|
1
|
|
|
# -*- coding: utf-8 -*- |
|
2
|
|
|
# emacs: -*- mode: python; py-indent-offset: 4; indent-tabs-mode: nil -*- |
|
3
|
|
|
# vi: set ft=python sts=4 ts=4 sw=4 et: |
|
4
|
|
|
""" |
|
5
|
|
|
Crumb manipulation utilities |
|
6
|
|
|
""" |
|
7
|
|
|
import os |
|
8
|
|
|
from string import Formatter |
|
9
|
|
|
from typing import Iterable, Tuple, Dict, Iterator |
|
10
|
|
|
|
|
11
|
|
|
_txt_idx = 0 |
|
12
|
|
|
_fld_idx = 1 |
|
13
|
|
|
_rgx_idx = 2 |
|
14
|
|
|
_cnv_idx = 3 |
|
15
|
|
|
|
|
16
|
|
|
|
|
17
|
|
|
def _yield_items(crumb_path: str, index=None) -> Iterator[str]: |
|
18
|
|
|
""" An iterator over the items in `crumb_path` given by string.Formatter.""" |
|
19
|
|
|
if index is None: |
|
20
|
|
|
return Formatter().parse(crumb_path) |
|
21
|
|
|
|
|
22
|
|
|
# for (literal_text, field_name, format_spec, conversion) in fmt.parse(crumb_path): |
|
23
|
|
|
# (txt, fld, fmt, conv) |
|
24
|
|
|
return (items[index] for items in Formatter().parse(crumb_path) if items[index] is not None) |
|
25
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
def _enum_items(crumb_path: str) -> Iterator[Tuple[int, str]]: |
|
28
|
|
|
""" An iterator over the enumerated items, i.e., (index, items) in |
|
29
|
|
|
`crumb_path` given by string.Formatter. """ |
|
30
|
|
|
yield from enumerate(Formatter().parse(crumb_path)) |
|
31
|
|
|
|
|
32
|
|
|
|
|
33
|
|
|
def _depth_items(crumb_path: str, index: int = None) -> Iterator[Tuple[int, str]]: |
|
34
|
|
|
""" Return a generator with (depth, items) in `crumb_path`. Being `depth` |
|
35
|
|
|
the place in the file path each argument is.""" |
|
36
|
|
|
if index is None: |
|
37
|
|
|
index = slice(_txt_idx, _cnv_idx + 1) |
|
38
|
|
|
|
|
39
|
|
|
depth = 0 |
|
40
|
|
|
for idx, items in _enum_items(crumb_path): |
|
41
|
|
|
if items[_fld_idx]: |
|
42
|
|
|
depth += items[_txt_idx].count(os.path.sep) |
|
43
|
|
|
yield depth, items[index] |
|
44
|
|
|
|
|
45
|
|
|
|
|
46
|
|
|
def _arg_names(crumb_path: str) -> Iterator[str]: |
|
47
|
|
|
""" Return an iterator over arg_name in crumb_path.""" |
|
48
|
|
|
yield from _yield_items(crumb_path, _fld_idx) |
|
49
|
|
|
|
|
50
|
|
|
|
|
51
|
|
|
def _depth_names(crumb_path: str) -> Iterator[Tuple[int, str]]: |
|
52
|
|
|
""" Return an iterator over (depth, arg_name).""" |
|
53
|
|
|
yield from _depth_items(crumb_path, _fld_idx) |
|
54
|
|
|
|
|
55
|
|
|
|
|
56
|
|
|
def _depth_names_regexes(crumb_path: str) -> Iterator[Tuple[int, str]]: |
|
57
|
|
|
""" Return an iterator over (depth, (arg_name, arg_regex)).""" |
|
58
|
|
|
yield from _depth_items(crumb_path, slice(_fld_idx, _cnv_idx)) |
|
59
|
|
|
|
|
60
|
|
|
|
|
61
|
|
|
def _build_path( |
|
62
|
|
|
crumb_path: str, |
|
63
|
|
|
arg_values: Dict[str, str], |
|
64
|
|
|
with_regex: bool=True, |
