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_missing_ro_user(self):
  64         with pytest.raises(UsageError, match='Missing read-only user.'):
  65             database_import.setup_database_skeleton(f'dbname={self.DBNAME}',
  66                                                     rouser='sdfwkjkjgdugu2;jgsafkljas;')
  67
  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 gazetteer' 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')
 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.*'):
 114         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 115
 116
 117 def test_import_osm_data_drop(table_factory, temp_db_conn, tmp_path, osm2pgsql_options):
 118     table_factory('place', content=((1, ), ))
 119     table_factory('planet_osm_nodes')
 120
 121     flatfile = tmp_path / 'flatfile'
 122     flatfile.write_text('touch')
 123
 124     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 125
 126     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 127
 128     assert not flatfile.exists()
 129     assert not temp_db_conn.table_exists('planet_osm_nodes')
 130
 131
 132 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options, capfd):
 133     table_factory('place', content=((1, ), ))
 134
 135     osm2pgsql_options['osm2pgsql_cache'] = 0
 136
 137     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 138     captured = capfd.readouterr()
 139
 140     assert f'--cache {osm2pgsql_options["osm2pgsql_cache"]}' in captured.out
 141
 142
 143 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory):
 144     tables = ('placex', 'place_addressline', 'location_area',
 145               'location_area_country',
 146               'location_property_tiger', 'location_property_osmline',
 147               'location_postcode', 'search_name', 'location_road_23')
 148     for table in tables:
 149         table_factory(table, content=((1, ), (2, ), (3, )))
 150         assert temp_db_cursor.table_rows(table) == 3
 151
 152     database_import.truncate_data_tables(temp_db_conn)
 153
 154     for table in tables:
 155         assert temp_db_cursor.table_rows(table) == 0
 156
 157
 158 @pytest.mark.parametrize("threads", (1, 5))
 159 def test_load_data(dsn, place_row, placex_table, osmline_table,
 160                    word_table, temp_db_cursor, threads):
 161     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 162         temp_db_cursor.execute("""CREATE FUNCTION {} (src TEXT)
 163                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 164                                """.format(func))
 165     for oid in range(100, 130):
 166         place_row(osm_id=oid)
 167     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 168               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 169
 170     database_import.load_data(dsn, threads)
 171
 172     assert temp_db_cursor.table_rows('placex') == 30
 173     assert temp_db_cursor.table_rows('location_property_osmline') == 1