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 @pytest.fixture(autouse=True)
  16 def read_config(def_config):
  17     country_info.setup_country_config(def_config)
  18
  19 @pytest.mark.parametrize("no_partitions", (True, False))
  20 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  21                               def_config, no_partitions):
  22     country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions)
  23
  24     assert temp_db_cursor.table_exists('country_name')
  25     assert temp_db_cursor.table_rows('country_name') == \
  26              temp_db_cursor.scalar('SELECT count(DISTINCT country_code) FROM country_name')
  27
  28     partitions = temp_db_cursor.row_set("SELECT DISTINCT partition FROM country_name")
  29     if no_partitions:
  30         assert partitions == {(0, )}
  31     else:
  32         assert len(partitions) > 10
  33
  34     assert temp_db_cursor.table_exists('country_osm_grid')
  35     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  36
  37
  38 @pytest.mark.parametrize("languages", (None, ' fr,en'))
  39 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  40                               table_factory, tokenizer_mock, languages):
  41
  42     table_factory('country_name', 'country_code varchar(2), name hstore',
  43                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  44                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  45
  46     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  47
  48     tokenizer = tokenizer_mock()
  49
  50     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  51
  52     assert len(tokenizer.analyser_cache['countries']) == 2
  53
  54     result_set = {k: set(v.values()) for k, v in tokenizer.analyser_cache['countries']}
  55
  56     if languages:
  57         assert result_set == {'us' : set(('us', 'us1', 'United States')),
  58                               'fr' : set(('fr', 'Fra', 'Fren'))}
  59     else:
  60         assert result_set == {'us' : set(('us', 'us1', 'us2', 'United States')),
  61                               'fr' : set(('fr', 'Fra', 'Fren'))}
  62
  63 @pytest.mark.parametrize("yaml_file_content", (
  64 """
  65 de:
  66     partition: 3
  67     names:
  68         name:
  69             default: Deutschland
  70 """,
  71 """
  72 de:
  73     partition: 3
  74     languages: de
  75     names:
  76 """,
  77 """
  78 de:
  79     partition: 3
  80     languages: de
  81 """
  82 ))
  83 def test_load(project_env, def_config, yaml_file_content):
  84     (project_env.project_dir / 'country_settings.yaml').write_text(yaml_file_content)
  85
  86     country_info._COUNTRY_INFO._info = def_config.load_sub_configuration(
  87         (project_env.project_dir / 'country_settings.yaml'))
  88
  89     for prop in country_info._COUNTRY_INFO._info.values():
  90         if 'languages' not in prop:
  91             prop['languages'] = []
  92             assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  93             'languages': [], 'names': {'name': {'default': 'Deutschland'}}}}
  94         if 'names' not in prop or prop['names'] is None:
  95             prop['names'] = {'name': {}}
  96             assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  97             'languages': 'de', 'names': {'name': {}}}}