|
65
|
|
|
regexes: Dict[str, str]=None |
|
66
|
|
|
) -> str: |
|
67
|
|
|
""" Build the crumb_path with the values in arg_values. |
|
68
|
|
|
Parameters |
|
69
|
|
|
---------- |
|
70
|
|
|
crumb_path: |
|
71
|
|
|
|
|
72
|
|
|
arg_values: |
|
73
|
|
|
arg_name -> arg_value |
|
74
|
|
|
|
|
75
|
|
|
with_regex: |
|
76
|
|
|
|
|
77
|
|
|
regexes: |
|
78
|
|
|
dict[arg_name] -> regex |
|
79
|
|
|
The regexes contained here will replace or be added as a regex for |
|
80
|
|
|
the corresponding arg_name. |
|
81
|
|
|
|
|
82
|
|
|
Returns |
|
83
|
|
|
------- |
|
84
|
|
|
built_path: |
|
85
|
|
|
""" |
|
86
|
|
|
if regexes is None: |
|
87
|
|
|
regexes = {} |
|
88
|
|
|
|
|
89
|
|
|
path = '' |
|
90
|
|
|
for txt, fld, rgx, conv in _yield_items(crumb_path): |
|
91
|
|
|
path += txt |
|
92
|
|
|
if fld is None: |
|
93
|
|
|
continue |
|
94
|
|
|
|
|
95
|
|
|
if fld in arg_values: |
|
96
|
|
|
path += arg_values[fld] |
|
97
|
|
|
else: |
|
98
|
|
|
regex = regexes.get(fld, rgx) if with_regex else '' |
|
99
|
|
|
path += _format_arg(fld, regex=regex) |
|
100
|
|
|
|
|
101
|
|
|
return path |
|
102
|
|
|
|
|
103
|
|
|
|
|
104
|
|
|
def is_valid(crumb_path: str) -> bool: |
|
105
|
|
|
""" Return True if `crumb_path` is a valid Crumb value, False otherwise. """ |
|
106
|
|
|
try: |
|
107
|
|
|
list(_depth_names_regexes(crumb_path)) |
|
108
|
|
|
except ValueError: |
|
109
|
|
|
return False |
|
110
|
|
|
else: |
|
111
|
|
|
return True |
|
112
|
|
|
|
|
113
|
|
|
|
|
114
|
|
|
def _first_txt(crumb_path: str) -> str: |
|
115
|
|
|
""" Return the first text part without arguments in `crumb_path`. """ |
|
116
|
|
|
for txt in _yield_items(crumb_path, index=_txt_idx): |
|
117
|
|
|
return txt |
|
118
|
|
|
|
|
119
|
|
|
|
|
120
|
|
|
def _find_arg_depth(crumb_path: str, arg_name: str) -> Tuple[int, str, str]: |
|
121
|
|
|
""" Return the depth, name and regex of the argument with name `arg_name`. |
|
122
|
|
|
""" |
|
123
|
|
|
for depth, (txt, fld, rgx, conv) in _depth_items(crumb_path): |
|
124
|
|
|
if fld == arg_name: |
|
125
|
|
|
return depth, fld, rgx |
|
126
|
|
|
|
|
127
|
|
|
|
|
128
|
|
|
def _has_arg(crumb_path: str, arg_name: str) -> bool: |
|
129
|
|
|
""" Return the True if the `arg_name` is found in `crumb_path`. """ |
|
130
|
|
|
for txt, fld, rgx, conv in _yield_items(crumb_path): |
|
131
|
|
|
if fld == arg_name: |
|
132
|
|
|
return True |
|
133
|
|
|
return False |
|
134
|
|
|
|
|
135
|
|
|
|
|
136
|
|
|
def _check(crumb_path: str) -> str: |
|
137
|
|
|
""" Raises some Errors if there is something wrong with `crumb_path`, if |
|
138
|
|
|
not the type needed or is not valid. |
|
139
|
|
|
Parameters |
|
140
|
|
|
---------- |
|
141
|
|
|
crumb_path: str |
|
142
|
|
|
|
|
143
|
|
|
Raises |
|
144
|
|
|
------ |
|
145
|
|
