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) 2024 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 class TestSplitName:
  18
  19     @pytest.fixture(autouse=True)
  20     def setup_country(self, def_config):
  21         self.config = def_config
  22
  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
  31     def sanitize_with_delimiter(self, delimiter, name):
  32         place = PlaceInfo({'name': {'name': name}})
  33         san = PlaceSanitizer([{'step': 'split-name-list', 'delimiters': delimiter}],
  34                              self.config)
  35         name, _ = san.process_names(place)
  36
  37         return sorted([p.name for p in name])
  38
  39
  40     def test_simple(self):
  41         assert self.run_sanitizer_on(name='ABC') == [('ABC', 'name', None)]
  42         assert self.run_sanitizer_on(name='') == [('', 'name', None)]
  43
  44
  45     def test_splits(self):
  46         assert self.run_sanitizer_on(name='A;B;C') == [('A', 'name', None),
  47                                                        ('B', 'name', None),
  48                                                        ('C', 'name', None)]
  49         assert self.run_sanitizer_on(short_name=' House, boat ') == [('House', 'short_name', None),
  50                                                                      ('boat', 'short_name', None)]
  51
  52
  53     def test_empty_fields(self):
  54         assert self.run_sanitizer_on(name='A;;B') == [('A', 'name', None),
  55                                                       ('B', 'name', None)]
  56         assert self.run_sanitizer_on(name='A; ,B') == [('A', 'name', None),
  57                                                        ('B', 'name', None)]
  58         assert self.run_sanitizer_on(name=' ;B') == [('B', 'name', None)]
  59         assert self.run_sanitizer_on(name='B,') == [('B', 'name', None)]
  60
  61
  62     def test_custom_delimiters(self):
  63         assert self.sanitize_with_delimiter(':', '12:45,3') == ['12', '45,3']
  64         assert self.sanitize_with_delimiter('\\', 'a;\\b!#@ \\') == ['a;', 'b!#@']
  65         assert self.sanitize_with_delimiter('[]', 'foo[to]be') == ['be', 'foo', 'to']
  66         assert self.sanitize_with_delimiter(' ', 'morning  sun') == ['morning', 'sun']
  67
  68
  69     def test_empty_delimiter_set(self):
  70         with pytest.raises(UsageError):
  71             self.sanitize_with_delimiter('', 'abc')
  72
  73
  74 def test_no_name_list(def_config):
  75     place = PlaceInfo({'address': {'housenumber': '3'}})
  76     name, address = PlaceSanitizer([{'step': 'split-name-list'}], def_config).process_names(place)
  77
  78     assert not name
  79     assert len(address) == 1