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.indexer.place_info import PlaceInfo
  14 from nominatim.tools 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', '  34009  '))
  47 def test_postcode_pass_good_format(sanitize, postcode):
  48     assert sanitize(country='de', postcode=postcode) == [('postcode', postcode.strip())]
  49
  50
  51 @pytest.mark.parametrize("postcode", ('123456', '', '   ', '.....'))
  52 @pytest.mark.sanitizer_params(convert_to_address=False)
  53 def test_postcode_drop_bad_format(sanitize, postcode):
  54     assert sanitize(country='de', postcode=postcode) == []
  55
  56
  57 @pytest.mark.parametrize("postcode", ('1234', '9435', '99000'))
  58 def test_postcode_cyprus_pass(sanitize, postcode):
  59     assert sanitize(country='cy', postcode=postcode) == [('postcode', postcode)]
  60
  61
  62 @pytest.mark.parametrize("postcode", ('91234', '99a45', '567'))
  63 @pytest.mark.sanitizer_params(convert_to_address=False)
  64 def test_postcode_cyprus_fail(sanitize, postcode):
  65     assert sanitize(country='cy', postcode=postcode) == []
  66
  67
  68 @pytest.mark.parametrize("postcode", ('123456', 'A33F2G7'))
  69 def test_postcode_kazakhstan_pass(sanitize, postcode):
  70     assert sanitize(country='kz', postcode=postcode) == [('postcode', postcode)]
  71
  72
  73 @pytest.mark.parametrize("postcode", ('V34T6Y923456', '99345'))
  74 @pytest.mark.sanitizer_params(convert_to_address=False)
  75 def test_postcode_kazakhstan_fail(sanitize, postcode):
  76     assert sanitize(country='kz', postcode=postcode) == []
  77