|
1
|
|
|
import sys |
|
|
|
|
|
|
2
|
|
|
import warnings |
|
|
|
|
|
|
3
|
|
|
from typing import Callable, Mapping, Optional, Sequence |
|
|
|
|
|
|
4
|
|
|
|
|
5
|
|
|
import regex |
|
|
|
|
|
|
6
|
|
|
|
|
7
|
|
|
from pocketutils.core.exceptions import * |
|
|
|
|
|
|
8
|
|
|
from pocketutils.tools.base_tools import BaseTools |
|
9
|
|
|
|
|
10
|
|
|
logger = logging.getLogger("pocketutils") |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
class PathTools(BaseTools): |
|
|
|
|
|
|
14
|
|
|
@classmethod |
|
15
|
|
|
def updir(cls, n: int, *parts) -> Path: |
|
|
|
|
|
|
16
|
|
|
""" |
|
17
|
|
|
Get an absolute path ``n`` parents from ``os.getcwd()``. |
|
18
|
|
|
Does not sanitize. |
|
19
|
|
|
|
|
20
|
|
|
Ex: In dir '/home/john/dir_a/dir_b': |
|
21
|
|
|
updir(2, 'dir1', 'dir2') # returns Path('/home/john/dir1/dir2') |
|
22
|
|
|
""" |
|
23
|
|
|
base = Path(os.getcwd()) |
|
24
|
|
|
for _ in range(n): |
|
25
|
|
|
base = base.parent |
|
26
|
|
|
for part in parts: |
|
27
|
|
|
base = base / part |
|
28
|
|
|
return base.resolve() |
|
29
|
|
|
|
|
30
|
|
|
@classmethod |
|
31
|
|
|
def guess_trash(cls) -> Path: |
|
32
|
|
|
""" |
|
33
|
|
|
Chooses a reasonable path for trash based on the OS. |
|
34
|
|
|
This is not reliable. For a more sophisticated solution, see https://github.com/hsoft/send2trash |
|
|
|
|
|
|
35
|
|
|
However, even that can fail. |
|
36
|
|
|
""" |
|
37
|
|
|
plat = sys.platform.lower() |
|
38
|
|
|
if "darwin" in plat: |
|
|
|
|
|
|
39
|
|
|
return Path.home() / ".Trash" |
|
40
|
|
|
elif "win" in plat: |
|
41
|
|
|
return Path(Path.home().root) / "$Recycle.Bin" |
|
42
|
|
|
else: |
|
43
|
|
|
return Path.home() / ".trash" |
|
44
|
|
|
|
|
45
|
|
|
@classmethod |
|
46
|
|
|
def prep_dir(cls, path: PathLike, exist_ok: bool = True) -> bool: |
|
47
|
|
|
""" |
|
48
|
|
|
Prepares a directory by making it if it doesn't exist. |
|
49
|
|
|
If exist_ok is False, calls logger.warning it already exists |
|
50
|
|
|
""" |
|
51
|
|
|
path = Path(path) |
|
52
|
|
|
exists = path.exists() |
|
53
|
|
|
# On some platforms we get generic exceptions like permissions errors, |
|
54
|
|
|
# so these are better |
|
55
|
|
|
if exists and not path.is_dir(): |
|
56
|
|
|
raise DirDoesNotExistError(f"Path {path} exists but is not a file") |
|
57
|
|
|
if exists and not exist_ok: |
|
58
|
|
|
logger.warning(f"Directory {path} already exists") |
|
|
|
|
|
|
59
|
|
|
if not exists: |
|
60
|
|
|
# NOTE! exist_ok in mkdir throws an error on Windows |
|
61
|
|
|
path.mkdir(parents=True) |
|
62
|
|
|
return exists |
|
63
|
|
|
|
|
64
|
|
|
@classmethod |
|
65
|
|
|
def prep_file(cls, path: PathLike, exist_ok: bool = True) -> None: |
|
66
|
|
|
""" |
|
67
|
|
|
Prepares a file path by making its parent directory. |
|
68
|
|
|
Same as ``pathlib.Path.mkdir`` but makes sure ``path`` is a file if it exists. |
|
69
|
|
|
""" |
|
70
|
|
|
# On some platforms we get generic exceptions like permissions errors, so these are better |
|
