test/python/tokenizer/sanitizers/test_clean_postcodes.py

   1 # SPDX-License-Identifier: GPL-2.0-only
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2022 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.tokenizer.place_sanitizer import PlaceSanitizer
  13 from nominatim.data.place_info import PlaceInfo
  14 from nominatim.data import country_info
  15
  16 @pytest.fixture
  17 def sanitize(def_config, request):
  18     country_info.setup_country_config(def_config)
  19     sanitizer_args = {'step': 'clean-postcodes'}
  20     for mark in request.node.iter_markers(name="sanitizer_params"):
  21         sanitizer_args.update({k.replace('_', '-') : v for k,v in mark.kwargs.items()})
  22
  23     def _run(country=None, **kwargs):
  24         pi = {'address': kwargs}
  25         if country is not None:
  26             pi['country_code'] = country
  27
  28         _, address = PlaceSanitizer([sanitizer_args]).process_names(PlaceInfo(pi))
  29
  30         return sorted([(p.kind, p.name) for p in address])
  31
  32     return _run
  33
  34
  35 @pytest.mark.parametrize("country", (None, 'ae'))
  36 def test_postcode_no_country(sanitize, country):
  37     assert sanitize(country=country, postcode='23231') == [('unofficial_postcode', '23231')]
  38
  39
  40 @pytest.mark.parametrize("country", (None, 'ae'))
  41 @pytest.mark.sanitizer_params(convert_to_address=False)
  42 def test_postcode_no_country_drop(sanitize, country):
  43     assert sanitize(country=country, postcode='23231') == []
  44
  45
  46 @pytest.mark.parametrize("postcode", ('12345', '  12345  ', 'de 12345',
  47                                       'DE12345', 'DE 12345', 'DE-12345'))
  48 def test_postcode_pass_good_format(sanitize, postcode):
  49     assert sanitize(country='de', postcode=postcode) == [('postcode', '12345')]
  50
  51
  52 @pytest.mark.parametrize("postcode", ('123456', '', '   ', '.....',
  53                                       'DE  12345', 'DEF12345', 'CH 12345'))
  54 @pytest.mark.sanitizer_params(convert_to_address=False)
  55 def test_postcode_drop_bad_format(sanitize, postcode):
  56     assert sanitize(country='de', postcode=postcode) == []
  57
  58
  59 @pytest.mark.parametrize("postcode", ('1234', '9435', '99000'))
  60 def test_postcode_cyprus_pass(sanitize, postcode):
  61     assert sanitize(country='cy', postcode=postcode) == [('postcode', postcode)]
  62
  63
  64 @pytest.mark.parametrize("postcode", ('91234', '99a45', '567'))
  65 @pytest.mark.sanitizer_params(convert_to_address=False)
  66 def test_postcode_cyprus_fail(sanitize, postcode):
  67     assert sanitize(country='cy', postcode=postcode) == []
  68
  69
  70 @pytest.mark.parametrize("postcode", ('123456', 'A33F2G7'))
  71 def test_postcode_kazakhstan_pass(sanitize, postcode):
  72     assert sanitize(country='kz', postcode=postcode) == [('postcode', postcode)]
  73
  74
  75 @pytest.mark.parametrize("postcode", ('V34T6Y923456', '99345'))
  76 @pytest.mark.sanitizer_params(convert_to_address=False)
  77 def test_postcode_kazakhstan_fail(sanitize, postcode):
  78     assert sanitize(country='kz', postcode=postcode) == []
  79
  80
  81 @pytest.mark.parametrize("postcode", ('675 34', '67534', 'SE-675 34', 'SE67534'))
  82 def test_postcode_sweden_pass(sanitize, postcode):
  83     assert sanitize(country='se', postcode=postcode) == [('postcode', '675 34')]
  84
  85
  86 @pytest.mark.parametrize("postcode", ('67 345', '671123'))
  87 @pytest.mark.sanitizer_params(convert_to_address=False)
  88 def test_postcode_sweden_fail(sanitize, postcode):
  89     assert sanitize(country='se', postcode=postcode) == []
  90
  91
  92 @pytest.mark.parametrize("postcode", ('AB1', '123-456-7890', '1 as 44'))
  93 @pytest.mark.sanitizer_params(default_pattern='[A-Z0-9- ]{3,12}')
  94 def test_postcode_default_pattern_pass(sanitize, postcode):
  95     assert sanitize(country='an', postcode=postcode) == [('postcode', postcode.upper())]
  96
  97
  98 @pytest.mark.parametrize("postcode", ('C', '12', 'ABC123DEF 456', '1234,5678', '11223;11224'))
  99 @pytest.mark.sanitizer_params(convert_to_address=False, default_pattern='[A-Z0-9- ]{3,12}')
 100 def test_postcode_default_pattern_fail(sanitize, postcode):
 101     assert sanitize(country='an', postcode=postcode) == []
 102