]> git.openstreetmap.org Git - nominatim.git/blob - test/python/tokenizer/sanitizers/test_clean_postcodes.py
Merge pull request #4130 from lonvia/avoid-creating-word-lookup-indexes
[nominatim.git] / test / python / tokenizer / sanitizers / test_clean_postcodes.py
1 # SPDX-License-Identifier: GPL-3.0-or-later
2 #
3 # This file is part of Nominatim. (https://nominatim.org)
4 #
5 # Copyright (C) 2026 by the Nominatim developer community.
6 # For a full list of authors see the git log.
7 """
8 Tests for the sanitizer that normalizes postcodes.
9 """
10 import pytest
11
12 from nominatim_db.tokenizer.place_sanitizer import PlaceSanitizer
13 from nominatim_db.data.place_info import PlaceInfo
14 from nominatim_db.data import country_info
15
16
17 @pytest.fixture
18 def sanitize(def_config, request):
19     country_info.setup_country_config(def_config)
20     sanitizer_args = {'step': 'clean-postcodes'}
21     for mark in request.node.iter_markers(name="sanitizer_params"):
22         sanitizer_args.update({k.replace('_', '-'): v for k, v in mark.kwargs.items()})
23
24     def _run(country=None, **kwargs):
25         pi = {'address': kwargs}
26         if country is not None:
27             pi['country_code'] = country
28
29         place = PlaceInfo(pi)
30         PlaceSanitizer([sanitizer_args], def_config).process_names(place)
31
32         return sorted([(p.kind, p.name) for p in place.searchable_address])
33
34     return _run
35
36
37 @pytest.mark.parametrize("country", (None, 'ae'))
38 def test_postcode_no_country(sanitize, country):
39     assert sanitize(country=country, postcode='23231') == [('unofficial_postcode', '23231')]
40
41
42 @pytest.mark.parametrize("country", (None, 'ae'))
43 @pytest.mark.sanitizer_params(convert_to_address=False)
44 def test_postcode_no_country_drop(sanitize, country):
45     assert sanitize(country=country, postcode='23231') == []
46
47
48 @pytest.mark.parametrize("postcode", ('12345', '  12345  ', 'de 12345',
49                                       'DE12345', 'DE 12345', 'DE-12345'))
50 def test_postcode_pass_good_format(sanitize, postcode):
51     assert sanitize(country='de', postcode=postcode) == [('postcode', '12345')]
52
53
54 @pytest.mark.parametrize("postcode", ('123456', '', '   ', '.....',
55                                       'DE  12345', 'DEF12345', 'CH 12345'))
56 @pytest.mark.sanitizer_params(convert_to_address=False)
57 def test_postcode_drop_bad_format(sanitize, postcode):
58     assert sanitize(country='de', postcode=postcode) == []
59
60
61 @pytest.mark.parametrize("postcode", ('1234', '9435', '99000'))
62 def test_postcode_cyprus_pass(sanitize, postcode):
63     assert sanitize(country='cy', postcode=postcode) == [('postcode', postcode)]
64
65
66 @pytest.mark.parametrize("postcode", ('91234', '99a45', '567'))
67 @pytest.mark.sanitizer_params(convert_to_address=False)
68 def test_postcode_cyprus_fail(sanitize, postcode):
69     assert sanitize(country='cy', postcode=postcode) == []
70
71
72 @pytest.mark.parametrize("postcode", ('123456', 'A33F2G7'))
73 def test_postcode_kazakhstan_pass(sanitize, postcode):
74     assert sanitize(country='kz', postcode=postcode) == [('postcode', postcode)]
75
76
77 @pytest.mark.parametrize("postcode", ('V34T6Y923456', '99345'))
78 @pytest.mark.sanitizer_params(convert_to_address=False)
79 def test_postcode_kazakhstan_fail(sanitize, postcode):
80     assert sanitize(country='kz', postcode=postcode) == []
81
82
83 @pytest.mark.parametrize("postcode", ('675 34', '67534', 'SE-675 34', 'SE67534'))
84 def test_postcode_sweden_pass(sanitize, postcode):
85     assert sanitize(country='se', postcode=postcode) == [('postcode', '675 34')]
86
