test/python/tokenizer/sanitizers/test_split_name_list.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 splits multivalue lists.
   9 """
  10 import pytest
  11
  12 from nominatim_db.tokenizer.place_sanitizer import PlaceSanitizer
  13 from nominatim_db.data.place_info import PlaceInfo
  14
  15 from nominatim_db.errors import UsageError
  16
  17
  18 class TestSplitName:
  19
  20     @pytest.fixture(autouse=True)
  21     def setup_country(self, def_config):
  22         self.config = def_config
  23
  24     def run_sanitizer_on(self, **kwargs):
  25         place = PlaceInfo({'name': kwargs})
  26         PlaceSanitizer([{'step': 'split-name-list'}], self.config).process_names(place)
  27
  28         return sorted([(p.name, p.kind, p.suffix) for p in place.searchable_names])
  29
  30     def sanitize_with_delimiter(self, delimiter, name):
  31         place = PlaceInfo({'name': {'name': name}})
  32         PlaceSanitizer([{'step': 'split-name-list', 'delimiters': delimiter}],
  33                        self.config).process_names(place)
  34
  35         return sorted([p.name for p in place.searchable_names])
  36
  37     def test_simple(self):
  38         assert self.run_sanitizer_on(name='ABC') == [('ABC', 'name', None)]
  39         assert self.run_sanitizer_on(name='') == [('', 'name', None)]
  40
  41     def test_splits(self):
  42         assert self.run_sanitizer_on(name='A;B;C') == [('A', 'name', None),
  43                                                        ('B', 'name', None),
  44                                                        ('C', 'name', None)]
  45         assert self.run_sanitizer_on(short_name=' House, boat ') == [('House', 'short_name', None),
  46                                                                      ('boat', 'short_name', None)]
  47
  48     def test_empty_fields(self):
  49         assert self.run_sanitizer_on(name='A;;B') == [('A', 'name', None),
  50                                                       ('B', 'name', None)]
  51         assert self.run_sanitizer_on(name='A; ,B') == [('A', 'name', None),
  52                                                        ('B', 'name', None)]
  53         assert self.run_sanitizer_on(name=' ;B') == [('B', 'name', None)]
  54         assert self.run_sanitizer_on(name='B,') == [('B', 'name', None)]
  55
  56     def test_custom_delimiters(self):
  57         assert self.sanitize_with_delimiter(':', '12:45,3') == ['12', '45,3']
  58         assert self.sanitize_with_delimiter('\\', 'a;\\b!#@ \\') == ['a;', 'b!#@']
  59         assert self.sanitize_with_delimiter('[]', 'foo[to]be') == ['be', 'foo', 'to']
  60         assert self.sanitize_with_delimiter(' ', 'morning  sun') == ['morning', 'sun']
  61
  62     def test_empty_delimiter_set(self):
  63         with pytest.raises(UsageError):
  64             self.sanitize_with_delimiter('', 'abc')
  65
  66
  67 def test_no_name_list(def_config):
  68     place = PlaceInfo({'address': {'housenumber': '3'}})
  69     PlaceSanitizer([{'step': 'split-name-list'}], def_config).process_names(place)
  70
  71     assert not place.searchable_names
  72     assert len(place.searchable_address) == 1