test/python/tools/test_database_import.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 functions to import a new database.
   9 """
  10 from pathlib import Path
  11
  12 import pytest
  13 import pytest_asyncio
  14 import psycopg
  15 from psycopg import sql as pysql
  16
  17 from nominatim_db.tools import database_import
  18 from nominatim_db.errors import UsageError
  19
  20 class TestDatabaseSetup:
  21     DBNAME = 'test_nominatim_python_unittest'
  22
  23     @pytest.fixture(autouse=True)
  24     def setup_nonexistant_db(self):
  25         with psycopg.connect(dbname='postgres', autocommit=True) as conn:
  26             with conn.cursor() as cur:
  27                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  28
  29             yield True
  30
  31             with conn.cursor() as cur:
  32                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  33
  34
  35     @pytest.fixture
  36     def cursor(self):
  37         with psycopg.connect(dbname=self.DBNAME) as conn:
  38             with conn.cursor() as cur:
  39                 yield cur
  40
  41
  42     def conn(self):
  43         return psycopg.connect(dbname=self.DBNAME)
  44
  45
  46     def test_setup_skeleton(self):
  47         database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  48
  49         # Check that all extensions are set up.
  50         with self.conn() as conn:
  51             with conn.cursor() as cur:
  52                 cur.execute('CREATE TABLE t (h HSTORE, geom GEOMETRY(Geometry, 4326))')
  53
  54
  55     def test_unsupported_pg_version(self, monkeypatch):
  56         monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  57
  58         with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  59             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  60
  61
  62     def test_create_db_explicit_ro_user(self):
  63         database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  64                                                 rouser='postgres')
  65
  66
  67     def test_create_db_missing_ro_user(self):
  68         with pytest.raises(UsageError, match='Missing read-only user.'):
  69             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  70                                                     rouser='sdfwkjkjgdugu2;jgsafkljas;')
  71
  72
  73     def test_setup_extensions_old_postgis(self, monkeypatch):
  74         monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  75
  76         with pytest.raises(UsageError, match='PostGIS is too old.'):
  77             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  78
  79
  80 def test_setup_skeleton_already_exists(temp_db):
  81     with pytest.raises(UsageError):
  82         database_import.setup_database_skeleton(f'dbname={temp_db}')
  83
  84
  85 def test_import_osm_data_simple(table_factory, osm2pgsql_options, capfd):
  86     table_factory('place', content=((1, ), ))
  87
  88     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
  89     captured = capfd.readouterr()
  90
  91     assert '--create' in captured.out
  92     assert '--output gazetteer' in captured.out
  93     assert f'--style {osm2pgsql_options["osm2pgsql_style"]}' in captured.out
  94     assert f'--number-processes {osm2pgsql_options["threads"]}' in captured.out
  95     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
  96     assert 'file.pbf' in captured.out
  97
  98
  99 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options, capfd):
 100     table_factory('place', content=((1, ), ))
 101     osm2pgsql_options['osm2pgsql_cache'] = 0
 102
 103     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
 104     for f in files:
 105         f.write_text('test')
 106
 107     database_import.import_osm_data(files, osm2pgsql_options)
 108     captured = capfd.readouterr()
 109
 110     assert 'file1.osm' in captured.out
 111     assert 'file2.osm' in captured.out
 112
 113
 114 def test_import_osm_data_simple_no_data(table_factory, osm2pgsql_options):
 115     table_factory('place')
 116
 117     with pytest.raises(UsageError, match='No data imported'):
 118         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 119
 120
 121 def test_import_osm_data_simple_ignore_no_data(table_factory, osm2pgsql_options):
 122     table_factory('place')
 123
 124     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 125                                     ignore_errors=True)
 126
 127
 128 def test_import_osm_data_drop(table_factory, temp_db_cursor, tmp_path, osm2pgsql_options):
 129     table_factory('place', content=((1, ), ))
 130     table_factory('planet_osm_nodes')
