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) 2024 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 @pytest.fixture
  16 def loaded_country(def_config):
  17     country_info.setup_country_config(def_config)
  18
  19
  20 @pytest.fixture
  21 def env_with_country_config(project_env):
  22
  23     def _mk_config(cfg):
  24         (project_env.project_dir / 'country_settings.yaml').write_text(dedent(cfg))
  25
  26         return project_env
  27
  28     return _mk_config
  29
  30
  31 @pytest.mark.parametrize("no_partitions", (True, False))
  32 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  33                               loaded_country, no_partitions):
  34     country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions)
  35
  36     assert temp_db_cursor.table_exists('country_name')
  37     assert temp_db_cursor.table_rows('country_name') == \
  38         temp_db_cursor.scalar(
  39             'SELECT count(DISTINCT country_code) FROM country_name')
  40
  41     partitions = temp_db_cursor.row_set(
  42         "SELECT DISTINCT partition FROM country_name")
  43     if no_partitions:
  44         assert partitions == {(0, )}
  45     else:
  46         assert len(partitions) > 10
  47
  48     assert temp_db_cursor.table_exists('country_osm_grid')
  49     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  50
  51
  52 @pytest.mark.parametrize("languages", (None, ['fr', 'en']))
  53 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  54                               table_factory, tokenizer_mock, languages, loaded_country):
  55
  56     table_factory('country_name', 'country_code varchar(2), name hstore',
  57                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  58                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  59
  60     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  61
  62     tokenizer = tokenizer_mock()
  63
  64     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  65
  66     assert len(tokenizer.analyser_cache['countries']) == 2
  67
  68     result_set = {k: set(v.values())
  69                   for k, v in tokenizer.analyser_cache['countries']}
  70
  71     if languages:
  72         assert result_set == {'us': set(('us', 'us1')),
  73                               'fr': set(('fr', 'Fra', 'Fren'))}
  74     else:
  75         assert result_set == {'us': set(('us', 'us1', 'us2')),
  76                               'fr': set(('fr', 'Fra', 'Fren'))}
  77
  78
  79 def test_setup_country_names_prefixes(env_with_country_config):
  80     config = env_with_country_config("""\
  81                                      es:
  82                                        names:
  83                                          name:
  84                                            en: Spain
  85                                            de: Spanien
  86                                            default: Espagñe
  87                                      us:
  88                                        names:
  89                                          short_name:
  90                                            default: USA
  91                                          name:
  92                                            default: United States
  93                                            en: United States
  94                                      """)
  95     info = country_info._CountryInfo()
  96     info.load(config)
  97
  98     assert info.get('es')['names'] == {"name": "Espagñe",
  99                                        "name:en": "Spain",
 100                                        "name:de": "Spanien"}
 101     assert info.get('us')['names'] == {"name": "United States",
 102                                        "name:en": "United States",
 103                                        "short_name": "USA"}
 104     assert 'names' not in info.get('xx')
 105
 106
 107 def test_setup_country_config_languages_not_loaded(env_with_country_config):
 108     config = env_with_country_config("""\
 109                                      de:
 110                                          partition: 3
 111                                          names:
 112                                              name:
 113                                                  default: Deutschland
 114                                      """)
 115     info = country_info._CountryInfo()
 116     info.load(config)
 117     assert dict(info.items()) == {'de': {'partition': 3,
 118                                   'languages': [],
 119                                   'names': {'name': 'Deutschland'}}}
 120
 121
 122 def test_setup_country_config_name_not_loaded(env_with_country_config):
 123     config = env_with_country_config("""\
 124                                      de:
 125                                          partition: 3
 126                                          languages: de
 127                                          names:
 128                                      """)
 129
 130     info = country_info._CountryInfo()
 131     info.load(config)
 132
 133     assert dict(info.items()) == {'de': {'partition': 3,
 134                                          'languages': ['de'],
 135                                          'names': {}
 136                                  }}
 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
 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'}
 171                                  }}