71
|
|
|
path = Path(path) |
|
72
|
|
|
# check for errors first; don't make the dirs and then fail |
|
73
|
|
|
if path.exists() and not path.is_file() and not path.is_symlink(): |
|
74
|
|
|
raise FileDoesNotExistError(f"Path {path} exists but is not a file") |
|
75
|
|
|
Path(path.parent).mkdir(parents=True, exist_ok=exist_ok) |
|
76
|
|
|
|
|
77
|
|
|
@classmethod |
|
78
|
|
|
def sanitize_path( |
|
79
|
|
|
cls, |
|
|
|
|
|
|
80
|
|
|
path: PathLike, |
|
|
|
|
|
|
81
|
|
|
is_file: Optional[bool] = None, |
|
|
|
|
|
|
82
|
|
|
*, |
|
|
|
|
|
|
83
|
|
|
show_warnings: Union[bool, Callable[[str], Any]] = True, |
|
|
|
|
|
|
84
|
|
|
) -> Path: |
|
85
|
|
|
r""" |
|
86
|
|
|
Sanitizes a path for major OSes and filesystems. |
|
87
|
|
|
Also see sanitize_path_nodes and sanitize_path_node. |
|
88
|
|
|
Platform-dependent. |
|
89
|
|
|
A corner case is drive letters in Linux: |
|
90
|
|
|
"C:\\Users\\john" is converted to '/C:/users/john' if os.name=='posix' |
|
91
|
|
|
""" |
|
92
|
|
|
# the idea is to sanitize for both Windows and Posix, regardless of the platform in use |
|
93
|
|
|
# the sanitization should be as uniform as possible for both platforms |
|
94
|
|
|
# this works for at least Windows+NTFS |
|
95
|
|
|
# tilde substitution for long filenames in Windows -- is unsupported |
|
96
|
|
|
w = {True: logger.warning, False: lambda _: None}.get(show_warnings, show_warnings) |
|
|
|
|
|
|
97
|
|
|
if path.startswith("\\\\?"): |
|
98
|
|
|
w(f"Long UNC Windows paths (\\\\? prefix) are not supported (path '{path}')", path=path) |
|
99
|
|
|
bits = str(path).strip().replace("\\", "/").split("/") |
|
100
|
|
|
new_path = cls.sanitize_path_nodes(bits, is_file=is_file) |
|
101
|
|
|
if new_path != path: |
|
102
|
|
|
w(f"Sanitized filename {path} → {new_path}") |
|
103
|
|
|
return Path(new_path) |
|
104
|
|
|
|
|
105
|
|
|
@classmethod |
|
106
|
|
|
def sanitize_path_nodes(cls, bits: Sequence[PathLike], is_file: Optional[bool] = None) -> Path: |
|
|
|
|
|
|
107
|
|
|
fixed_bits = [ |
|
108
|
|
|
bit + os.sep |
|
109
|
|
|
if i == 0 and bit.strip() in ["", ".", ".."] |
|
110
|
|
|
else cls.sanitize_path_node( |
|
111
|
|
|
bit, |
|
112
|
|
|
is_file=(False if i < len(bits) - 1 else is_file), |
|
113
|
|
|
is_root_or_drive=(None if i == 0 else False), |
|
114
|
|
|
) |
|
115
|
|
|
for i, bit in enumerate(bits) |
|
116
|
|
|
if bit.strip() not in ["", "."] |
|
117
|
|
|
or i == 0 # ignore // (empty) just like Path does (but fail on sanitize_path_node(' ')) |
|
118
|
|
|
] |
|
119
|
|
|
fixed_bits = [bit for i, bit in enumerate(fixed_bits) if i == 0 or bit not in ["", "."]] |
|
120
|
|
|
# unfortunately POSIX turns Path('C:\', '5') into C:\/5 |
|
121
|
|
|
# this isn't an ideal way to fix it, but it works |
|
122
|
|
|
pat = regex.compile(r"^([A-Z]:)(?:\\)?$", flags=regex.V1) |
|
123
|
|
|
if os.name == "posix" and len(fixed_bits) > 0 and pat.fullmatch(fixed_bits[0]): |
|
124
|
|
|
