test/python/tools/test_country_info.py

   1 # SPDX-License-Identifier: GPL-2.0-only
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2022 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
  11 import pytest
  12
  13 from nominatim.tools import country_info
  14
  15
  16 def read_config(def_config):
  17     country_info.setup_country_config(def_config)
  18
  19
  20 @pytest.mark.parametrize("no_partitions", (True, False))
  21 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  22                               def_config, no_partitions):
  23     read_config(def_config)
  24     country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions)
  25
  26     assert temp_db_cursor.table_exists('country_name')
  27     assert temp_db_cursor.table_rows('country_name') == \
  28         temp_db_cursor.scalar(
  29             'SELECT count(DISTINCT country_code) FROM country_name')
  30
  31     partitions = temp_db_cursor.row_set(
  32         "SELECT DISTINCT partition FROM country_name")
  33     if no_partitions:
  34         assert partitions == {(0, )}
  35     else:
  36         assert len(partitions) > 10
  37
  38     assert temp_db_cursor.table_exists('country_osm_grid')
  39     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  40
  41
  42 @pytest.mark.parametrize("languages", (None, ' fr,en'))
  43 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  44                               table_factory, tokenizer_mock, languages, def_config):
  45     read_config(def_config)
  46
  47     table_factory('country_name', 'country_code varchar(2), name hstore',
  48                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  49                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  50
  51     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  52
  53     tokenizer = tokenizer_mock()
  54
  55     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  56
  57     assert len(tokenizer.analyser_cache['countries']) == 2
  58
  59     result_set = {k: set(v.values())
  60                   for k, v in tokenizer.analyser_cache['countries']}
  61
  62     if languages:
  63         assert result_set == {'us': set(('us', 'us1', 'United States')),
  64                               'fr': set(('fr', 'Fra', 'Fren'))}
  65     else:
  66         assert result_set == {'us': set(('us', 'us1', 'us2', 'United States')),
  67                               'fr': set(('fr', 'Fra', 'Fren'))}
  68
  69
  70 def test_setup_country_config_languages_not_loaded(project_env):
  71     (project_env.project_dir / 'country_settings.yaml').write_text("""
  72 de:
  73     partition: 3
  74     names:
  75         name:
  76             default: Deutschland
  77 """)
  78     country_info._COUNTRY_INFO._info = None
  79     country_info.setup_country_config(project_env)
  80     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  81                                                        'languages': [], 'names': {'name': {'default': 'Deutschland'}}}}
  82
  83
  84 def test_setup_country_config_name_not_loaded(project_env):
  85     (project_env.project_dir / 'country_settings.yaml').write_text("""
  86 de:
  87     partition: 3
  88     languages: de
  89     names:
  90 """)
  91     country_info._COUNTRY_INFO._info = None
  92     country_info.setup_country_config(project_env)
  93     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  94                                                        'languages': ['de'], 'names': {'name': {}}}}
  95
  96
  97 def test_setup_country_config_names_not_loaded(project_env):
  98     (project_env.project_dir / 'country_settings.yaml').write_text("""
  99 de:
 100     partition: 3
 101     languages: de
 102 """)
 103     country_info._COUNTRY_INFO._info = None
 104     country_info.setup_country_config(project_env)
 105     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
 106                                                        'languages': ['de'], 'names': {'name': {}}}}
 107
 108
 109 def test_setup_country_config_special_character(project_env):
 110     (project_env.project_dir / 'country_settings.yaml').write_text("""
 111 bq:
 112     partition: 250
 113     languages: nl
 114     names:
 115         name:
 116             default: "\\N"
 117 """)
 118     country_info._COUNTRY_INFO._info = None
 119     country_info.setup_country_config(project_env)
 120     assert country_info._COUNTRY_INFO._info == {'bq': {'partition': 250,
 121                                                        'languages': ['nl'], 'names': {'name': {'default': '\x85'}}}}