test/python/test_tools_database_import.py

   1 """
   2 Tests for functions to import a new database.
   3 """
   4 import pytest
   5 import psycopg2
   6 import sys
   7 from pathlib import Path
   8
   9 from nominatim.tools import database_import
  10 from nominatim.errors import UsageError
  11
  12 @pytest.fixture
  13 def nonexistant_db():
  14     dbname = 'test_nominatim_python_unittest'
  15
  16     conn = psycopg2.connect(database='postgres')
  17
  18     conn.set_isolation_level(0)
  19     with conn.cursor() as cur:
  20         cur.execute('DROP DATABASE IF EXISTS {}'.format(dbname))
  21
  22     yield dbname
  23
  24     with conn.cursor() as cur:
  25         cur.execute('DROP DATABASE IF EXISTS {}'.format(dbname))
  26
  27 @pytest.mark.parametrize("no_partitions", (True, False))
  28 def test_setup_skeleton(src_dir, nonexistant_db, no_partitions):
  29     database_import.setup_database_skeleton('dbname=' + nonexistant_db,
  30                                             src_dir / 'data', no_partitions)
  31
  32     conn = psycopg2.connect(database=nonexistant_db)
  33
  34     try:
  35         with conn.cursor() as cur:
  36             cur.execute("SELECT distinct partition FROM country_name")
  37             partitions = set([r[0] for r in list(cur)])
  38             if no_partitions:
  39                 assert partitions == set([0])
  40             else:
  41                 assert len(partitions) > 10
  42     finally:
  43         conn.close()
  44
  45
  46 def test_create_db_success(nonexistant_db):
  47     database_import.create_db('dbname=' + nonexistant_db, rouser='www-data')
  48
  49     conn = psycopg2.connect(database=nonexistant_db)
  50     conn.close()
  51
  52
  53 def test_create_db_already_exists(temp_db):
  54     with pytest.raises(UsageError):
  55         database_import.create_db('dbname=' + temp_db)
  56
  57
  58 def test_create_db_unsupported_version(nonexistant_db, monkeypatch):
  59     monkeypatch.setattr(database_import, 'POSTGRESQL_REQUIRED_VERSION', (100, 4))
  60
  61     with pytest.raises(UsageError, match='PostgreSQL server is too old.'):
  62         database_import.create_db('dbname=' + nonexistant_db)
  63
  64
  65 def test_create_db_missing_ro_user(nonexistant_db):
  66     with pytest.raises(UsageError, match='Missing read-only user.'):
  67         database_import.create_db('dbname=' + nonexistant_db, rouser='sdfwkjkjgdugu2;jgsafkljas;')
  68
  69
  70 def test_setup_extensions(temp_db_conn, temp_db_cursor):
  71     database_import.setup_extensions(temp_db_conn)
  72
  73     temp_db_cursor.execute('CREATE TABLE t (h HSTORE, geom GEOMETRY(Geometry, 4326))')
  74
  75
  76 def test_setup_extensions_old_postgis(temp_db_conn, monkeypatch):
  77     monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  78
  79     with pytest.raises(UsageError, match='PostGIS version is too old.'):
  80         database_import.setup_extensions(temp_db_conn)
  81
  82
  83 def test_install_module(tmp_path):
  84     src_dir = tmp_path / 'source'
  85     src_dir.mkdir()
  86     (src_dir / 'nominatim.so').write_text('TEST nomiantim.so')
  87
  88     project_dir = tmp_path / 'project'
  89     project_dir.mkdir()
  90
  91     database_import.install_module(src_dir, project_dir, '')
  92
  93     outfile = project_dir / 'module' / 'nominatim.so'
  94
  95     assert outfile.exists()
  96     assert outfile.read_text() == 'TEST nomiantim.so'
  97     assert outfile.stat().st_mode == 33261
  98
  99
 100 def test_install_module_custom(tmp_path):
 101     (tmp_path / 'nominatim.so').write_text('TEST nomiantim.so')
 102
 103     database_import.install_module(tmp_path, tmp_path, str(tmp_path.resolve()))
 104
 105     assert not (tmp_path / 'module').exists()
 106
 107
 108 def test_install_module_fail_access(temp_db_conn, tmp_path):
 109     (tmp_path / 'nominatim.so').write_text('TEST nomiantim.so')
 110
 111     with pytest.raises(UsageError, match='.*module cannot be accessed.*'):
 112         database_import.install_module(tmp_path, tmp_path, '',
