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 from contextlib import closing
  12
  13 import pytest
  14 import psycopg2
  15
  16 from nominatim_db.tools import database_import
  17 from nominatim_db.errors import UsageError
  18
  19 class TestDatabaseSetup:
  20     DBNAME = 'test_nominatim_python_unittest'
  21
  22     @pytest.fixture(autouse=True)
  23     def setup_nonexistant_db(self):
  24         conn = psycopg2.connect(database='postgres')
  25
  26         try:
  27             conn.set_isolation_level(0)
  28             with conn.cursor() as cur:
  29                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  30
  31             yield True
  32
  33             with conn.cursor() as cur:
  34                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  35         finally:
  36             conn.close()
  37
  38     @pytest.fixture
  39     def cursor(self):
  40         conn = psycopg2.connect(database=self.DBNAME)
  41
  42         try:
  43             with conn.cursor() as cur:
  44                 yield cur
  45         finally:
  46             conn.close()
  47
  48
  49     def conn(self):
  50         return closing(psycopg2.connect(database=self.DBNAME))
  51
  52
  53     def test_setup_skeleton(self):
  54         database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  55
  56         # Check that all extensions are set up.
  57         with self.conn() as conn:
  58             with conn.cursor() as cur:
  59                 cur.execute('CREATE TABLE t (h HSTORE, geom GEOMETRY(Geometry, 4326))')
  60
  61
  62     def test_unsupported_pg_version(self, monkeypatch):
  63         monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  64
  65         with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  66             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  67
  68
  69     def test_create_db_explicit_ro_user(self):
  70         database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  71                                                 rouser='postgres')
  72
  73
  74     def test_create_db_missing_ro_user(self):
  75         with pytest.raises(UsageError, match='Missing read-only user.'):
  76             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  77                                                     rouser='sdfwkjkjgdugu2;jgsafkljas;')
  78
  79
  80     def test_setup_extensions_old_postgis(self, monkeypatch):
  81         monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  82
  83         with pytest.raises(UsageError, match='PostGIS is too old.'):
  84             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  85
  86
  87 def test_setup_skeleton_already_exists(temp_db):
  88     with pytest.raises(UsageError):
  89         database_import.setup_database_skeleton(f'dbname={temp_db}')
  90
  91
  92 def test_import_osm_data_simple(table_factory, osm2pgsql_options, capfd):
  93     table_factory('place', content=((1, ), ))
  94
  95     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
  96     captured = capfd.readouterr()
  97
  98     assert '--create' in captured.out
  99     assert '--output gazetteer' in captured.out
 100     assert f'--style {osm2pgsql_options["osm2pgsql_style"]}' in captured.out
 101     assert f'--number-processes {osm2pgsql_options["threads"]}' in captured.out
 102     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 103     assert 'file.pbf' in captured.out
 104
 105
 106 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options, capfd):
 107     table_factory('place', content=((1, ), ))
 108     osm2pgsql_options['osm2pgsql_cache'] = 0
 109
 110     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
 111     for f in files:
 112         f.write_text('test')
 113
 114     database_import.import_osm_data(files, osm2pgsql_options)
 115     captured = capfd.readouterr()
 116
 117     assert 'file1.osm' in captured.out
 118     assert 'file2.osm' in captured.out
 119
 120
 121 def test_import_osm_data_simple_no_data(table_factory, osm2pgsql_options):
 122     table_factory('place')
 123
 124     with pytest.raises(UsageError, match='No data imported'):
 125         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 126
 127
 128 def test_import_osm_data_simple_ignore_no_data(table_factory, osm2pgsql_options):
 129     table_factory('place')
 130
 131     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 132                                     ignore_errors=True)
 133
 134
 135 def test_import_osm_data_drop(table_factory, temp_db_cursor, tmp_path, osm2pgsql_options):
