X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/bb175cc95898de420781867973b38d033c187e81..18864afa8aee710a5aa7fe65565711119ca7a663:/test/python/tools/test_country_info.py?ds=sidebyside diff --git a/test/python/tools/test_country_info.py b/test/python/tools/test_country_info.py index 66f785c2..3f00d54e 100644 --- a/test/python/tools/test_country_info.py +++ b/test/python/tools/test_country_info.py @@ -1,25 +1,45 @@ +# SPDX-License-Identifier: GPL-2.0-only +# +# This file is part of Nominatim. (https://nominatim.org) +# +# Copyright (C) 2022 by the Nominatim developer community. +# For a full list of authors see the git log. """ Tests for function that handle country properties. """ - +from textwrap import dedent import pytest from nominatim.tools import country_info -@pytest.fixture(autouse=True) -def read_config(def_config): +@pytest.fixture +def loaded_country(def_config): country_info.setup_country_config(def_config) + +@pytest.fixture +def env_with_country_config(project_env): + + def _mk_config(cfg): + (project_env.project_dir / 'country_settings.yaml').write_text(dedent(cfg)) + + return project_env + + return _mk_config + + @pytest.mark.parametrize("no_partitions", (True, False)) def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cursor, - def_config, no_partitions): + loaded_country, no_partitions): country_info.setup_country_tables(dsn, src_dir / 'data', no_partitions) assert temp_db_cursor.table_exists('country_name') assert temp_db_cursor.table_rows('country_name') == \ - temp_db_cursor.scalar('SELECT count(DISTINCT country_code) FROM country_name') + temp_db_cursor.scalar( + 'SELECT count(DISTINCT country_code) FROM country_name') - partitions = temp_db_cursor.row_set("SELECT DISTINCT partition FROM country_name") + partitions = temp_db_cursor.row_set( + "SELECT DISTINCT partition FROM country_name") if no_partitions: assert partitions == {(0, )} else: @@ -29,9 +49,9 @@ def test_setup_country_tables(src_dir, temp_db_with_extensions, dsn, temp_db_cur assert temp_db_cursor.table_rows('country_osm_grid') > 100 -@pytest.mark.parametrize("languages", (None, ' fr,en')) +@pytest.mark.parametrize("languages", (None, ['fr', 'en'])) def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor, - table_factory, tokenizer_mock, languages): + table_factory, tokenizer_mock, languages, loaded_country): table_factory('country_name', 'country_code varchar(2), name hstore', content=(('us', '"name"=>"us1","name:af"=>"us2"'), @@ -45,11 +65,107 @@ def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cur assert len(tokenizer.analyser_cache['countries']) == 2 - result_set = {k: set(v.values()) for k, v in tokenizer.analyser_cache['countries']} + result_set = {k: set(v.values()) + for k, v in tokenizer.analyser_cache['countries']} if languages: - assert result_set == {'us' : set(('us', 'us1', 'United States')), - 'fr' : set(('fr', 'Fra', 'Fren'))} + assert result_set == {'us': set(('us', 'us1')), + 'fr': set(('fr', 'Fra', 'Fren'))} else: - assert result_set == {'us' : set(('us', 'us1', 'us2', 'United States')), - 'fr' : set(('fr', 'Fra', 'Fren'))} + assert result_set == {'us': set(('us', 'us1', 'us2')), + 'fr': set(('fr', 'Fra', 'Fren'))} + + +def test_setup_country_names_prefixes(env_with_country_config): + config = env_with_country_config("""\ + es: + names: + name: + en: Spain + de: Spanien + default: Espagñe + us: + names: + short_name: + default: USA + name: + default: United States + en: United States + """) + info = country_info._CountryInfo() + info.load(config) + + assert info.get('es')['names'] == {"name": "Espagñe", + "name:en": "Spain", + "name:de": "Spanien"} + assert info.get('us')['names'] == {"name": "United States", + "name:en": "United States", + "short_name": "USA"} + assert 'names' not in info.get('xx') + + +def test_setup_country_config_languages_not_loaded(env_with_country_config): + config = env_with_country_config("""\ + de: + partition: 3 + names: + name: + default: Deutschland + """) + info = country_info._CountryInfo() + info.load(config) + assert dict(info.items()) == {'de': {'partition': 3, + 'languages': [], + 'names': {'name': 'Deutschland'}}} + + +def test_setup_country_config_name_not_loaded(env_with_country_config): + config = env_with_country_config("""\ + de: + partition: 3 + languages: de + names: + """) + + info = country_info._CountryInfo() + info.load(config) + + assert dict(info.items()) == {'de': {'partition': 3, + 'languages': ['de'], + 'names': {} + }} + + +def test_setup_country_config_names_not_loaded(env_with_country_config): + config = env_with_country_config(""" + de: + partition: 3 + languages: de + """) + + info = country_info._CountryInfo() + info.load(config) + + assert dict(info.items()) == {'de': {'partition': 3, + 'languages': ['de'], + 'names': {} + }} + + +def test_setup_country_config_special_character(env_with_country_config): + config = env_with_country_config(""" + bq: + partition: 250 + languages: nl + names: + name: + default: "\\N" + """) + + info = country_info._CountryInfo() + info.load(config) + + assert dict(info.items()) == {'bq': {'partition': 250, + 'languages': ['nl'], + 'names': {'name': '\x85'} + }}