|
|
55
|
|
|
|
|
56
|
|
|
def descend_and_clean(directory, file_types, dir_names): |
|
57
|
|
|
""" |
|
58
|
|
|
Walk and descend a directory tree, cleaning up files of a certain type |
|
59
|
|
|
along the way. Only delete directories if they are empty, in the end. |
|
60
|
|
|
""" |
|
61
|
|
|
for child in sorted(os.scandir(directory), key=lambda e: e.name): |
|
62
|
|
|
if child.is_file(): |
|
63
|
|
|
if Path(child.path).suffix in file_types: |
|
64
|
|
|
Runner.unlink(Path(child.path)) |
|
65
|
|
|
elif child.is_dir(): |
|
66
|
|
|
if should_ignore(child.path, Runner.ignore): |
|
67
|
|
|
log.debug('Skipping %s', child.name) |
|
68
|
|
|
else: |
|
69
|
|
|
descend_and_clean(child.path, file_types, dir_names) |
|
70
|
|
|
|
|
71
|
|
|
if child.name in dir_names: |
|
72
|
|
|
Runner.rmdir(Path(child.path)) |
|
73
|
|
|
else: |
|
74
|
|
|
log.debug('Ignoring %s (neither a file nor a folder)', child.name) |
|
75
|
|
|
|
bittner /
pyclean
