|
1
|
|
|
# -*- coding: utf-8 -*- |
|
2
|
|
|
|
|
3
|
|
|
from benedict.serializers import JSONSerializer |
|
4
|
|
|
from benedict.utils import type_util |
|
5
|
|
|
|
|
6
|
|
|
from datetime import datetime |
|
7
|
|
|
from dateutil import parser as date_parser |
|
8
|
|
|
from decimal import Decimal, DecimalException |
|
9
|
|
|
from MailChecker import MailChecker |
|
10
|
|
|
from phonenumbers import phonenumberutil, PhoneNumberFormat |
|
11
|
|
|
from six import text_type |
|
12
|
|
|
from slugify import slugify |
|
13
|
|
|
|
|
14
|
|
|
import ftfy |
|
15
|
|
|
import phonenumbers |
|
16
|
|
|
import re |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
def _parse_with(val, type_checker, parser, **kwargs): |
|
20
|
|
|
if val is None: |
|
21
|
|
|
return None |
|
22
|
|
|
if callable(type_checker) and type_checker(val): |
|
23
|
|
|
return val |
|
24
|
|
|
s = text_type(val) |
|
25
|
|
|
if not len(s): |
|
26
|
|
|
return None |
|
27
|
|
|
return parser(s, **kwargs) |
|
28
|
|
|
|
|
29
|
|
|
|
|
30
|
|
|
def _parse_bool(val): |
|
31
|
|
|
val = val.lower() |
|
32
|
|
|
if val in ['1', 'true', 'yes', 'ok', 'on']: |
|
33
|
|
|
return True |
|
34
|
|
|
elif val in ['0', 'false', 'no', 'ko', 'off']: |
|
35
|
|
|
return False |
|
36
|
|
|
return None |
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
def parse_bool(val): |
|
40
|
|
|
return _parse_with(val, type_util.is_bool, _parse_bool) |
|
41
|
|
|
|
|
42
|
|
|
|
|
43
|
|
|
def parse_date(val, format=None): |
|
44
|
|
|
val = parse_datetime(val, format) |
|
45
|
|
|
if val: |
|
46
|
|
|
return val.date() |
|
47
|
|
|
return None |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
def _parse_datetime_with_format(val, format): |
|
51
|
|
|
try: |
|
52
|
|
|
return datetime.strptime(val, format) |
|
53
|
|
|
except Exception: |
|
54
|
|
|
return None |
|
55
|
|
|
|
|
56
|
|
|
|
|
57
|
|
|
def _parse_datetime_without_format(val): |
|
58
|
|
|
try: |
|
59
|
|
|
return date_parser.parse(val) |
|
60
|
|
|
except Exception: |
|
61
|
|
|
return _parse_datetime_from_timestamp(val) |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
def _parse_datetime_from_timestamp(val): |
|
65
|
|
|
try: |
|
66
|
|
|
return datetime.fromtimestamp(float(val)) |
|
67
|
|
|
except Exception: |
|
68
|
|
|
return None |
|
69
|
|
|
|
|
70
|
|
|
|
|
71
|
|
|
def parse_datetime(val, format=None): |
|
72
|
|
|
if type_util.is_datetime(val): |
|
73
|
|
|
return val |
|
74
|
|
|
s = text_type(val) |
|
75
|
|
|
if format: |
|
76
|
|
|
return _parse_datetime_with_format(s, format) |
|
77
|
|
|
else: |
|
78
|
|
|
return _parse_datetime_without_format(s) |
|
79
|
|
|
|
|
80
|
|
|
|
|
81
|
|
|
def _parse_decimal(val): |
|
82
|
|
|
try: |
|
83
|
|
|
return Decimal(val) |
|
84
|
|
|
except (ValueError, DecimalException): |
|
85
|
|
|
return None |
|
86
|
|
|
|
|
87
|
|
|
|
|
88
|
|
|
def parse_decimal(val): |
|
89
|
|
|
return _parse_with(val, type_util.is_decimal, _parse_decimal) |
|
90
|
|
|
|
|
91
|
|
|
|
|
92
|
|
|
def _parse_dict(val): |
|
93
|
|
|
serializer = JSONSerializer() |
|
94
|
|
|
try: |
|
95
|
|
|
d = serializer.decode(val) |
|
96
|
|
|
if type_util.is_dict(d): |
|
97
|
|
|
return d |
|
98
|
|
|
return None |
|
99
|
|
|
except Exception: |
|
100
|
|
|
return None |
|
101
|
|
|
|
|
102
|
|
|
|
|
103
|
|
|
def parse_dict(val): |
|
104
|
|
|
return _parse_with(val, type_util.is_dict, _parse_dict) |
|
105
|
|
|
|
|
106
|
|
|
|
|
107
|
|
|
def _parse_float(val): |
|
108
|
|
|
try: |
|
109
|
|
|
return float(val) |
|
110
|
|
|
except ValueError: |
|
111
|
|
|
return None |
|
112
|
|
|
|
|
113
|
|
|
|
|
114
|
|
|
def parse_float(val): |
|
115
|
|
|
return _parse_with(val, type_util.is_float, _parse_float) |
|
116
|
|
|
|
