test/python/tokenizer/sanitizers/test_derive_names.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 creates derived names using regular expressions.
   9 """
  10 import pytest
  11
  12 from nominatim_db.errors import UsageError
  13 from nominatim_db.data.place_info import PlaceInfo
  14 from nominatim_db.tokenizer.place_sanitizer import PlaceSanitizer
  15
  16
  17 @pytest.fixture
  18 def mk_sanitizer(def_config):
  19     def _f(**kwargs):
  20         args = {k.replace('_', '-'): v for k, v in kwargs.items()}
  21         return PlaceSanitizer([args | {'step': 'derive-names'}], def_config)
  22
  23     return _f
  24
  25
  26 def test_no_parameters(mk_sanitizer):
  27     with pytest.raises(UsageError, match='name-pattern is missing'):
  28         mk_sanitizer()
  29
  30
  31 @pytest.mark.parametrize('prim,out', [('name', 0), ('address', 1)])
  32 @pytest.mark.parametrize('keep', [True, False])
  33 def test_name_deletion(mk_sanitizer, prim, out, keep):
  34     san = mk_sanitizer(name_pattern='A.*B', type=prim, keep_original=keep)
  35
  36     names = dict(name='AltBoom', alt_name='foo', loc_name='AltB')
  37     place = PlaceInfo({'name': names, 'address': names,
  38                        'country_code': 'de', 'rank_address': 30})
  39
  40     san.process_names(place)
  41     res = [place.searchable_names, place.searchable_address]
  42
  43     assert len(res[(out + 1) % 2]) == 3
  44     if keep:
  45         assert len(res[out]) == 3
  46     else:
  47         assert {(p.kind, p.name) for p in res[out]} == {('name', 'AltBoom'),
  48                                                         ('alt_name', 'foo')}
  49
  50
  51 @pytest.fixture
  52 def simple_replace(mk_sanitizer):
  53     def _impl(name, pattern, *variants, keep=True):
  54         san = mk_sanitizer(name_pattern=pattern, type='name', keep_original=keep,
  55                            variants=variants)
  56         place = PlaceInfo({'name': {'name': name}, 'country_code': 'de', 'rank_address': 30})
  57         san.process_names(place)
  58
  59         return {p.name for p in place.searchable_names}
  60     return _impl
  61
  62
  63 def test_variant_single_name(simple_replace):
  64     assert simple_replace('A', 'A', 'The A') == {'A', 'The A'}
  65
  66
  67 def test_replace_single_name(simple_replace):
  68     assert simple_replace('A', 'A', 'The A', keep=False) == {'The A'}
  69
  70
  71 def test_variant_with_multiple_names(simple_replace):
  72     assert simple_replace('A', 'A', 'A1', 'A2', 'A1') == {'A', 'A1', 'A2'}
  73
  74
  75 def test_variant_with_group_replacement(simple_replace):
  76     assert simple_replace('abc X', '(.*) X', r'\1 Y') == {'abc X', 'abc Y'}