test/python/test_tools_database_import.py

   1 """
   2 Tests for functions to import a new database.
   3 """
   4 from pathlib import Path
   5
   6 import pytest
   7 import psycopg2
   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, table_factory):
  71     database_import.setup_extensions(temp_db_conn)
  72
  73     # Use table creation to check that hstore and geometry types are available.
  74     table_factory('t', 'h HSTORE, geom GEOMETRY(Geometry, 4326)')
  75
  76
  77 def test_setup_extensions_old_postgis(temp_db_conn, monkeypatch):
  78     monkeypatch.setattr(database_import, 'POSTGIS_REQUIRED_VERSION', (50, 50))
  79
  80     with pytest.raises(UsageError, match='PostGIS version is too old.'):
  81         database_import.setup_extensions(temp_db_conn)
  82
  83
  84 def test_import_base_data(dsn, src_dir, temp_db_with_extensions, temp_db_cursor):
  85     database_import.import_base_data(dsn, src_dir / 'data')
  86
  87     assert temp_db_cursor.table_rows('country_name') > 0
  88
  89
  90 def test_import_base_data_ignore_partitions(dsn, src_dir, temp_db_with_extensions,
  91                                             temp_db_cursor):
  92     database_import.import_base_data(dsn, src_dir / 'data', ignore_partitions=True)
  93
  94     assert temp_db_cursor.table_rows('country_name') > 0
  95     assert temp_db_cursor.table_rows('country_name', where='partition != 0') == 0
  96
  97
  98 def test_import_osm_data_simple(table_factory, osm2pgsql_options):
  99     table_factory('place', content=((1, ), ))
 100
 101     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 102
 103
 104 def test_import_osm_data_multifile(table_factory, tmp_path, osm2pgsql_options):
 105     table_factory('place', content=((1, ), ))
 106     osm2pgsql_options['osm2pgsql_cache'] = 0
 107
 108     files = [tmp_path / 'file1.osm', tmp_path / 'file2.osm']
 109     for f in files:
 110         f.write_text('test')
 111
 112     database_import.import_osm_data(files, osm2pgsql_options)
 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.*'):
 119         database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options)
 120
 121
 122 def test_import_osm_data_drop(table_factory, temp_db_conn, tmp_path, osm2pgsql_options):
 123     table_factory('place', content=((1, ), ))
 124     table_factory('planet_osm_nodes')
 125
 126     flatfile = tmp_path / 'flatfile'
 127     flatfile.write_text('touch')
 128
 129     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 130
 131     database_import.import_osm_data(Path('file.pbf'), osm2pgsql_options, drop=True)
 132
 133     assert not flatfile.exists()
 134     assert not temp_db_conn.table_exists('planet_osm_nodes')
 135
 136
 137 def test_import_osm_data_default_cache(table_factory, osm2pgsql_options):
 138     table_factory('place', content=((1, ), ))
 139
 140     osm2pgsql_options['osm2pgsql_cache'] = 0
 141
 142     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 143
 144
 145 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory):
 146     tables = ('placex', 'place_addressline', 'location_area',
 147               'location_area_country',
 148               'location_property_tiger', 'location_property_osmline',
 149               'location_postcode', 'search_name', 'location_road_23')
 150     for table in tables:
 151         table_factory(table, content=((1, ), (2, ), (3, )))
 152         assert temp_db_cursor.table_rows(table) == 3
 153
 154     database_import.truncate_data_tables(temp_db_conn)
 155
 156     for table in tables:
 157         assert temp_db_cursor.table_rows(table) == 0
 158
 159
 160 @pytest.mark.parametrize("threads", (1, 5))
 161 def test_load_data(dsn, place_row, placex_table, osmline_table,
 162                    word_table, temp_db_cursor, threads):
 163     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 164         temp_db_cursor.execute("""CREATE FUNCTION {} (src TEXT)
 165                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 166                                """.format(func))
 167     for oid in range(100, 130):
 168         place_row(osm_id=oid)
 169     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 170               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 171
 172     database_import.load_data(dsn, threads)
 173
 174     assert temp_db_cursor.table_rows('placex') == 30
 175     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 176
 177
 178 @pytest.mark.parametrize("languages", (None, ' fr,en'))
 179 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
 180                               table_factory, tokenizer_mock, languages):
 181
 182     table_factory('country_name', 'country_code varchar(2), name hstore',
 183                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
 184                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
 185
 186     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
 187
 188     tokenizer = tokenizer_mock()
 189
 190     database_import.create_country_names(temp_db_conn, tokenizer, languages)
 191
 192     assert len(tokenizer.analyser_cache['countries']) == 2
 193
 194     result_set = {k: set(v.values()) for k, v in tokenizer.analyser_cache['countries']}
 195
 196     if languages:
 197         assert result_set == {'us' : set(('us', 'us1', 'United States')),
 198                               'fr' : set(('fr', 'Fra', 'Fren'))}
 199     else:
 200         assert result_set == {'us' : set(('us', 'us1', 'us2', 'United States')),
 201                               'fr' : set(('fr', 'Fra', 'Fren'))}