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) 2025 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  # noqa
  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
  21 class TestDatabaseSetup:
  22     DBNAME = 'test_nominatim_python_unittest'
  23
  24     @pytest.fixture(autouse=True)
  25     def setup_nonexistant_db(self):
  26         with psycopg.connect(dbname='postgres', autocommit=True) as conn:
  27             with conn.cursor() as cur:
  28                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  29
  30             yield True
  31
  32             with conn.cursor() as cur:
  33                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  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     def conn(self):
  42         return psycopg.connect(dbname=self.DBNAME)
  43
  44     def test_setup_skeleton(self):
  45         database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  46
  47         # Check that all extensions are set up.
  48         with self.conn() as conn:
  49             with conn.cursor() as cur:
  50                 cur.execute('CREATE TABLE t (h HSTORE, geom GEOMETRY(Geometry, 4326))')
  51
  52     def test_unsupported_pg_version(self, monkeypatch):
  53         monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  54
  55         with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  56             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  57
  58     def test_create_db_explicit_ro_user(self):
  59         database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  60                                                 rouser='postgres')
  61
  62     def test_create_db_missing_ro_user(self):
  63         with pytest.raises(UsageError, match='Missing read-only user.'):
  64             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  65                                                     rouser='sdfwkjkjgdugu2;jgsafkljas;')
  66
  67     def test_setup_extensions_old_postgis(self, monkeypatch):
  68         monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  69
  70         with pytest.raises(UsageError, match='PostGIS is too old.'):
  71             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  72
  73
  74 def test_setup_skeleton_already_exists(temp_db):
  75     with pytest.raises(UsageError):
  76         database_import.setup_database_skeleton(f'dbname={temp_db}')
  77
  78
  79 def test_import_osm_data_simple(table_factory, osm2pgsql_options, capfd):
  80     table_factory('place', content=((1, ), ))
  81
  82     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
  83     captured = capfd.readouterr()
  84
  85     assert '--create' in captured.out
  86     assert '--output gazetteer' in captured.out
  87     assert f'--style {osm2pgsql_options["osm2pgsql_style"]}' in captured.out
  88     assert f'--number-processes {osm2pgsql_options["threads"]}' in captured.out
  89     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
  90     assert 'file.pbf' in captured.out
  91
  92
  93 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options, capfd):
  94     table_factory('place', content=((1, ), ))
  95     osm2pgsql_options['osm2pgsql_cache'] = 0
  96
  97     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
  98     for f in files:
  99         f.write_text('test')
 100
 101     database_import.import_osm_data(files, osm2pgsql_options)
 102     captured = capfd.readouterr()
 103
 104     assert 'file1.osm' in captured.out
 105     assert 'file2.osm' in captured.out
 106
 107
 108 def test_import_osm_data_simple_no_data(table_factory, osm2pgsql_options):
 109     table_factory('place')
 110
 111     with pytest.raises(UsageError, match='No data imported'):
 112         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 113
 114
 115 def test_import_osm_data_simple_ignore_no_data(table_factory, osm2pgsql_options):
 116     table_factory('place')
 117
 118     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 119                                     ignore_errors=True)
 120
 121
 122 def test_import_osm_data_drop(table_factory, temp_db_cursor, tmp_path, osm2pgsql_options):
 123     table_factory('place', content=((1, ), ))
 124     table_factory('planet_osm_nodes')
 125
 126     flatfile = tmp_path / 'flatfile'
 127     flatfile.write_text('touch')
 128
 129     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 130
 131     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 132
 133     assert not flatfile.exists()
 134     assert not temp_db_cursor.table_exists('planet_osm_nodes')
 135
 136
 137 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 138     table_factory('place', content=((1, ), ))
 139
 140     osm2pgsql_options['osm2pgsql_cache'] = 0
 141
 142     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 143     captured = capfd.readouterr()
 144
 145     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 146
 147
 148 @pytest.mark.parametrize("with_search", (True, False))
 149 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 150     tables = ['placex', 'place_addressline', 'location_area',
 151               'location_area_country',
 152               'location_property_tiger', 'location_property_osmline',
 153               'location_postcode', 'location_road_23']
 154     if with_search:
 155         tables.append('search_name')
 156
 157     for table in tables:
 158         table_factory(table, content=((1, ), (2, ), (3, )))
 159         assert temp_db_cursor.table_rows(table) == 3
 160
 161     database_import.truncate_data_tables(temp_db_conn)
 162
 163     for table in tables:
 164         assert temp_db_cursor.table_rows(table) == 0
 165
 166
 167 @pytest.mark.parametrize("threads", (1, 5))
 168 @pytest.mark.asyncio
 169 async def test_load_data(dsn, place_row, placex_table, osmline_table,
 170                          temp_db_cursor, threads):
 171     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 172         temp_db_cursor.execute(pysql.SQL("""CREATE FUNCTION {} (src TEXT)
 173                                             RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 174                                          """).format(pysql.Identifier(func)))
 175     for oid in range(100, 130):
 176         place_row(osm_id=oid)
 177     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 178               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 179
 180     await database_import.load_data(dsn, threads)
 181
 182     assert temp_db_cursor.table_rows('placex') == 30
 183     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 184
 185
 186 class TestSetupSQL:
 187
 188     @pytest.fixture(autouse=True)
 189     def init_env(self, temp_db, tmp_path, def_config, sql_preprocessor_cfg):
 190         def_config.lib_dir.sql = tmp_path / 'sql'
 191         def_config.lib_dir.sql.mkdir()
 192
 193         self.config = def_config
 194
 195     def write_sql(self, fname, content):
 196         (self.config.lib_dir.sql / fname).write_text(content)
 197
 198     @pytest.mark.parametrize("reverse", [True, False])
 199     def test_create_tables(self, temp_db_conn, temp_db_cursor, reverse):
 200         self.write_sql('tables.sql',
 201                        """CREATE FUNCTION test() RETURNS bool
 202                           AS $$ SELECT {{db.reverse_only}} $$ LANGUAGE SQL""")
 203
 204         database_import.create_tables(temp_db_conn, self.config, reverse)
 205
 206         temp_db_cursor.scalar('SELECT test()') == reverse
 207
 208     def test_create_table_triggers(self, temp_db_conn, temp_db_cursor):
 209         self.write_sql('table-triggers.sql',
 210                        """CREATE FUNCTION test() RETURNS TEXT
 211                           AS $$ SELECT 'a'::text $$ LANGUAGE SQL""")
 212
 213         database_import.create_table_triggers(temp_db_conn, self.config)
 214
 215         temp_db_cursor.scalar('SELECT test()') == 'a'
 216
 217     def test_create_partition_tables(self, temp_db_conn, temp_db_cursor):
 218         self.write_sql('partition-tables.src.sql',
 219                        """CREATE FUNCTION test() RETURNS TEXT
 220                           AS $$ SELECT 'b'::text $$ LANGUAGE SQL""")
 221
 222         database_import.create_partition_tables(temp_db_conn, self.config)
 223
 224         temp_db_cursor.scalar('SELECT test()') == 'b'
 225
 226     @pytest.mark.parametrize("drop", [True, False])
 227     @pytest.mark.asyncio
 228     async def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop):
 229         self.write_sql('indices.sql',
 230                        """CREATE FUNCTION test() RETURNS bool
 231                           AS $$ SELECT {{drop}} $$ LANGUAGE SQL""")
 232
 233         await database_import.create_search_indices(temp_db_conn, self.config, drop)
 234
 235         temp_db_cursor.scalar('SELECT test()') == drop