1
|
|
|
# -*- coding: utf-8 -*- |
2
|
|
|
"""Collection of functions to coerce conversion of types with an intelligent guess.""" |
3
|
|
|
from __future__ import absolute_import, division, print_function |
4
|
|
|
|
5
|
|
|
from collections import Mapping |
6
|
|
|
from itertools import chain |
7
|
|
|
from re import IGNORECASE, compile |
8
|
|
|
|
9
|
|
|
from .compat import NoneType, integer_types, isiterable, iteritems, string_types, text_type |
10
|
|
|
from .decorators import memoize, memoizeproperty |
11
|
|
|
from .exceptions import AuxlibError |
12
|
|
|
|
13
|
|
|
__all__ = ["boolify", "typify", "maybecall", "listify", "numberify"] |
14
|
|
|
|
15
|
|
|
BOOLISH_TRUE = ("true", "yes", "on", "y") |
16
|
|
|
BOOLISH_FALSE = ("false", "off", "n", "no", "non", "none", "") |
17
|
|
|
NULL_STRINGS = ("none", "~", "null", "\0") |
18
|
|
|
BOOL_COERCEABLE_TYPES = integer_types + (bool, float, complex, list, set, dict, tuple) |
19
|
|
|
NUMBER_TYPES = integer_types + (float, complex) |
20
|
|
|
NUMBER_TYPES_SET = frozenset(NUMBER_TYPES) |
21
|
|
|
STRING_TYPES_SET = frozenset(string_types) |
22
|
|
|
BOOLNULL_TYPE_SET = frozenset([bool, NoneType]) |
23
|
|
|
BOOLSTRING_TYPES_SET = STRING_TYPES_SET | frozenset([bool]) |
24
|
|
|
|
25
|
|
|
NO_MATCH = object() |
26
|
|
|
|
27
|
|
|
|
28
|
|
|
class TypeCoercionError(AuxlibError, ValueError): |
29
|
|
|
|
30
|
|
|
def __init__(self, value, msg, *args, **kwargs): |
31
|
|
|
self.value = value |
32
|
|
|
super(TypeCoercionError, self).__init__(msg, *args, **kwargs) |
33
|
|
|
|
34
|
|
|
|
35
|
|
|
class _Regex(object): |
36
|
|
|
|
37
|
|
|
@memoizeproperty |
38
|
|
|
def BOOLEAN_TRUE(self): |
39
|
|
|
return compile(r'^true$|^yes$|^on$', IGNORECASE), True |
40
|
|
|
|
41
|
|
|
@memoizeproperty |
42
|
|
|
def BOOLEAN_FALSE(self): |
43
|
|
|
return compile(r'^false$|^no$|^off$', IGNORECASE), False |
44
|
|
|
|
45
|
|
|
@memoizeproperty |
46
|
|
|
def NONE(self): |
47
|
|
|
return compile(r'^none$|^null$', IGNORECASE), None |
48
|
|
|
|
49
|
|
|
@memoizeproperty |
50
|
|
|
def INT(self): |
51
|
|
|
return compile(r'^[-+]?\d+$'), int |
52
|
|
|
|
53
|
|
|
@memoizeproperty |
54
|
|
|
def BIN(self): |
55
|
|
|
return compile(r'^[-+]?0[bB][01]+$'), bin |
56
|
|
|
|
57
|
|
|
@memoizeproperty |
58
|
|
|
def OCT(self): |
59
|
|
|
return compile(r'^[-+]?0[oO][0-7]+$'), oct |
60
|
|
|
|
61
|
|
|
@memoizeproperty |
62
|
|
|
def HEX(self): |
63
|
|
|
return compile(r'^[-+]?0[xX][0-9a-fA-F]+$'), hex |
64
|
|
|
|
65
|
|
|
@memoizeproperty |
66
|
|
|
def FLOAT(self): |
67
|
|
|
return compile(r'^[-+]?(\d+(\.\d*)?|\.\d+)([eE][-+]?\d+)?$'), float |
68
|
|
|
|
69
|
|
|
@memoizeproperty |
70
|
|
|
def COMPLEX(self): |
71
|
|
|
return (compile(r'^(?:[-+]?(\d+(\.\d*)?|\.\d+)([eE][-+]?\d+)?)?' # maybe first float |
