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) 2026 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(pysql.SQL('DROP DATABASE IF EXISTS ')
  29                             + pysql.Identifier(self.DBNAME))
  30
  31             yield True
  32
  33             with conn.cursor() as cur:
  34                 cur.execute(pysql.SQL('DROP DATABASE IF EXISTS ')
  35                             + pysql.Identifier(self.DBNAME))
  36
  37     @pytest.fixture
  38     def cursor(self):
  39         with psycopg.connect(dbname=self.DBNAME) as conn:
  40             with conn.cursor() as cur:
  41                 yield cur
  42
  43     def conn(self):
  44         return psycopg.connect(dbname=self.DBNAME)
  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     def test_unsupported_pg_version(self, monkeypatch):
  55         monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  56
  57         with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  58             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  59
  60     def test_create_db_explicit_ro_user(self):
  61         database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  62                                                 rouser='postgres')
  63
  64     def test_create_db_missing_ro_user(self):
  65         with pytest.raises(UsageError, match='Missing read-only user.'):
  66             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  67                                                     rouser='sdfwkjkjgdugu2jgsafkljas')
  68
  69     def test_setup_extensions_old_postgis(self, monkeypatch):
  70         monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  71
  72         with pytest.raises(UsageError, match='PostGIS is too old.'):
  73             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  74
  75
  76 def test_setup_skeleton_already_exists(temp_db):
  77     with pytest.raises(UsageError):
  78         database_import.setup_database_skeleton(f'dbname={temp_db}')
  79
  80
  81 def test_import_osm_data_simple(table_factory, osm2pgsql_options, capfd):
  82     table_factory('place', content=((1, ), ))
  83
  84     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
  85     captured = capfd.readouterr()
  86
  87     assert '--create' in captured.out
  88     assert '--output flex' in captured.out
  89     assert f'--style {osm2pgsql_options["osm2pgsql_style"]}' in captured.out
  90     assert f'--number-processes {osm2pgsql_options["threads"]}' in captured.out
  91     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
  92     assert 'file.pbf' in captured.out
  93
  94
  95 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options, capfd):
  96     table_factory('place', content=((1, ), ))
  97     osm2pgsql_options['osm2pgsql_cache'] = 0
  98
  99     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
 100     for f in files:
 101         f.write_text('test', encoding='utf-8')
 102
 103     database_import.import_osm_data(files, osm2pgsql_options)
 104     captured = capfd.readouterr()
 105
 106     assert 'file1.osm' in captured.out
 107     assert 'file2.osm' in captured.out
 108
 109
 110 def test_import_osm_data_simple_no_data(table_factory, osm2pgsql_options):
 111     table_factory('place')
 112
 113     with pytest.raises(UsageError, match='No data imported'):
 114         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 115
 116
 117 def test_import_osm_data_simple_ignore_no_data(table_factory, osm2pgsql_options):
 118     table_factory('place')
 119
 120     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 121                                     ignore_errors=True)
 122
 123
 124 def test_import_osm_data_drop(table_factory, temp_db_cursor, tmp_path, osm2pgsql_options):
 125     table_factory('place', content=((1, ), ))
 126     table_factory('planet_osm_nodes')
 127
 128     flatfile = tmp_path / 'flatfile'
 129     flatfile.write_text('touch', encoding='utf-8')
 130
 131     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 132
 133     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 134
 135     assert not flatfile.exists()
 136     assert not temp_db_cursor.table_exists('planet_osm_nodes')
 137
 138
 139 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 140     table_factory('place', content=((1, ), ))
 141
 142     osm2pgsql_options['osm2pgsql_cache'] = 0
 143
 144     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 145     captured = capfd.readouterr()
 146
 147     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 148
 149
 150 @pytest.mark.parametrize("with_search", (True, False))
 151 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 152     tables = ['placex', 'place_addressline', 'location_area',
 153               'location_area_country',
 154               'location_property_tiger', 'location_property_osmline',
 155               'location_postcodes', 'location_road_23']
 156     if with_search:
 157         tables.append('search_name')
 158
 159     for table in tables:
 160         table_factory(table, content=((1, ), (2, ), (3, )))
 161         assert temp_db_cursor.table_rows(table) == 3
 162
 163     database_import.truncate_data_tables(temp_db_conn)
 164
 165     for table in tables:
 166         assert temp_db_cursor.table_rows(table) == 0
 167
 168
 169 @pytest.mark.parametrize("threads", (1, 5))
 170 @pytest.mark.asyncio
 171 async def test_load_data(dsn, place_row, placex_table, osmline_table,
 172                          temp_db_cursor, threads):
 173     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 174         temp_db_cursor.execute(pysql.SQL("""CREATE FUNCTION {} (src TEXT)
 175                                             RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 176                                          """).format(pysql.Identifier(func)))
 177     for oid in range(100, 130):
 178         place_row(osm_id=oid)
 179     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 180               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 181
 182     await database_import.load_data(dsn, threads)
 183
 184     assert temp_db_cursor.table_rows('placex') == 30
 185     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 186
 187
 188 class TestSetupSQL:
 189
 190     @pytest.fixture(autouse=True)
 191     def init_env(self, temp_db, tmp_path, def_config, sql_preprocessor_cfg):
 192         def_config.lib_dir.sql = tmp_path / 'sql'
 193         def_config.lib_dir.sql.mkdir()
 194
 195         self.config = def_config
 196
 197     def write_sql(self, fname, content):
 198         (self.config.lib_dir.sql / fname).write_text(content, encoding='utf-8')
 199
 200     @pytest.mark.parametrize("reverse", [True, False])
 201     def test_create_tables(self, temp_db_conn, temp_db_cursor, reverse):
 202         self.write_sql('tables.sql',
 203                        """CREATE FUNCTION test() RETURNS bool
 204                           AS $$ SELECT {{db.reverse_only}} $$ LANGUAGE SQL""")
 205
 206         self.write_sql('grants.sql', "-- Mock grants file for testing\n")
 207
 208         database_import.create_tables(temp_db_conn, self.config, reverse)
 209
 210         temp_db_cursor.scalar('SELECT test()') == reverse
 211
 212     def test_create_table_triggers(self, temp_db_conn, temp_db_cursor):
 213         self.write_sql('table-triggers.sql',
 214                        """CREATE FUNCTION test() RETURNS TEXT
 215                           AS $$ SELECT 'a'::text $$ LANGUAGE SQL""")
 216
 217         database_import.create_table_triggers(temp_db_conn, self.config)
 218
 219         temp_db_cursor.scalar('SELECT test()') == 'a'
 220
 221     def test_create_partition_tables(self, temp_db_conn, temp_db_cursor):
 222         self.write_sql('partition-tables.src.sql',
 223                        """CREATE FUNCTION test() RETURNS TEXT
 224                           AS $$ SELECT 'b'::text $$ LANGUAGE SQL""")
 225
 226         database_import.create_partition_tables(temp_db_conn, self.config)
 227
 228         temp_db_cursor.scalar('SELECT test()') == 'b'
 229
 230     @pytest.mark.parametrize("drop", [True, False])
 231     @pytest.mark.asyncio
 232     async def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop):
 233         self.write_sql('indices.sql',
 234                        """CREATE FUNCTION test() RETURNS bool
 235                           AS $$ SELECT {{drop}} $$ LANGUAGE SQL""")
 236
 237         await database_import.create_search_indices(temp_db_conn, self.config, drop)
 238
 239         temp_db_cursor.scalar('SELECT test()') == drop