test/python/data/test_country_info.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 function that handle country properties.
   9 """
  10 from textwrap import dedent
  11 import pytest
  12
  13 from nominatim_db.data import country_info
  14
  15
  16 @pytest.fixture
  17 def loaded_country(def_config):
  18     country_info.setup_country_config(def_config)
  19
  20
  21 @pytest.fixture
  22 def env_with_country_config(project_env):
  23
  24     def _mk_config(cfg):
  25         (project_env.project_dir / 'country_settings.yaml').write_text(dedent(cfg))
  26
  27         return project_env
  28
  29     return _mk_config
  30
  31
  32 @pytest.mark.parametrize("no_partitions", (True, False))
  33 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  34                               loaded_country, no_partitions):
  35     country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions)
  36
  37     assert temp_db_cursor.table_exists('country_name')
  38     assert temp_db_cursor.table_rows('country_name') == \
  39         temp_db_cursor.scalar(
  40             'SELECT count(DISTINCT country_code) FROM country_name')
  41
  42     partitions = temp_db_cursor.row_set(
  43         "SELECT DISTINCT partition FROM country_name")
  44     if no_partitions:
  45         assert partitions == {(0, )}
  46     else:
  47         assert len(partitions) > 10
  48
  49     assert temp_db_cursor.table_exists('country_osm_grid')
  50     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  51
  52
  53 @pytest.mark.parametrize("languages", (None, ['fr', 'en']))
  54 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  55                               table_factory, tokenizer_mock, languages, loaded_country):
  56
  57     table_factory('country_name', 'country_code varchar(2), name hstore',
  58                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  59                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  60
  61     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  62
  63     tokenizer = tokenizer_mock()
  64
  65     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  66
  67     assert len(tokenizer.analyser_cache['countries']) == 2
  68
  69     result_set = {k: set(v.values())
  70                   for k, v in tokenizer.analyser_cache['countries']}
  71
  72     if languages:
  73         assert result_set == {'us': set(('us', 'us1')),
  74                               'fr': set(('fr', 'Fra', 'Fren'))}
  75     else:
  76         assert result_set == {'us': set(('us', 'us1', 'us2')),
  77                               'fr': set(('fr', 'Fra', 'Fren'))}
  78
  79
  80 def test_setup_country_names_prefixes(env_with_country_config):
  81     config = env_with_country_config("""\
  82                                      es:
  83                                        names:
  84                                          name:
  85                                            en: Spain
  86                                            de: Spanien
  87                                            default: Espagñe
  88                                      us:
  89                                        names:
  90                                          short_name:
  91                                            default: USA
  92                                          name:
  93                                            default: United States
  94                                            en: United States
  95                                      """)
  96     info = country_info._CountryInfo()
  97     info.load(config)
  98
  99     assert info.get('es')['names'] == {"name": "Espagñe",
 100                                        "name:en": "Spain",
 101                                        "name:de": "Spanien"}
 102     assert info.get('us')['names'] == {"name": "United States",
 103                                        "name:en": "United States",
 104                                        "short_name": "USA"}
 105     assert 'names' not in info.get('xx')
 106
 107
 108 def test_setup_country_config_languages_not_loaded(env_with_country_config):
 109     config = env_with_country_config("""\
 110                                      de:
 111                                          partition: 3
 112                                          names:
 113                                              name:
 114                                                  default: Deutschland
 115                                      """)
 116     info = country_info._CountryInfo()
 117     info.load(config)
 118     assert dict(info.items()) == {'de': {'partition': 3,
 119                                          'languages': [],
 120                                          'names': {'name': 'Deutschland'}}}
 121
 122
 123 def test_setup_country_config_name_not_loaded(env_with_country_config):
 124     config = env_with_country_config("""\
 125                                      de:
 126                                          partition: 3
 127                                          languages: de
 128                                          names:
 129                                      """)
 130
 131     info = country_info._CountryInfo()
 132     info.load(config)
 133
 134     assert dict(info.items()) == {'de': {'partition': 3,
 135                                          'languages': ['de'],
 136                                          'names': {}}}
 137
 138
 139 def test_setup_country_config_names_not_loaded(env_with_country_config):
 140     config = env_with_country_config("""
 141                                      de:
 142                                          partition: 3
 143                                          languages: de
 144                                      """)
 145
 146     info = country_info._CountryInfo()
 147     info.load(config)
 148
 149     assert dict(info.items()) == {'de': {'partition': 3,
 150                                          'languages': ['de'],
 151                                          'names': {}}}
 152
 153
 154 def test_setup_country_config_special_character(env_with_country_config):
 155     config = env_with_country_config("""
 156                                      bq:
 157                                          partition: 250
 158                                          languages: nl
 159                                          names:
 160                                              name:
 161                                                  default: "\\N"
 162                                      """)
 163
 164     info = country_info._CountryInfo()
 165     info.load(config)
 166
 167     assert dict(info.items()) == {'bq': {'partition': 250,
 168                                          'languages': ['nl'],
 169                                          'names': {'name': '\x85'}}}