|
117
|
|
|
|
|
118
|
|
|
def _parse_email(val, check_blacklist=True): |
|
119
|
|
|
val = val.lower() |
|
120
|
|
|
if check_blacklist: |
|
121
|
|
|
if not MailChecker.is_valid(val): |
|
122
|
|
|
return None |
|
123
|
|
|
else: |
|
124
|
|
|
if not MailChecker.is_valid_email_format(val): |
|
125
|
|
|
return None |
|
126
|
|
|
return val |
|
127
|
|
|
|
|
128
|
|
|
|
|
129
|
|
|
def parse_email(val, check_blacklist=True): |
|
130
|
|
|
return _parse_with( |
|
131
|
|
|
val, None, _parse_email, check_blacklist=check_blacklist) |
|
132
|
|
|
|
|
133
|
|
|
|
|
134
|
|
|
def _parse_int(val): |
|
135
|
|
|
try: |
|
136
|
|
|
return int(val) |
|
137
|
|
|
except ValueError: |
|
138
|
|
|
return None |
|
139
|
|
|
|
|
140
|
|
|
|
|
141
|
|
|
def parse_int(val): |
|
142
|
|
|
return _parse_with(val, type_util.is_integer, _parse_int) |
|
143
|
|
|
|
|
144
|
|
|
|
|
145
|
|
|
def _parse_list(val, separator=None): |
|
146
|
|
|
if val.startswith('{') and val.endswith('}') or val.startswith('[') and val.endswith(']'): |
|
147
|
|
|
try: |
|
148
|
|
|
serializer = JSONSerializer() |
|
149
|
|
|
l = serializer.decode(val) |
|
150
|
|
|
if type_util.is_list(l): |
|
151
|
|
|
return l |
|
152
|
|
|
return None |
|
153
|
|
|
except Exception: |
|
154
|
|
|
pass |
|
155
|
|
|
if separator: |
|
156
|
|
|
l = list(val.split(separator)) |
|
157
|
|
|
return l |
|
158
|
|
|
return None |
|
159
|
|
|
|
|
160
|
|
|
|
|
161
|
|
|
def parse_list(val, separator=None): |
|
162
|
|
|
val = _parse_with(val, type_util.is_list_or_tuple, |
|
163
|
|
|
_parse_list, separator=separator) |
|
164
|
|
|
return list(val) if type_util.is_list_or_tuple(val) else val |
|
165
|
|
|
|
|
166
|
|
|
|
|
167
|
|
|
def _parse_phonenumber(val, country_code=None): |
|
168
|
|
|
try: |
|
169
|
|
|
phone_obj = phonenumbers.parse(val, country_code) |
|
170
|
|
|
if phonenumbers.is_valid_number(phone_obj): |
|
171
|
|
|
return { |
|
172
|
|
|
'e164': phonenumbers.format_number( |
|
173
|
|
|
phone_obj, PhoneNumberFormat.E164), |
|
174
|
|
|
'international': phonenumbers.format_number( |
|
175
|
|
|
phone_obj, PhoneNumberFormat.INTERNATIONAL), |
|
176
|
|
|
'national': phonenumbers.format_number( |
|
177
|
|
|
phone_obj, PhoneNumberFormat.NATIONAL), |
|
178
|
|
|
} |
|
179
|
|
|
return None |
|
180
|
|
|
except phonenumberutil.NumberParseException: |
|
181
|
|
|
return None |
|
182
|
|
|
|
|
183
|
|
|
|
|
184
|
|
|
def parse_phonenumber(val, country_code=None): |
|
185
|
|
|
s = parse_str(val) |
|
186
|
|
|
if not s: |
|
187
|
|
|
return None |
|
188
|
|
|
phone_raw = re.sub(r'[^0-9\+]', ' ', s) |
|
189
|
|
|
phone_raw = phone_raw.strip() |
|
190
|
|
|
if phone_raw.startswith('00'): |
|
191
|
|
|
phone_raw = '+{}'.format(phone_raw[2:]) |
|
192
|
|
|
if country_code and len(country_code) >= 2: |
|
193
|
|
|
country_code = country_code[0:2].upper() |
|
194
|
|
|
return _parse_with( |
|
195
|
|
|
phone_raw, None, _parse_phonenumber, country_code=country_code) |
|
196
|
|
|
|
|
197
|
|
|
|
|
198
|
|
|
def _parse_slug(val): |
|
199
|
|
|
return slugify(val) |
|
200
|
|
|
|
|
201
|
|
|
|
|
202
|
|
|
def parse_slug(val): |
|
203
|
|
|
s = parse_str(val) |
|
204
|
|
|
return _parse_slug(s) |
|
205
|
|
|
|
|
206
|
|
|
|
|
207
|
|
|
def parse_str(val): |
|
208
|
|
|
if type_util.is_string(val): |
|
209
|
|
|
try: |
|
210
|
|
|
val = ftfy.fix_text(val) |
|
211
|
|
|
except UnicodeError: |
|
212
|
|
|
pass |
|
213
|
|
|
else: |
|
214
|
|
|
val = text_type(val) |
|
215
|
|
|
val = val.strip() |
|
216
|
|
|
val = ' '.join(val.split()) |
|
217
|
|
|
return val |
|
218
|
|
|
|
|
219
|
|
|
|
|
220
|
|
|
def parse_uuid(val): |
|
221
|
|
|
s = parse_str(val) |
|
222
|
|
|
return s if type_util.is_uuid(s) else None |
|
223
|
|
|
|