fixed_bits[0] = fixed_bits[0].rstrip("\\") |
|
125
|
|
|
fixed_bits.insert(0, "/") |
|
126
|
|
|
return Path(*fixed_bits) |
|
127
|
|
|
|
|
128
|
|
|
@classmethod |
|
129
|
|
|
def sanitize_path_node( |
|
130
|
|
|
cls, |
|
|
|
|
|
|
131
|
|
|
bit: PathLike, |
|
|
|
|
|
|
132
|
|
|
is_file: Optional[bool] = None, |
|
|
|
|
|
|
133
|
|
|
is_root_or_drive: Optional[bool] = None, |
|
|
|
|
|
|
134
|
|
|
include_fat: bool = False, |
|
|
|
|
|
|
135
|
|
|
) -> str: |
|
136
|
|
|
r""" |
|
137
|
|
|
Sanitizes a path node such that it will be fine for major OSes and filesystems. |
|
138
|
|
|
For example: |
|
139
|
|
|
- 'plums;and/or;apples' becomes 'plums_and_or_apples' (escaped ; and /) |
|
140
|
|
|
- 'null.txt' becomes '_null_.txt' ('null' is forbidden in Windows) |
|
141
|
|
|
- 'abc ' becomes 'abc' (no trailing spaces) |
|
142
|
|
|
The behavior is platform-independent -- os, sys, and pathlib are not used. |
|
143
|
|
|
For ex, calling sanitize_path_node(r'C:\') returns r'C:\' on both Windows and Linux |
|
144
|
|
|
If you want to sanitize a whole path, see sanitize_path instead. |
|
145
|
|
|
|
|
146
|
|
|
Args: |
|
147
|
|
|
bit: The node |
|
148
|
|
|
is_file: False for directories, True otherwise, None if unknown |
|
149
|
|
|
is_root_or_drive: True if known to be the root ('/') or a drive ('C:\'), None if unknown |
|
150
|
|
|
include_fat: Also make compatible with FAT filesystems |
|
151
|
|
|
|
|
152
|
|
|
Returns: |
|
153
|
|
|
A string |
|
154
|
|
|
""" |
|
155
|
|
|
# since is_file and is_root_or_drive are both Optional[bool], let's be explicit and use 'is' for clarity |
|
|
|
|
|
|
156
|
|
|
if is_file is True and is_root_or_drive is True: |
|
157
|
|
|
raise ContradictoryRequestError("is_file and is_root_or_drive are both true") |
|
158
|
|
|
if is_file is True and is_root_or_drive is None: |
|
159
|
|
|
is_root_or_drive = False |
|
160
|
|
|
if is_root_or_drive is True and is_file is None: |
|
161
|
|
|
is_file = False |
|
162
|
|
|
source_bit = copy(str(bit)) |
|
163
|
|
|
bit = str(bit).strip() |
|
164
|
|
|
# first, catch root or drive as long as is_root_or_drive is not false |
|
165
|
|
|
# if is_root_or_drive is True (which is a weird call), then fail if it's not |
|
166
|
|
|
# otherwise, it's not a root or drive letter, so keep going |
|
167
|
|
|
if is_root_or_drive is not False: |
|
168
|
|
|
# \ is allowed in Windows |
|
169
|
|
|
if bit in ["/", "\\"]: |
|
170
|
|
|
return bit |
|
171
|
|
|
m = regex.compile(r"^([A-Z]:)(?:\\)?$", flags=regex.V1).fullmatch(bit) |
|
|
|
|
|
|
172
|
|
|
# this is interesting |
|
173
|
|
|
# for bit=='C:' and is_root_or_drive=None, |
|
174
|
|
|
# it could be either a drive letter |
|
175
|
|
|
# or a file path that should be corrected to 'C_' |
|
176
|
|
|
# I guess here we're going with a drive letter |
|
177
|
|
|
if m is not None: |
|
178
|
|
|
# we need C:\ and not C: because: |
|
179
|
|
|
# Path('C:\\', '5').is_absolute() is True |
|
180
|
|
|
