test/python/tools/test_database_import.py

   1 """
   2 Tests for functions to import a new database.
   3 """
   4 from pathlib import Path
   5 from contextlib import closing
   6
   7 import pytest
   8 import psycopg2
   9
  10 from nominatim.tools import database_import
  11 from nominatim.errors import UsageError
  12
  13 class TestDatabaseSetup:
  14     DBNAME = 'test_nominatim_python_unittest'
  15
  16     @pytest.fixture(autouse=True)
  17     def setup_nonexistant_db(self):
  18         conn = psycopg2.connect(database='postgres')
  19
  20         try:
  21             conn.set_isolation_level(0)
  22             with conn.cursor() as cur:
  23                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  24
  25             yield True
  26
  27             with conn.cursor() as cur:
  28                 cur.execute(f'DROP DATABASE IF EXISTS {self.DBNAME}')
  29         finally:
  30             conn.close()
  31
  32     @pytest.fixture
  33     def cursor(self):
  34         conn = psycopg2.connect(database=self.DBNAME)
  35
  36         try:
  37             with conn.cursor() as cur:
  38                 yield cur
  39         finally:
  40             conn.close()
  41
  42
  43     def conn(self):
  44         return closing(psycopg2.connect(database=self.DBNAME))
  45
  46
  47     def test_setup_skeleton(self):
  48         database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  49
  50         # Check that all extensions are set up.
  51         with self.conn() as conn:
  52             with conn.cursor() as cur:
  53                 cur.execute('CREATE TABLE t (h HSTORE, geom GEOMETRY(Geometry, 4326))')
  54
  55
  56     def test_unsupported_pg_version(self, monkeypatch):
  57         monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  58
  59         with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  60             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  61
  62
  63     def test_create_db_explicit_ro_user(self):
  64         database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  65                                                 rouser='postgres')
  66
  67
  68     def test_create_db_missing_ro_user(self):
  69         with pytest.raises(UsageError, match='Missing read-only user.'):
  70             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  71                                                     rouser='sdfwkjkjgdugu2;jgsafkljas;')
  72
  73
  74     def test_setup_extensions_old_postgis(self, monkeypatch):
  75         monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  76
  77         with pytest.raises(UsageError, match='PostGIS is too old.'):
  78             database_import.setup_database_skeleton(f'dbname={self.DBNAME}')
  79
  80
  81 def test_setup_skeleton_already_exists(temp_db):
  82     with pytest.raises(UsageError):
  83         database_import.setup_database_skeleton(f'dbname={temp_db}')
  84
  85
  86 def test_import_osm_data_simple(table_factory, osm2pgsql_options, capfd):
  87     table_factory('place', content=((1, ), ))
  88
  89     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
  90     captured = capfd.readouterr()
  91
  92     assert '--create' in captured.out
  93     assert '--output gazetteer' in captured.out
  94     assert f'--style {osm2pgsql_options["osm2pgsql_style"]}' in captured.out
  95     assert f'--number-processes {osm2pgsql_options["threads"]}' in captured.out
  96     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
  97     assert 'file.pbf' in captured.out
  98
  99
 100 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options, capfd):
 101     table_factory('place', content=((1, ), ))
 102     osm2pgsql_options['osm2pgsql_cache'] = 0
 103
 104     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
 105     for f in files:
 106         f.write_text('test')
 107
 108     database_import.import_osm_data(files, osm2pgsql_options)
 109     captured = capfd.readouterr()
 110
 111     assert 'file1.osm' in captured.out
 112     assert 'file2.osm' in captured.out
 113
 114
 115 def test_import_osm_data_simple_no_data(table_factory, osm2pgsql_options):
 116     table_factory('place')
 117
 118     with pytest.raises(UsageError, match='No data imported'):
 119         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 120
 121
 122 def test_import_osm_data_simple_ignore_no_data(table_factory, osm2pgsql_options):
 123     table_factory('place')
 124
 125     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options,
 126                                     ignore_errors=True)
 127
 128
 129 def test_import_osm_data_drop(table_factory, temp_db_conn, tmp_path, osm2pgsql_options):
 130     table_factory('place', content=((1, ), ))
 131     table_factory('planet_osm_nodes')
 132
 133     flatfile = tmp_path / 'flatfile'
 134     flatfile.write_text('touch')
 135
 136     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 137
 138     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 139
 140     assert not flatfile.exists()
 141     assert not temp_db_conn.table_exists('planet_osm_nodes')
 142
 143
 144 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 145     table_factory('place', content=((1, ), ))
 146
 147     osm2pgsql_options['osm2pgsql_cache'] = 0
 148
 149     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 150     captured = capfd.readouterr()
 151
 152     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 153
 154
 155 @pytest.mark.parametrize("with_search", (True, False))
 156 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory, with_search):
 157     tables = ['placex', 'place_addressline', 'location_area',
 158               'location_area_country',
 159               'location_property_tiger', 'location_property_osmline',
 160               'location_postcode', 'location_road_23']
 161     if with_search:
 162         tables.append('search_name')
 163
 164     for table in tables:
 165         table_factory(table, content=((1, ), (2, ), (3, )))
 166         assert temp_db_cursor.table_rows(table) == 3
 167
 168     database_import.truncate_data_tables(temp_db_conn)
 169
 170     for table in tables:
 171         assert temp_db_cursor.table_rows(table) == 0
 172
 173
 174 @pytest.mark.parametrize("threads", (1, 5))
 175 def test_load_data(dsn, place_row, placex_table, osmline_table,
 176                    word_table, temp_db_cursor, threads):
 177     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 178         temp_db_cursor.execute(f"""CREATE FUNCTION {func} (src TEXT)
 179                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 180                                """)
 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     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     def test_create_search_indices(self, temp_db_conn, temp_db_cursor, drop):
 239         self.write_sql('indices.sql',
 240                        """CREATE FUNCTION test() RETURNS bool
 241                           AS $$ SELECT {{drop}} $$ LANGUAGE SQL""")
 242
 243         database_import.create_search_indices(temp_db_conn, self.config, drop)
 244
 245         temp_db_cursor.scalar('SELECT test()') == drop