87
88 @pytest.mark.parametrize("postcode", ('67 345', '671123'))
89 @pytest.mark.sanitizer_params(convert_to_address=False)
90 def test_postcode_sweden_fail(sanitize, postcode):
91     assert sanitize(country='se', postcode=postcode) == []
92
93
94 @pytest.mark.parametrize("postcode", ('AD123', '123', 'AD 123', 'AD-123'))
95 def test_postcode_andorra_pass(sanitize, postcode):
96     assert sanitize(country='ad', postcode=postcode) == [('postcode', 'AD123')]
97
98
99 @pytest.mark.parametrize("postcode", ('AD1234', 'AD AD123', 'XX123'))
100 @pytest.mark.sanitizer_params(convert_to_address=False)
101 def test_postcode_andorra_fail(sanitize, postcode):
102     assert sanitize(country='ad', postcode=postcode) == []
103
104
105 @pytest.mark.parametrize("postcode", ('AI-2640', '2640', 'AI 2640'))
106 def test_postcode_anguilla_pass(sanitize, postcode):
107     assert sanitize(country='ai', postcode=postcode) == [('postcode', 'AI-2640')]
108
109
110 @pytest.mark.parametrize("postcode", ('AI-2000', 'AI US-2640', 'AI AI-2640'))
111 @pytest.mark.sanitizer_params(convert_to_address=False)
112 def test_postcode_anguilla_fail(sanitize, postcode):
113     assert sanitize(country='ai', postcode=postcode) == []
114
115
116 @pytest.mark.parametrize("postcode", ('BN1111', 'BN 1111', 'BN BN1111', 'BN BN 1111'))
117 def test_postcode_brunei_pass(sanitize, postcode):
118     assert sanitize(country='bn', postcode=postcode) == [('postcode', 'BN1111')]
119
120
121 @pytest.mark.parametrize("postcode", ('BN-1111', 'BNN1111'))
122 @pytest.mark.sanitizer_params(convert_to_address=False)
123 def test_postcode_brunei_fail(sanitize, postcode):
124     assert sanitize(country='bn', postcode=postcode) == []
125
126
127 @pytest.mark.parametrize("postcode", ('IM1 1AA', 'IM11AA', 'IM IM11AA'))
128 def test_postcode_isle_of_man_pass(sanitize, postcode):
129     assert sanitize(country='im', postcode=postcode) == [('postcode', 'IM1 1AA')]
130
131
132 @pytest.mark.parametrize("postcode", ('IZ1 1AA', 'IM1 AA'))
133 @pytest.mark.sanitizer_params(convert_to_address=False)
134 def test_postcode_isle_of_man_fail(sanitize, postcode):
135     assert sanitize(country='im', postcode=postcode) == []
136
137
138 @pytest.mark.parametrize("postcode", ('JE5 0LA', 'JE50LA', 'JE JE50LA', 'je JE5 0LA'))
139 def test_postcode_jersey_pass(sanitize, postcode):
140     assert sanitize(country='je', postcode=postcode) == [('postcode', 'JE5 0LA')]
141
142
143 @pytest.mark.parametrize("postcode", ('gb JE5 0LA', 'IM50LA', 'IM5 012'))
144 @pytest.mark.sanitizer_params(convert_to_address=False)
145 def test_postcode_jersey_fail(sanitize, postcode):
146     assert sanitize(country='je', postcode=postcode) == []
147
148
149 @pytest.mark.parametrize("postcode", ('KY1-1234', '1-1234', 'KY 1-1234'))
150 def test_postcode_cayman_islands_pass(sanitize, postcode):
151     assert sanitize(country='ky', postcode=postcode) == [('postcode', 'KY1-1234')]
152
153
154 @pytest.mark.parametrize("postcode", ('KY-1234', 'KZ1-1234', 'KY1 1234', 'KY1-123', 'KY KY1-1234'))
155 @pytest.mark.sanitizer_params(convert_to_address=False)
156 def test_postcode_cayman_islands_fail(sanitize, postcode):
157     assert sanitize(country='ky', postcode=postcode) == []
158
159
160 @pytest.mark.parametrize("postcode", ('LC11 222', '11 222', '11222', 'LC 11 222'))
161 def test_postcode_saint_lucia_pass(sanitize, postcode):
162     assert sanitize(country='lc', postcode=postcode) == [('postcode', 'LC11 222')]
163
164
165 @pytest.mark.parametrize("postcode", ('11 2222', 'LC LC11 222'))
166 @pytest.mark.sanitizer_params(convert_to_address=False)
167 def test_postcode_saint_lucia_fail(sanitize, postcode):