72
|
|
|
r'[-+]?(\d+(\.\d*)?|\.\d+)([eE][-+]?\d+)?j$'), # second float with j |
73
|
|
|
complex) |
74
|
|
|
|
75
|
|
|
@property |
76
|
|
|
def numbers(self): |
77
|
|
|
yield self.INT |
78
|
|
|
yield self.FLOAT |
79
|
|
|
yield self.BIN |
80
|
|
|
yield self.OCT |
81
|
|
|
yield self.HEX |
82
|
|
|
yield self.COMPLEX |
83
|
|
|
|
84
|
|
|
@property |
85
|
|
|
def boolean(self): |
86
|
|
|
yield self.BOOLEAN_TRUE |
87
|
|
|
yield self.BOOLEAN_FALSE |
88
|
|
|
|
89
|
|
|
@property |
90
|
|
|
def none(self): |
91
|
|
|
yield self.NONE |
92
|
|
|
|
93
|
|
|
def convert_number(self, value_string): |
94
|
|
|
return self._convert(value_string, (self.numbers, )) |
95
|
|
|
|
96
|
|
|
def convert(self, value_string): |
97
|
|
|
return self._convert(value_string, (self.boolean, self.none, self.numbers, )) |
98
|
|
|
|
99
|
|
|
def _convert(self, value_string, type_list): |
100
|
|
|
return next((typish(value_string) if callable(typish) else typish |
101
|
|
|
for regex, typish in chain.from_iterable(type_list) |
102
|
|
|
if regex.match(value_string)), |
103
|
|
|
NO_MATCH) |
104
|
|
|
|
105
|
|
|
_REGEX = _Regex() |
106
|
|
|
|
107
|
|
|
|
108
|
|
|
def numberify(value): |
109
|
|
|
""" |
110
|
|
|
|
111
|
|
|
Examples: |
112
|
|
|
>>> [numberify(x) for x in ('1234', 1234, '0755', 0o0755, False, 0, '0', True, 1, '1')] |
113
|
|
|
[1234, 1234, 755, 493, 0, 0, 0, 1, 1, 1] |
114
|
|
|
>>> [numberify(x) for x in ('12.34', 12.34, 1.2+3.5j, '1.2+3.5j')] |
115
|
|
|
[12.34, 12.34, (1.2+3.5j), (1.2+3.5j)] |
116
|
|
|
|
117
|
|
|
""" |
118
|
|
|
if isinstance(value, bool): |
119
|
|
|
return int(value) |
120
|
|
|
if isinstance(value, NUMBER_TYPES): |
121
|
|
|
return value |
122
|
|
|
candidate = _REGEX.convert_number(value) |
123
|
|
|
if candidate is not NO_MATCH: |
124
|
|
|
return candidate |
125
|
|
|
raise TypeCoercionError(value, "Cannot convert {0} to a number.".format(value)) |
126
|
|
|
|
127
|
|
|
|
128
|
|
|
def boolify(value, nullable=False, return_string=False): |
129
|
|
|
"""Convert a number, string, or sequence type into a pure boolean. |
130
|
|
|
|
131
|
|
|
Args: |
132
|
|
|
value (number, string, sequence): pretty much anything |
133
|
|
|
|
134
|
|
|
Returns: |
135
|
|
|
bool: boolean representation of the given value |
136
|
|
|
|
137
|
|
|
Examples: |
138
|
|
|
>>> [boolify(x) for x in ('yes', 'no')] |
139
|
|
|
[True, False] |
140
|
|
|
>>> [boolify(x) for x in (0.1, 0+0j, True, '0', '0.0', '0.1', '2')] |
141
|
|
|
[True, False, True, False, False, True, True] |
142
|
|
|
>>> [boolify(x) for x in ("true", "yes", "on", "y")] |
143
|
|
|
[True, True, True, True] |
144
|
|
|
>>> [boolify(x) for x in ("no", "non", "none", "off", "")] |
145
|
|
|
[False, False, False, False, False] |