 131
 132     flatfile = tmp_path / 'flatfile'
 133     flatfile.write_text('touch')
 134
 135     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 136
 137     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 138
 139     assert not flatfile.exists()
 140     assert not temp_db_cursor.table_exists('planet_osm_nodes')
 141
 142
 143 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 144     table_factory('place', content=((1, ), ))
 145
 146     osm2pgsql_options['osm2pgsql_cache'] = 0
 147
 148     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 149     captured = capfd.readouterr()
 150
 151     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 152
 153
 154 @pytest.mark.parametrize("with_search", (True, False))
 155 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 156     tables = ['placex', 'place_addressline', 'location_area',
 157               'location_area_country',
 158               'location_property_tiger', 'location_property_osmline',
 159               'location_postcode', 'location_road_23']
 160     if with_search:
 161         tables.append('search_name')
 162
 163     for table in tables:
 164         table_factory(table, content=((1, ), (2, ), (3, )))
 165         assert temp_db_cursor.table_rows(table) == 3
 166
 167     database_import.truncate_data_tables(temp_db_conn)
 168
 169     for table in tables:
 170         assert temp_db_cursor.table_rows(table) == 0
 171
 172
 173 @pytest.mark.parametrize("threads", (1, 5))
 174 @pytest.mark.asyncio
 175 async def test_load_data(dsn, place_row, placex_table, osmline_table,
 176                    temp_db_cursor, threads):
 177     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 178         temp_db_cursor.execute(pysql.SQL("""CREATE FUNCTION {} (src TEXT)
 179                                             RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 180                                          """).format(pysql.Identifier(func)))
 181     for oid in range(100, 130):
 182         place_row(osm_id=oid)
 183     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 184               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 185
 186     await database_import.load_data(dsn, threads)
 187
 188     assert temp_db_cursor.table_rows('placex') == 30
 189     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 190
 191
 192 class TestSetupSQL:
 193
 194     @pytest.fixture(autouse=True)
 195     def init_env(self, temp_db, tmp_path, def_config, sql_preprocessor_cfg):
 196         def_config.lib_dir.sql = tmp_path / 'sql'
 197         def_config.lib_dir.sql.mkdir()
 198
 199         self.config = def_config
 200
 201
 202     def write_sql(self, fname, content):
 203         (self.config.lib_dir.sql / fname).write_text(content)
 204
 205
 206     @pytest.mark.parametrize("reverse", [True, False])
 207     def test_create_tables(self, temp_db_conn, temp_db_cursor, reverse):
 208         self.write_sql('tables.sql',
 209                        """CREATE FUNCTION test() RETURNS bool
 210                           AS $$ SELECT {{db.reverse_only}} $$ LANGUAGE SQL""")
 211
 212         database_import.create_tables(temp_db_conn, self.config, reverse)
 213
 214         temp_db_cursor.scalar('SELECT test()') == reverse
 215
 216
 217     def test_create_table_triggers(self, temp_db_conn, temp_db_cursor):
 218         self.write_sql('table-triggers.sql',
 219                        """CREATE FUNCTION test() RETURNS TEXT
 220                           AS $$ SELECT 'a'::text $$ LANGUAGE SQL""")
 221
 222         database_import.create_table_triggers(temp_db_conn, self.config)
 223
 224         temp_db_cursor.scalar('SELECT test()') == 'a'
 225
 226
 227     def test_create_partition_tables(self, temp_db_conn, temp_db_cursor):
 228         self.write_sql('partition-tables.src.sql',
 229                        """CREATE FUNCTION test() RETURNS TEXT
 230                           AS $$ SELECT 'b'::text $$ LANGUAGE SQL""")
 231
 232         database_import.create_partition_tables(temp_db_conn, self.config)
 233
 234         temp_db_cursor.scalar('SELECT test()') == 'b'
 235
 236
 237     @pytest.mark.parametrize("drop", [True, False])
 238     @pytest.mark.asyncio
 239     async def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop):
 240         self.write_sql('indices.sql',
 241                        """CREATE FUNCTION test() RETURNS bool
 242                           AS $$ SELECT {{drop}} $$ LANGUAGE SQL""")
 243
 244         await database_import.create_search_indices(temp_db_conn, self.config, drop)
 245
 246         temp_db_cursor.scalar('SELECT test()') == drop