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(place_row, osm2pgsql_options, capfd):
  82     place_row()
  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(place_row, tmp_path, osm2pgsql_options, capfd):
  96     place_row()
  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(place_row, osm2pgsql_options):
 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(place_table, osm2pgsql_options):
 116     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 117                                     ignore_errors=True)
 118
 119
 120 def test_import_osm_data_drop(place_row, table_factory, temp_db_cursor,
 121                               tmp_path, osm2pgsql_options):
 122     place_row()
 123     table_factory('planet_osm_nodes')
 124
 125     flatfile = tmp_path / 'flatfile'
 126     flatfile.write_text('touch', encoding='utf-8')
 127
 128     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 129
 130     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 131
 132     assert not flatfile.exists()
 133     assert not temp_db_cursor.table_exists('planet_osm_nodes')
 134
 135
 136 def test_import_osm_data_default_cache(place_row, osm2pgsql_options, capfd):
 137     place_row()
 138
 139     osm2pgsql_options['osm2pgsql_cache'] = 0
 140
 141     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 142     captured = capfd.readouterr()
 143
 144     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 145
 146
 147 @pytest.mark.parametrize("with_search", (True, False))
 148 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 149     tables = ['placex', 'place_addressline', 'location_area',
 150               'location_area_country',
 151               'location_property_tiger', 'location_property_osmline',
 152               'location_postcodes', 'location_road_23']
 153     if with_search:
 154         tables.append('search_name')
 155
 156     for table in tables:
 157         table_factory(table, content=((1, ), (2, ), (3, )))
 158         assert temp_db_cursor.table_rows(table) == 3
 159
 160     database_import.truncate_data_tables(temp_db_conn)
 161
 162     for table in tables:
 163         assert temp_db_cursor.table_rows(table) == 0
 164
 165
 166 @pytest.mark.parametrize("threads", (1, 5))
 167 @pytest.mark.asyncio
 168 async def test_load_data(dsn, place_row, placex_table, osmline_table,
 169                          temp_db_cursor, threads):
 170     for oid in range(100, 130):
 171         place_row(osm_id=oid)
 172     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 173               geom='LINESTRING(0 0, 10 10)')
 174
 175     temp_db_cursor.execute("""
 176         CREATE OR REPLACE FUNCTION placex_insert() RETURNS TRIGGER AS $$
 177         BEGIN
 178           NEW.place_id := nextval('seq_place');
 179           NEW.indexed_status := 1;
 180           NEW.centroid := ST_Centroid(NEW.geometry);
 181           NEW.partition := 0;
 182           NEW.geometry_sector := 2424;
 183           NEW.rank_address := 30;
 184           NEW.rank_search := 30;
 185         RETURN NEW;
 186         END; $$ LANGUAGE plpgsql STABLE PARALLEL SAFE;
 187
 188         CREATE OR REPLACE FUNCTION osmline_insert() RETURNS TRIGGER AS $$
 189         BEGIN
 190           NEW.place_id := nextval('seq_place');
 191           IF NEW.indexed_status IS NULL THEN
 192             NEW.indexed_status := 1;
 193             NEW.partition := 0;
 194             NEW.geometry_sector := 2424;
 195           END IF;
 196         RETURN NEW;
 197         END; $$ LANGUAGE plpgsql STABLE PARALLEL SAFE;
 198
 199         CREATE TRIGGER placex_before_insert BEFORE INSERT ON placex
 200         FOR EACH ROW EXECUTE PROCEDURE placex_insert();
 201
 202         CREATE TRIGGER osmline_before_insert BEFORE INSERT ON location_property_osmline
 203         FOR EACH ROW EXECUTE PROCEDURE osmline_insert();
 204     """)
 205
 206     await database_import.load_data(dsn, threads)
 207
 208     assert temp_db_cursor.table_rows('placex') == 30
 209     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 210
 211
 212 class TestSetupSQL:
 213
 214     @pytest.fixture(autouse=True)
 215     def osm2ppsql_skel(self, def_config, temp_db_with_extensions, place_row,
 216                        country_table, table_factory, temp_db_conn):
 217         self.config = def_config
 218         place_row()
 219         table_factory('osm2pgsql_properties', 'property TEXT, value TEXT',
 220                       (('db_format', 2),))
 221
 222         table_factory('planet_osm_rels', 'id BIGINT, members JSONB, tags JSONB')
 223         temp_db_conn.execute("""
 224             CREATE OR REPLACE FUNCTION planet_osm_member_ids(jsonb, character)
 225               RETURNS bigint[] AS $$
 226                 SELECT array_agg((el->>'ref')::int8)
 227                 FROM jsonb_array_elements($1) AS el WHERE el->>'type' = $2
 228               $$ LANGUAGE sql IMMUTABLE;
 229             """)
 230
 231     @pytest.mark.parametrize("reverse", [True, False])
 232     def test_create_tables(self, table_factory, temp_db_conn, temp_db_cursor, reverse):
 233         table_factory('country_osm_grid')
 234
 235         database_import.create_tables(temp_db_conn, self.config, reverse)
 236
 237         assert temp_db_cursor.table_exists('placex')
 238         assert not reverse == temp_db_cursor.table_exists('search_name')
 239
 240     def test_create_table_triggers(self, temp_db_conn, placex_table, osmline_table,
 241                                    postcode_table, load_sql):
 242         load_sql('functions.sql')
 243
 244         database_import.create_table_triggers(temp_db_conn, self.config)
 245
 246     def test_create_partition_tables(self, country_row, temp_db_conn, temp_db_cursor, load_sql):
 247         for i in range(3):
 248             country_row(partition=i)
 249         load_sql('tables/location_area.sql')
 250
 251         database_import.create_partition_tables(temp_db_conn, self.config)
 252
 253         for i in range(3):
 254             assert temp_db_cursor.table_exists(f"location_area_large_{i}")
 255             assert temp_db_cursor.table_exists(f"search_name_{i}")
 256
 257     @pytest.mark.parametrize("drop", [True, False])
 258     @pytest.mark.asyncio
 259     async def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop, load_sql):
 260         load_sql('tables.sql', 'functions/ranking.sql')
 261         await database_import.create_search_indices(temp_db_conn, self.config, drop)
 262
 263         assert temp_db_cursor.index_exists('placex', 'idx_placex_geometry')
 264         assert not drop == temp_db_cursor.index_exists('placex', 'idx_placex_geometry_buildings')