146
|
|
|
>>> [boolify(x) for x in ([], set(), dict(), tuple())] |
147
|
|
|
[False, False, False, False] |
148
|
|
|
>>> [boolify(x) for x in ([1], set([False]), dict({'a': 1}), tuple([2]))] |
149
|
|
|
[True, True, True, True] |
150
|
|
|
|
151
|
|
|
""" |
152
|
|
|
# cast number types naturally |
153
|
|
|
if isinstance(value, BOOL_COERCEABLE_TYPES): |
154
|
|
|
return bool(value) |
155
|
|
|
# try to coerce string into number |
156
|
|
|
val = text_type(value).strip().lower().replace('.', '', 1) |
157
|
|
|
if val.isnumeric(): |
158
|
|
|
return bool(float(val)) |
159
|
|
|
elif val in BOOLISH_TRUE: |
160
|
|
|
return True |
161
|
|
|
elif nullable and val in NULL_STRINGS: |
162
|
|
|
return None |
163
|
|
|
elif val in BOOLISH_FALSE: |
164
|
|
|
return False |
165
|
|
|
else: # must be False |
166
|
|
|
try: |
167
|
|
|
return bool(complex(val)) |
168
|
|
|
except ValueError: |
169
|
|
|
if isinstance(value, string_types) and return_string: |
170
|
|
|
return value |
171
|
|
|
raise TypeCoercionError(value, "The value %r cannot be boolified." % value) |
172
|
|
|
|
173
|
|
|
|
174
|
|
|
def boolify_truthy_string_ok(value): |
175
|
|
|
try: |
176
|
|
|
return boolify(value) |
177
|
|
|
except ValueError: |
178
|
|
|
assert isinstance(value, string_types), repr(value) |
179
|
|
|
return True |
180
|
|
|
|
181
|
|
|
|
182
|
|
|
@memoize |
183
|
|
|
def typify(value, type_hint=None): |
184
|
|
|
"""Take a primitive value, usually a string, and try to make a more relevant type out of it. |
185
|
|
|
An optional type_hint will try to coerce the value to that type. |
186
|
|
|
|
187
|
|
|
Args: |
188
|
|
|
value (Any): Usually a string, not a sequence |
189
|
|
|
type_hint (type or Tuple[type]): |
190
|
|
|
|
191
|
|
|
Examples: |
192
|
|
|
>>> typify('32') |
193
|
|
|
32 |
194
|
|
|
>>> typify('32', float) |
195
|
|
|
32.0 |
196
|
|
|
>>> typify('32.0') |
197
|
|
|
32.0 |
198
|
|
|
>>> typify('32.0.0') |
199
|
|
|
'32.0.0' |
200
|
|
|
>>> [typify(x) for x in ('true', 'yes', 'on')] |
201
|
|
|
[True, True, True] |
202
|
|
|
>>> [typify(x) for x in ('no', 'FALSe', 'off')] |
203
|
|
|
[False, False, False] |
204
|
|
|
>>> [typify(x) for x in ('none', 'None', None)] |
205
|
|
|
[None, None, None] |
206
|
|
|
|
207
|
|
|
""" |
208
|
|
|
# value must be a string, or there at least needs to be a type hint |
209
|
|
|
if isinstance(value, string_types): |
210
|
|
|
value = value.strip() |
211
|
|
|
elif type_hint is None: |
212
|
|
|
# can't do anything because value isn't a string and there's no type hint |
213
|
|
|
return value |
214
|
|
|
|
215
|
|
|
# now we either have a stripped string, a type hint, or both |
216
|
|
|
# use the hint if it exists |
217
|
|
|
if type_hint is None: |
218
|
|
|
