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_import_base_data(src_dir, temp_db, temp_db_cursor):
  84     temp_db_cursor.execute('CREATE EXTENSION hstore')
  85     temp_db_cursor.execute('CREATE EXTENSION postgis')
  86     database_import.import_base_data('dbname=' + temp_db, src_dir / 'data')
  87
  88     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name') > 0
  89
  90
  91 def test_import_base_data_ignore_partitions(src_dir, temp_db, temp_db_cursor):
  92     temp_db_cursor.execute('CREATE EXTENSION hstore')
  93     temp_db_cursor.execute('CREATE EXTENSION postgis')
  94     database_import.import_base_data('dbname=' + temp_db, src_dir / 'data',
  95                                      ignore_partitions=True)
  96
  97     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name') > 0
  98     assert temp_db_cursor.scalar('SELECT count(*) FROM country_name WHERE partition != 0') == 0
  99
 100
 101 def test_import_osm_data_simple(temp_db_cursor,osm2pgsql_options):
 102     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 103     temp_db_cursor.execute('INSERT INTO place values (1)')
 104
 105     database_import.import_osm_data('file.pdf', osm2pgsql_options)
 106
 107
 108 def test_import_osm_data_simple_no_data(temp_db_cursor,osm2pgsql_options):
 109     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 110
 111     with pytest.raises(UsageError, match='No data.*'):
 112         database_import.import_osm_data('file.pdf', osm2pgsql_options)
 113
 114
 115 def test_import_osm_data_drop(temp_db_conn, temp_db_cursor, tmp_path, osm2pgsql_options):
 116     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 117     temp_db_cursor.execute('CREATE TABLE planet_osm_nodes (id INT)')
 118     temp_db_cursor.execute('INSERT INTO place values (1)')
 119
 120     flatfile = tmp_path / 'flatfile'
 121     flatfile.write_text('touch')
 122
 123     osm2pgsql_options['flatnode_file'] = str(flatfile.resolve())
 124
 125     database_import.import_osm_data('file.pdf', osm2pgsql_options, drop=True)
 126
 127     assert not flatfile.exists()
 128     assert not temp_db_conn.table_exists('planet_osm_nodes')
 129
 130
 131 def test_import_osm_data_default_cache(temp_db_cursor,osm2pgsql_options):
 132     temp_db_cursor.execute('CREATE TABLE place (id INT)')
 133     temp_db_cursor.execute('INSERT INTO place values (1)')
 134
 135     osm2pgsql_options['osm2pgsql_cache'] = 0
 136
 137     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
 138
 139
 140 def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory):
 141     tables = ('placex', 'place_addressline', 'location_area',
 142               'location_area_country',
 143               'location_property_tiger', 'location_property_osmline',
 144               'location_postcode', 'search_name', 'location_road_23')
 145     for table in tables:
 146         table_factory(table, content=((1, ), (2, ), (3, )))
 147         assert temp_db_cursor.table_rows(table) == 3
 148
 149     database_import.truncate_data_tables(temp_db_conn)
 150
 151     for table in tables:
 152         assert temp_db_cursor.table_rows(table) == 0
 153
 154
 155 @pytest.mark.parametrize("threads", (1, 5))
 156 def test_load_data(dsn, src_dir, place_row, placex_table, osmline_table,
 157                    word_table, temp_db_cursor, threads):
 158     for func in ('precompute_words', 'getorcreate_housenumber_id', 'make_standard_name'):
 159         temp_db_cursor.execute("""CREATE FUNCTION {} (src TEXT)
 160                                   RETURNS TEXT AS $$ SELECT 'a'::TEXT $$ LANGUAGE SQL
 161                                """.format(func))
 162     for oid in range(100, 130):
 163         place_row(osm_id=oid)
 164     place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
 165               geom='SRID=4326;LINESTRING(0 0, 10 10)')
 166
 167     database_import.load_data(dsn, threads)
 168
 169     assert temp_db_cursor.table_rows('placex') == 30
 170     assert temp_db_cursor.table_rows('location_property_osmline') == 1
 171
 172
 173 @pytest.mark.parametrize("languages", (None, ' fr,en'))
 174 def test_create_country_names(temp_db_with_extensions, temp_db_conn, temp_db_cursor,
 175                               table_factory, tokenizer_mock, languages):
 176
 177     table_factory('country_name', 'country_code varchar(2), name hstore',
 178                   content=(('us', '"name"=>"us1","name:af"=>"us2"'),
 179                            ('fr', '"name"=>"Fra", "name:en"=>"Fren"')))
 180
 181     assert temp_db_cursor.scalar("SELECT count(*) FROM country_name") == 2
 182
 183     tokenizer = tokenizer_mock()
 184
 185     database_import.create_country_names(temp_db_conn, tokenizer, languages)
 186
 187     assert len(tokenizer.analyser_cache['countries']) == 2
 188
 189     result_set = {k: set(v) for k, v in tokenizer.analyser_cache['countries']}
 190
 191     if languages:
 192         assert result_set == {'us' : set(('us', 'us1', 'United States')),
 193                               'fr' : set(('fr', 'Fra', 'Fren'))}
 194     else:
 195         assert result_set == {'us' : set(('us', 'us1', 'us2', 'United States')),
 196                               'fr' : set(('fr', 'Fra', 'Fren'))}