# but Path('C:', '5').is_absolute() is False |
|
181
|
|
|
# unfortunately, doing Path('C:\\', '5') on Linux gives 'C:\\/5' |
|
182
|
|
|
# I can't handle that here, but sanitize_path() will account for it |
|
183
|
|
|
return m.group(1) + "\\" |
|
184
|
|
|
if is_root_or_drive is True: |
|
185
|
|
|
raise IllegalPathError(f"Node '{bit}' is not the root or a drive letter") |
|
186
|
|
|
# note that we can't call WindowsPath.is_reserved because it can't be instantiated on non-Linux |
|
|
|
|
|
|
187
|
|
|
# also, these appear to be different from the ones defined there |
|
188
|
|
|
bad_chars = { |
|
189
|
|
|
"<", |
|
190
|
|
|
">", |
|
191
|
|
|
":", |
|
192
|
|
|
'"', |
|
193
|
|
|
"|", |
|
194
|
|
|
"?", |
|
195
|
|
|
"*", |
|
196
|
|
|
"\\", |
|
197
|
|
|
"/", |
|
198
|
|
|
*{chr(c) for c in range(128, 128 + 33)}, |
|
199
|
|
|
*{chr(c) for c in range(0, 32)}, |
|
200
|
|
|
"\t", |
|
201
|
|
|
} |
|
202
|
|
|
# don't handle Long UNC paths |
|
203
|
|
|
# also cannot be blank or whitespace |
|
204
|
|
|
# the $ suffixed ones are for FAT |
|
205
|
|
|
# no CLOCK$, even with an ext |
|
206
|
|
|
# also no SCREEN$ |
|
207
|
|
|
bad_strs = { |
|
208
|
|
|
"CON", |
|
209
|
|
|
"PRN", |
|
210
|
|
|
"AUX", |
|
211
|
|
|
"NUL", |
|
212
|
|
|
"COM1", |
|
213
|
|
|
"COM2", |
|
214
|
|
|
"COM3", |
|
215
|
|
|
"COM4", |
|
216
|
|
|
"COM5", |
|
217
|
|
|
"COM6", |
|
218
|
|
|
"COM7", |
|
219
|
|
|
"COM8", |
|
220
|
|
|
"COM9", |
|
221
|
|
|
"LPT1", |
|
222
|
|
|
"LPT2", |
|
223
|
|
|
"LPT3", |
|
224
|
|
|
"LPT4", |
|
225
|
|
|
"LPT5", |
|
226
|
|
|
"LPT6", |
|
227
|
|
|
"LPT7", |
|
228
|
|
|
"LPT8", |
|
229
|
|
|
"LPT9", |
|
230
|
|
|
} |
|
231
|
|
|
if include_fat: |
|
232
|
|
|
bad_strs += {"$IDLE$", "CONFIG$", "KEYBD$", "SCREEN$", "CLOCK$", "LST"} |
|
233
|
|
|
# just dots is invalid |
|
234
|
|
|
if set(bit.replace(" ", "")) == "." and bit not in ["..", "."]: |
|
235
|
|
|
raise IllegalPathError(f"Node '{source_bit}' is invalid") |
|
236
|
|
|
for q in bad_chars: |
|
|
|
|
|
|
237
|
|
|
bit = bit.replace(q, "_") |
|
238
|
|
|
if bit.upper() in bad_strs: |
|
239
|
|
|
# arbitrary decision |
|
240
|
|
|
bit = "_" + bit + "_" |
|
241
|
|
|
else: |
|
242
|
|
|
stub, ext = os.path.splitext(bit) |
|
243
|
|
|
if stub.upper() in bad_strs: |
|
244
|
|
|
bit = "_" + stub + "_" + ext |
|
245
|
|
|
if bit.strip() == "": |
|
246
|
|
|
raise IllegalPathError(f"Node '{source_bit}' is empty or contains only whitespace") |
|
247
|
|
|
# do this after |
|
248
|
|
|
if len(bit) > 254: |
|
249
|
|
|
raise IllegalPathError(f"Node '{source_bit}' has more than 254 characters") |
|
250
|
|
|
bit = bit.strip() |
|
251
|
|
|
if is_file is not True and (bit == "." or bit == ".."): |
|
|
|
|
|
|
252
|
|
|
return bit |
|
253
|
|
|
# never allow '.' (or ' ') to end a filename |
|
254
|
|
|
bit = bit.rstrip(".") |
|
255
|
|
|
return bit |
|
256
|
|
|
|
|
257
|
|
|
|
|
258
|
|
|
__all__ = ["PathTools"] |
|
259
|
|
|
|
dmyersturnbull /
pocketutils
Loading history...