|
- ValueError if the path of the Crumb has errors using `self.is_valid`. |
|
146
|
|
|
- TypeError if the crumb_path is not a str or a Crumb. |
|
147
|
|
|
""" |
|
148
|
|
|
if not isinstance(crumb_path, str): |
|
149
|
|
|
raise TypeError("Expected `crumb_path` to be a {}, " |
|
150
|
|
|
"got {}.".format(str, type(crumb_path))) |
|
151
|
|
|
|
|
152
|
|
|
if not is_valid(crumb_path): |
|
153
|
|
|
raise ValueError("The current crumb path has errors, " |
|
154
|
|
|
"got {}.".format(crumb_path)) |
|
155
|
|
|
|
|
156
|
|
|
return crumb_path |
|
157
|
|
|
|
|
158
|
|
|
|
|
159
|
|
|
def _get_path(crumb_path: str) -> str: |
|
160
|
|
|
""" Return the path string from `crumb_path`. |
|
161
|
|
|
Parameters |
|
162
|
|
|
---------- |
|
163
|
|
|
crumb_path: str or Crumb |
|
164
|
|
|
|
|
165
|
|
|
Returns |
|
166
|
|
|
------- |
|
167
|
|
|
path: str |
|
168
|
|
|
""" |
|
169
|
|
|
if hasattr(crumb_path, '_path'): |
|
170
|
|
|
crumb_path = crumb_path._path |
|
171
|
|
|
|
|
172
|
|
|
if not isinstance(crumb_path, str): |
|
173
|
|
|
raise TypeError( |
|
174
|
|
|
"Expected `crumb_path` to be a string, got {}.".format(type(crumb_path)) |
|
175
|
|
|
) |
|
176
|
|
|
|
|
177
|
|
|
return crumb_path |
|
178
|
|
|
|
|
179
|
|
|
|
|
180
|
|
|
def _is_crumb_arg(crumb_arg: str) -> bool: |
|
181
|
|
|
""" Return True if `crumb_arg` is a well formed crumb argument, i.e., |
|
182
|
|
|
is a string that starts with `start_sym` and ends with `end_sym`. |
|
183
|
|
|
False otherwise. |
|
184
|
|
|
""" |
|
185
|
|
|
if not isinstance(crumb_arg, str): |
|
186
|
|
|
return False |
|
187
|
|
|
start_sym, end_sym = ('{', '}') |
|
188
|
|
|
return crumb_arg.startswith(start_sym) and crumb_arg.endswith(end_sym) |
|
189
|
|
|
|
|
190
|
|
|
|
|
191
|
|
|
def _format_arg(arg_name: str, regex: str='') -> str: |
|
192
|
|
|
""" Return the crumb argument for its string `format()` representation. """ |
|
193
|
|
|
start_sym, end_sym = ('{', '}') |
|
194
|
|
|
reg_sym = ':' |
|
195
|
|
|
|
|
196
|
|
|
arg_fmt = start_sym + arg_name |
|
197
|
|
|
if regex: |
|
198
|
|
|
arg_fmt += reg_sym + regex |
|
199
|
|
|
arg_fmt += end_sym |
|
200
|
|
|
|
|
201
|
|
|
return arg_fmt |
|
202
|
|
|
|
|
203
|
|
|
|
|
204
|
|
|
def has_crumbs(crumb_path: str) -> bool: |
|
205
|
|
|
""" Return True if the `crumb_path.split(os.path.sep)` has item which is a |
|
206
|
|
|
crumb argument that starts with '{' and ends with '}'.""" |
|
207
|
|
|
crumb_path = _get_path(crumb_path) |
|
208
|
|
|
|
|
209
|
|
|
splt = crumb_path.split(os.path.sep) |
|
210
|
|
|
for i in splt: |
|
211
|
|
|
if _is_crumb_arg(i): |
|
212
|
|
|
return True |
|
213
|
|
|
|
|
214
|
|
|
return False |
|
215
|
|
|
|
|
216
|
|
|
|
|
217
|
|
|
def _split(crumb_path: str) -> Tuple[str, str]: |
|
218
|
|
|
