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, request):
  17     if 'custom_country_config' in request.keywords:
  18         return
  19     country_info.setup_country_config(def_config)
  20
  21 @pytest.mark.parametrize("no_partitions", (True, False))
  22 def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor,
  23                               def_config, no_partitions):
  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('SELECT count(DISTINCT country_code) FROM country_name')
  29
  30     partitions = temp_db_cursor.row_set("SELECT DISTINCT partition FROM country_name")
  31     if no_partitions:
  32         assert partitions == {(0, )}
  33     else:
  34         assert len(partitions) > 10
  35
  36     assert temp_db_cursor.table_exists('country_osm_grid')
  37     assert temp_db_cursor.table_rows('country_osm_grid') > 100
  38
  39
  40 @pytest.mark.parametrize("languages", (None, ' fr,en'))
  41 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
  42                               table_factory, tokenizer_mock, languages):
  43
  44     table_factory('country_name', 'country_code varchar(2), name hstore',
  45                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
  46                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
  47
  48     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
  49
  50     tokenizer = tokenizer_mock()
  51
  52     country_info.create_country_names(temp_db_conn, tokenizer, languages)
  53
  54     assert len(tokenizer.analyser_cache['countries']) == 2
  55
  56     result_set = {k: set(v.values()) for k, v in tokenizer.analyser_cache['countries']}
  57
  58     if languages:
  59         assert result_set == {'us' : set(('us', 'us1', 'United States')),
  60                               'fr' : set(('fr', 'Fra', 'Fren'))}
  61     else:
  62         assert result_set == {'us' : set(('us', 'us1', 'us2', 'United States')),
  63                               'fr' : set(('fr', 'Fra', 'Fren'))}
  64
  65 @pytest.mark.custom_country_config
  66 def test_setup_country_config_languages_not_loaded(project_env):
  67     (project_env.project_dir / 'country_settings.yaml').write_text("""
  68 de:
  69     partition: 3
  70     names:
  71         name:
  72             default: Deutschland
  73 """)
  74     country_info.setup_country_config(project_env)
  75     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  76             'languages': [], 'names': {'name': {'default': 'Deutschland'}}}}
  77
  78 @pytest.mark.custom_country_config
  79 def test_setup_country_config_name_not_loaded(project_env):
  80     (project_env.project_dir / 'country_settings.yaml').write_text("""
  81 de:
  82     partition: 3
  83     languages: de
  84     names:
  85 """)
  86     country_info._COUNTRY_INFO._info = None
  87     country_info.setup_country_config(project_env)
  88     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
  89             'languages': ['de'], 'names': {'name': {}}}}
  90
  91 @pytest.mark.custom_country_config
  92 def test_setup_country_config_names_not_loaded(project_env):
  93     (project_env.project_dir / 'country_settings.yaml').write_text("""
  94 de:
  95     partition: 3
  96     languages: de
  97 """)
  98     country_info._COUNTRY_INFO._info = None
  99     country_info.setup_country_config(project_env)
 100     assert country_info._COUNTRY_INFO._info == {'de': {'partition': 3,
 101             'languages': ['de'], 'names': {'name': {}}}}