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(
  26             dedent(cfg), encoding='utf-8')
  27
  28         return project_env
  29
  30     return _mk_config
  31
  32
  33 @pytest.mark.parametrize("no_partitions", (True, False))
  34 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  35                               loaded_country, no_partitions):
  36     country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions)
  37
  38     assert temp_db_cursor.table_exists('country_name')
  39     assert temp_db_cursor.table_rows('country_name') == \
  40         temp_db_cursor.scalar(
  41             'SELECT count(DISTINCT country_code) FROM country_name')
  42
  43     partitions = temp_db_cursor.row_set(
  44         "SELECT DISTINCT partition FROM country_name")
  45     if no_partitions:
  46         assert partitions == {(0, )}
  47     else:
  48         assert len(partitions) > 10
  49
  50     assert temp_db_cursor.table_exists('country_osm_grid')
  51     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  52
  53
  54 @pytest.mark.parametrize("languages", (None, ['fr', 'en']))
  55 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  56                               table_factory, tokenizer_mock, languages, loaded_country):
  57
  58     table_factory('country_name', 'country_code varchar(2), name hstore',
  59                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  60                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  61
  62     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  63
  64     tokenizer = tokenizer_mock()
  65
  66     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  67
  68     assert len(tokenizer.analyser_cache['countries']) == 2
  69
  70     result_set = {k: set(v.values())
  71                   for k, v in tokenizer.analyser_cache['countries']}
  72
  73     if languages:
  74         assert result_set == {'us': set(('us', 'us1')),
  75                               'fr': set(('fr', 'Fra', 'Fren'))}
  76     else:
  77         assert result_set == {'us': set(('us', 'us1', 'us2')),
  78                               'fr': set(('fr', 'Fra', 'Fren'))}
  79
  80
  81 def test_setup_country_names_prefixes(env_with_country_config):
  82     config = env_with_country_config("""\
  83                                      es:
  84                                        names:
  85                                          name:
  86                                            en: Spain
  87                                            de: Spanien
  88                                            default: Espagñe
  89                                      us:
  90                                        names:
  91                                          short_name:
  92                                            default: USA
  93                                          name:
  94                                            default: United States
  95                                            en: United States
  96                                      """)
  97     info = country_info._CountryInfo()
  98     info.load(config)
  99
 100     assert info.get('es')['names'] == {"name": "Espagñe",
 101                                        "name:en": "Spain",
 102                                        "name:de": "Spanien"}
 103     assert info.get('us')['names'] == {"name": "United States",
 104                                        "name:en": "United States",
 105                                        "short_name": "USA"}
 106     assert 'names' not in info.get('xx')
 107
 108
 109 def test_setup_country_config_languages_not_loaded(env_with_country_config):
 110     config = env_with_country_config("""\
 111                                      de:
 112                                          partition: 3
 113                                          names:
 114                                              name:
 115                                                  default: Deutschland
 116                                      """)
 117     info = country_info._CountryInfo()
 118     info.load(config)
 119     assert dict(info.items()) == {'de': {'partition': 3,
 120                                          'languages': [],
 121                                          'names': {'name': 'Deutschland'}}}
 122
 123
 124 def test_setup_country_config_name_not_loaded(env_with_country_config):
 125     config = env_with_country_config("""\
 126                                      de:
 127                                          partition: 3
 128                                          languages: de
 129                                          names:
 130                                      """)
 131
 132     info = country_info._CountryInfo()
 133     info.load(config)
 134
 135     assert dict(info.items()) == {'de': {'partition': 3,
 136                                          'languages': ['de'],
 137                                          'names': {}}}
 138
 139
 140 def test_setup_country_config_names_not_loaded(env_with_country_config):
 141     config = env_with_country_config("""
 142                                      de:
 143                                          partition: 3
 144                                          languages: de
 145                                      """)
 146
 147     info = country_info._CountryInfo()
 148     info.load(config)
 149
 150     assert dict(info.items()) == {'de': {'partition': 3,
 151                                          'languages': ['de'],
 152                                          'names': {}}}
 153
 154
 155 def test_setup_country_config_special_character(env_with_country_config):
 156     config = env_with_country_config("""
 157                                      bq:
 158                                          partition: 250
 159                                          languages: nl
 160                                          names:
 161                                              name:
 162                                                  default: "\\N"
 163                                      """)
 164
 165     info = country_info._CountryInfo()
 166     info.load(config)
 167
 168     assert dict(info.items()) == {'bq': {'partition': 250,
 169                                          'languages': ['nl'],
 170                                          'names': {'name': '\x85'}}}