# no type hint, but we know value is a string, so try to match with the regex patterns |
219
|
|
|
return _regex_typify_string(value) |
220
|
|
|
else: |
221
|
|
|
return _typify_with_hint(value, type_hint) |
222
|
|
|
|
223
|
|
|
|
224
|
|
|
def _typify_with_hint(value, type_hint): |
225
|
|
|
if isiterable(type_hint): |
226
|
|
|
return _typify_with_iterable_hint(value, type_hint) |
227
|
|
|
else: |
228
|
|
|
# coerce using the type hint, or use boolify for bool |
229
|
|
|
try: |
230
|
|
|
return boolify(value) if type_hint == bool else type_hint(value) |
231
|
|
|
except ValueError as e: |
232
|
|
|
# ValueError: invalid literal for int() with base 10: 'nope' |
233
|
|
|
raise TypeCoercionError(value, text_type(e)) |
234
|
|
|
|
235
|
|
|
|
236
|
|
|
def _typify_with_iterable_hint(value, type_hint): |
237
|
|
|
type_hint = frozenset(type_hint) |
238
|
|
|
if not (type_hint - NUMBER_TYPES_SET): |
239
|
|
|
return numberify(value) |
240
|
|
|
elif not (type_hint - STRING_TYPES_SET): |
241
|
|
|
return text_type(value) |
242
|
|
|
elif not (type_hint - BOOLNULL_TYPE_SET): |
243
|
|
|
return boolify(value, nullable=True) |
244
|
|
|
elif not (type_hint - BOOLSTRING_TYPES_SET): |
245
|
|
|
return boolify(value, return_string=True) |
246
|
|
|
else: |
247
|
|
|
raise NotImplementedError() |
248
|
|
|
|
249
|
|
|
|
250
|
|
|
def _regex_typify_string(string): |
251
|
|
|
candidate = _REGEX.convert(string) |
252
|
|
|
if candidate is not NO_MATCH: |
253
|
|
|
return candidate |
254
|
|
|
else: |
255
|
|
|
# nothing has caught so far; give up, and return the value that was given |
256
|
|
|
return string |
257
|
|
|
|
258
|
|
|
|
259
|
|
|
def typify_data_structure(value, type_hint=None): |
260
|
|
|
if isinstance(value, Mapping): |
261
|
|
|
return type(value)((k, typify(v, type_hint)) for k, v in iteritems(value)) |
262
|
|
|
elif isiterable(value): |
263
|
|
|
return type(value)(typify(v, type_hint) for v in value) |
264
|
|
|
else: |
265
|
|
|
return typify(value, type_hint) |
266
|
|
|
|
267
|
|
|
|
268
|
|
|
def maybecall(value): |
269
|
|
|
return value() if callable(value) else value |
270
|
|
|
|
271
|
|
|
|
272
|
|
|
def listify(val, return_type=tuple): |
273
|
|
|
""" |
274
|
|
|
Examples: |
275
|
|
|
>>> listify('abc', return_type=list) |
276
|
|
|
['abc'] |
277
|
|
|
>>> listify(None) |
278
|
|
|
() |
279
|
|
|
>>> listify(False) |
280
|
|
|
(False,) |
281
|
|
|
>>> listify(('a', 'b', 'c'), return_type=list) |
282
|
|
|
['a', 'b', 'c'] |
283
|
|
|
""" |
284
|
|
|
# TODO: flatlistify((1, 2, 3), 4, (5, 6, 7)) |
285
|
|
|
if val is None: |
286
|
|
|
return return_type() |
287
|
|
|
elif isiterable(val): |
288
|
|
|
return return_type(val) |
289
|
|
|
else: |
290
|
|
|
return return_type((val, )) |
291
|
|
|
|