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