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) 2025 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         name, _ = PlaceSanitizer([{'step': 'split-name-list'}], self.config).process_names(place)
  27
  28         return sorted([(p.name, p.kind, p.suffix) for p in name])
  29
  30     def sanitize_with_delimiter(self, delimiter, name):
  31         place = PlaceInfo({'name': {'name': name}})
  32         san = PlaceSanitizer([{'step': 'split-name-list', 'delimiters': delimiter}],
  33                              self.config)
  34         name, _ = san.process_names(place)
  35
  36         return sorted([p.name for p in name])
  37
  38     def test_simple(self):
  39         assert self.run_sanitizer_on(name='ABC') == [('ABC', 'name', None)]
  40         assert self.run_sanitizer_on(name='') == [('', 'name', None)]
  41
  42     def test_splits(self):
  43         assert self.run_sanitizer_on(name='A;B;C') == [('A', 'name', None),
  44                                                        ('B', 'name', None),
  45                                                        ('C', 'name', None)]
  46         assert self.run_sanitizer_on(short_name=' House, boat ') == [('House', 'short_name', None),
  47                                                                      ('boat', 'short_name', None)]
  48
  49     def test_empty_fields(self):
  50         assert self.run_sanitizer_on(name='A;;B') == [('A', 'name', None),
  51                                                       ('B', 'name', None)]
  52         assert self.run_sanitizer_on(name='A; ,B') == [('A', 'name', None),
  53                                                        ('B', 'name', None)]
  54         assert self.run_sanitizer_on(name=' ;B') == [('B', 'name', None)]
  55         assert self.run_sanitizer_on(name='B,') == [('B', 'name', None)]
  56
  57     def test_custom_delimiters(self):
  58         assert self.sanitize_with_delimiter(':', '12:45,3') == ['12', '45,3']
  59         assert self.sanitize_with_delimiter('\\', 'a;\\b!#@ \\') == ['a;', 'b!#@']
  60         assert self.sanitize_with_delimiter('[]', 'foo[to]be') == ['be', 'foo', 'to']
  61         assert self.sanitize_with_delimiter(' ', 'morning  sun') == ['morning', 'sun']
  62
  63     def test_empty_delimiter_set(self):
  64         with pytest.raises(UsageError):
  65             self.sanitize_with_delimiter('', 'abc')
  66
  67
  68 def test_no_name_list(def_config):
  69     place = PlaceInfo({'address': {'housenumber': '3'}})
  70     name, address = PlaceSanitizer([{'step': 'split-name-list'}], def_config).process_names(place)
  71
  72     assert not name
  73     assert len(address) == 1