 113                                        conn=temp_db_conn)
 114
 115
 116 def test_import_base_data(src_dir, temp_db, temp_db_cursor):
 117     temp_db_cursor.execute('CREATE EXTENSION hstore')
 118     temp_db_cursor.execute('CREATE EXTENSION postgis')
 119     database_import.import_base_data('dbname=' + temp_db, src_dir / 'data')
 120
 121     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name') > 0
 122
 123
 124 def test_import_base_data_ignore_partitions(src_dir, temp_db, temp_db_cursor):
 125     temp_db_cursor.execute('CREATE EXTENSION hstore')
 126     temp_db_cursor.execute('CREATE EXTENSION postgis')
 127     database_import.import_base_data('dbname=' + temp_db, src_dir / 'data',
 128                                      ignore_partitions=True)
 129
 130     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name') > 0
 131     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name WHERE partition != 0') == 0
 132
 133
 134 def test_import_osm_data_simple(temp_db_cursor,osm2pgsql_options):
 135     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 136     temp_db_cursor.execute('INSERT INTO place values (1)')
 137
 138     database_import.import_osm_data('file.pdf', osm2pgsql_options)
 139
 140
 141 def test_import_osm_data_simple_no_data(temp_db_cursor,osm2pgsql_options):
 142     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 143
 144     with pytest.raises(UsageError, match='No data.*'):
 145         database_import.import_osm_data('file.pdf', osm2pgsql_options)
 146
 147
 148 def test_import_osm_data_drop(temp_db_conn, temp_db_cursor, tmp_path, osm2pgsql_options):
 149     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 150     temp_db_cursor.execute('CREATE TABLE planet_osm_nodes (id INT)')
 151     temp_db_cursor.execute('INSERT INTO place values (1)')
 152
 153     flatfile = tmp_path / 'flatfile'
 154     flatfile.write_text('touch')
 155
 156     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 157
 158     database_import.import_osm_data('file.pdf', osm2pgsql_options, drop=True)
 159
 160     assert not flatfile.exists()
 161     assert not temp_db_conn.table_exists('planet_osm_nodes')
 162
 163
 164 def test_import_osm_data_default_cache(temp_db_cursor,osm2pgsql_options):
 165     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 166     temp_db_cursor.execute('INSERT INTO place values (1)')
 167
 168     osm2pgsql_options['osm2pgsql_cache'] = 0
 169
 170     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 171
 172
 173 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory):
 174     tables = ('word', 'placex', 'place_addressline', 'location_area',
 175               'location_area_country', 'location_property',
 176               'location_property_tiger', 'location_property_osmline',
 177               'location_postcode', 'search_name', 'location_road_23')
 178     for table in tables:
 179         table_factory(table, content=(1, 2, 3))
 180
 181     database_import.truncate_data_tables(temp_db_conn, max_word_frequency=23)
 182
 183     for table in tables:
 184         assert temp_db_cursor.table_rows(table) == 0
 185
 186
 187 @pytest.mark.parametrize("threads", (1, 5))
 188 def test_load_data(dsn, src_dir, place_row, placex_table, osmline_table, word_table,
 189                    temp_db_cursor, threads):
 190     for func in ('make_keywords', 'getorcreate_housenumber_id', 'make_standard_name'):
 191         temp_db_cursor.execute("""CREATE FUNCTION {} (src TEXT)
 192                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 193                                """.format(func))
 194     for oid in range(100, 130):
 195         place_row(osm_id=oid)
 196     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 197               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 198
 199     database_import.load_data(dsn, src_dir / 'data', threads)
 200
 201     assert temp_db_cursor.table_rows('placex') == 30
 202     assert temp_db_cursor.table_rows('location_property_osmline') == 1