168     assert sanitize(country='lc', postcode=postcode) == []
169
170
171 @pytest.mark.parametrize("postcode", ('LV-1111', '1111', 'LV 1111', 'LV1111',))
172 def test_postcode_latvia_pass(sanitize, postcode):
173     assert sanitize(country='lv', postcode=postcode) == [('postcode', 'LV-1111')]
174
175
176 @pytest.mark.parametrize("postcode", ('111', '11111', 'LV LV-1111'))
177 @pytest.mark.sanitizer_params(convert_to_address=False)
178 def test_postcode_latvia_fail(sanitize, postcode):
179     assert sanitize(country='lv', postcode=postcode) == []
180
181
182 @pytest.mark.parametrize("postcode", ('MD-1111', '1111', 'MD 1111', 'MD1111'))
183 def test_postcode_moldova_pass(sanitize, postcode):
184     assert sanitize(country='md', postcode=postcode) == [('postcode', 'MD-1111')]
185
186
187 @pytest.mark.parametrize("postcode", ("MD MD-1111", "MD MD1111", "MD MD 1111"))
188 @pytest.mark.sanitizer_params(convert_to_address=False)
189 def test_postcode_moldova_fail(sanitize, postcode):
190     assert sanitize(country='md', postcode=postcode) == []
191
192
193 @pytest.mark.parametrize("postcode", ('VLT 1117', 'GDJ 1234', 'BZN 2222'))
194 def test_postcode_malta_pass(sanitize, postcode):
195     assert sanitize(country='mt', postcode=postcode) == [('postcode', postcode)]
196
197
198 @pytest.mark.parametrize("postcode", ('MTF 1111', 'MT MTF 1111', 'MTF1111', 'MT MTF1111'))
199 def test_postcode_malta_mtarfa_pass(sanitize, postcode):
200     assert sanitize(country='mt', postcode=postcode) == [('postcode', 'MTF 1111')]
201
202
203 @pytest.mark.parametrize("postcode", ('1111', 'MTMT 1111'))
204 @pytest.mark.sanitizer_params(convert_to_address=False)
205 def test_postcode_malta_fail(sanitize, postcode):
206     assert sanitize(country='mt', postcode=postcode) == []
207
208
209 @pytest.mark.parametrize("postcode", ('VC1111', '1111', 'VC-1111', 'VC 1111'))
210 def test_postcode_saint_vincent_pass(sanitize, postcode):
211     assert sanitize(country='vc', postcode=postcode) == [('postcode', 'VC1111')]
212
213
214 @pytest.mark.parametrize("postcode", ('VC11', 'VC VC1111'))
215 @pytest.mark.sanitizer_params(convert_to_address=False)
216 def test_postcode_saint_vincent_fail(sanitize, postcode):
217     assert sanitize(country='vc', postcode=postcode) == []
218
219
220 @pytest.mark.parametrize("postcode", ('VG1111', '1111', 'VG 1111', 'VG-1111'))
221 def test_postcode_virgin_islands_pass(sanitize, postcode):
222     assert sanitize(country='vg', postcode=postcode) == [('postcode', 'VG1111')]
223
224
225 @pytest.mark.parametrize("postcode", ('111', '11111', 'VG VG1111'))
226 @pytest.mark.sanitizer_params(convert_to_address=False)
227 def test_postcode_virgin_islands_fail(sanitize, postcode):
228     assert sanitize(country='vg', postcode=postcode) == []
229
230
231 @pytest.mark.parametrize("postcode", ('AB1', '123-456-7890', '1 as 44'))
232 @pytest.mark.sanitizer_params(default_pattern='[A-Z0-9- ]{3,12}')
233 def test_postcode_default_pattern_pass(sanitize, postcode):
234     assert sanitize(country='an', postcode=postcode) == [('postcode', postcode.upper())]
235
236
237 @pytest.mark.parametrize("postcode", ('C', '12', 'ABC123DEF 456', '1234,5678', '11223;11224'))
238 @pytest.mark.sanitizer_params(convert_to_address=False, default_pattern='[A-Z0-9- ]{3,12}')
239 def test_postcode_default_pattern_fail(sanitize, postcode):
240     assert sanitize(country='an', postcode=postcode) == []
241
242
243 @pytest.mark.parametrize("postcode", ('00000', '00-000', 'PL-00000', 'PL 00-000'))
244 @pytest.mark.sanitizer_params(convert_to_address=False)
245 def test_postcode_zeros(sanitize, postcode):
246     assert sanitize(country='pl', postcode=postcode) == []