""" Split `crumb_path` in two parts, the first is the base folder without |
|
219
|
|
|
any crumb argument and the second is the rest of `crumb_path` beginning |
|
220
|
|
|
with the first crumb argument. |
|
221
|
|
|
If `crumb_path` starts with an argument, will return ('', crumb_path). |
|
222
|
|
|
""" |
|
223
|
|
|
crumb_path = _get_path(crumb_path) |
|
224
|
|
|
|
|
225
|
|
|
if not has_crumbs(crumb_path): |
|
226
|
|
|
return crumb_path, '' |
|
227
|
|
|
|
|
228
|
|
|
if not is_valid(crumb_path): |
|
229
|
|
|
raise ValueError('Crumb path {} is not valid.'.format(crumb_path)) |
|
230
|
|
|
|
|
231
|
|
|
start_sym = '{' |
|
232
|
|
|
if crumb_path.startswith(start_sym): |
|
233
|
|
|
base = '' |
|
234
|
|
|
rest = crumb_path |
|
235
|
|
|
else: |
|
236
|
|
|
idx = crumb_path.find(start_sym) |
|
237
|
|
|
base = crumb_path[0:idx] |
|
238
|
|
|
if base.endswith(os.path.sep): |
|
239
|
|
|
base = base[:-1] |
|
240
|
|
|
|
|
241
|
|
|
rest = crumb_path[idx:] |
|
242
|
|
|
|
|
243
|
|
|
return base, rest |
|
244
|
|
|
|
|
245
|
|
|
|
|
246
|
|
|
def _touch(crumb_path: str, exist_ok: bool=True) -> str: |
|
247
|
|
|
""" Create a leaf directory and all intermediate ones |
|
248
|
|
|
using the non crumbed part of `crumb_path`. |
|
249
|
|
|
If the target directory already exists, raise an IOError |
|
250
|
|
|
if exist_ok is False. Otherwise no exception is raised. |
|
251
|
|
|
Parameters |
|
252
|
|
|
---------- |
|
253
|
|
|
crumb_path: |
|
254
|
|
|
|
|
255
|
|
|
exist_ok: |
|
256
|
|
|
Default = True |
|
257
|
|
|
|
|
258
|
|
|
Returns |
|
259
|
|
|
------- |
|
260
|
|
|
nupath: |
|
261
|
|
|
The new path created. |
|
262
|
|
|
""" |
|
263
|
|
|
if has_crumbs(crumb_path): |
|
264
|
|
|
nupath = _split(crumb_path)[0] |
|
265
|
|
|
else: |
|
266
|
|
|
nupath = crumb_path |
|
267
|
|
|
|
|
268
|
|
|
os.makedirs(nupath, exist_ok=exist_ok) |
|
269
|
|
|
return nupath |
|
270
|
|
|
|
|
271
|
|
|
|
|
272
|
|
|
def _split_exists(crumb_path: str) -> bool: |
|
273
|
|
|
""" Return True if the part without crumb arguments of `crumb_path` |
|
274
|
|
|
is an existing path or a symlink, False otherwise. |
|
275
|
|
|
""" |
|
276
|
|
|
if has_crumbs(crumb_path): |
|
277
|
|
|
rpath = _split(crumb_path)[0] |
|
278
|
|
|
else: |
|
279
|
|
|
rpath = str(crumb_path) |
|
280
|
|
|
|
|
281
|
|
|
return os.path.exists(rpath) or os.path.islink(rpath) |
|
282
|
|
|
|
|
283
|
|
|
|
|
284
|
|
|
def _check_is_subset(list1: Iterable[str], list2: Iterable[str]): |
|
285
|
|
|
""" Raise an error if `list1` is not a subset of `list2`.""" |
|
286
|
|
|
if not set(list1).issubset(set(list2)): |
|
287
|
|
|
raise KeyError( |
|
288
|
|
|
'The `list1` argument should be a subset of `list2` ' |
|
289
|
|
|
'got {} and {}.'.format(list1, list2) |
|
290
|
|
|
) |
|
291
|
|
|
|