 136     table_factory('place', content=((1, ), ))
 137     table_factory('planet_osm_nodes')
 138
 139     flatfile = tmp_path / 'flatfile'
 140     flatfile.write_text('touch')
 141
 142     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 143
 144     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 145
 146     assert not flatfile.exists()
 147     assert not temp_db_cursor.table_exists('planet_osm_nodes')
 148
 149
 150 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 151     table_factory('place', content=((1, ), ))
 152
 153     osm2pgsql_options['osm2pgsql_cache'] = 0
 154
 155     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 156     captured = capfd.readouterr()
 157
 158     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 159
 160
 161 @pytest.mark.parametrize("with_search", (True, False))
 162 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 163     tables = ['placex', 'place_addressline', 'location_area',
 164               'location_area_country',
 165               'location_property_tiger', 'location_property_osmline',
 166               'location_postcode', 'location_road_23']
 167     if with_search:
 168         tables.append('search_name')
 169
 170     for table in tables:
 171         table_factory(table, content=((1, ), (2, ), (3, )))
 172         assert temp_db_cursor.table_rows(table) == 3
 173
 174     database_import.truncate_data_tables(temp_db_conn)
 175
 176     for table in tables:
 177         assert temp_db_cursor.table_rows(table) == 0
 178
 179
 180 @pytest.mark.parametrize("threads", (1, 5))
 181 def test_load_data(dsn, place_row, placex_table, osmline_table,
 182                    temp_db_cursor, threads):
 183     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 184         temp_db_cursor.execute(f"""CREATE FUNCTION {func} (src TEXT)
 185                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 186                                """)
 187     for oid in range(100, 130):
 188         place_row(osm_id=oid)
 189     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 190               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 191
 192     database_import.load_data(dsn, threads)
 193
 194     assert temp_db_cursor.table_rows('placex') == 30
 195     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 196
 197
 198 class TestSetupSQL:
 199
 200     @pytest.fixture(autouse=True)
 201     def init_env(self, temp_db, tmp_path, def_config, sql_preprocessor_cfg):
 202         def_config.lib_dir.sql = tmp_path / 'sql'
 203         def_config.lib_dir.sql.mkdir()
 204
 205         self.config = def_config
 206
 207
 208     def write_sql(self, fname, content):
 209         (self.config.lib_dir.sql / fname).write_text(content)
 210
 211
 212     @pytest.mark.parametrize("reverse", [True, False])
 213     def test_create_tables(self, temp_db_conn, temp_db_cursor, reverse):
 214         self.write_sql('tables.sql',
 215                        """CREATE FUNCTION test() RETURNS bool
 216                           AS $$ SELECT {{db.reverse_only}} $$ LANGUAGE SQL""")
 217
 218         database_import.create_tables(temp_db_conn, self.config, reverse)
 219
 220         temp_db_cursor.scalar('SELECT test()') == reverse
 221
 222
 223     def test_create_table_triggers(self, temp_db_conn, temp_db_cursor):
 224         self.write_sql('table-triggers.sql',
 225                        """CREATE FUNCTION test() RETURNS TEXT
 226                           AS $$ SELECT 'a'::text $$ LANGUAGE SQL""")
 227
 228         database_import.create_table_triggers(temp_db_conn, self.config)
 229
 230         temp_db_cursor.scalar('SELECT test()') == 'a'
 231
 232
 233     def test_create_partition_tables(self, temp_db_conn, temp_db_cursor):
 234         self.write_sql('partition-tables.src.sql',
 235                        """CREATE FUNCTION test() RETURNS TEXT
 236                           AS $$ SELECT 'b'::text $$ LANGUAGE SQL""")
 237
 238         database_import.create_partition_tables(temp_db_conn, self.config)
 239
 240         temp_db_cursor.scalar('SELECT test()') == 'b'
 241
 242
 243     @pytest.mark.parametrize("drop", [True, False])
 244     def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop):
 245         self.write_sql('indices.sql',
 246                        """CREATE FUNCTION test() RETURNS bool
 247                           AS $$ SELECT {{drop}} $$ LANGUAGE SQL""")
 248
 249         database_import.create_search_indices(temp_db_conn, self.config, drop)
 250
 251         temp_db_cursor.scalar('SELECT test()') == drop