test/python/test_tools_replication.py

   1 """
   2 Tests for replication functionality.
   3 """
   4 import datetime as dt
   5 import time
   6
   7 import pytest
   8 from osmium.replication.server import OsmosisState
   9
  10 import nominatim.tools.replication
  11 import nominatim.db.status as status
  12 from nominatim.errors import UsageError
  13
  14 OSM_NODE_DATA = """\
  15 <osm version="0.6" generator="OpenStreetMap server" copyright="OpenStreetMap and contributors" attribution="http://www.openstreetmap.org/copyright" license="http://opendatacommons.org/licenses/odbl/1-0/">
  16 <node id="100" visible="true" version="1" changeset="2047" timestamp="2006-01-27T22:09:10Z" user="Foo" uid="111" lat="48.7586670" lon="8.1343060">
  17 </node>
  18 </osm>
  19 """
  20
  21 ### init replication
  22
  23 def test_init_replication_bad_base_url(monkeypatch, status_table, place_row, temp_db_conn):
  24     place_row(osm_type='N', osm_id=100)
  25
  26     monkeypatch.setattr(nominatim.db.status, "get_url", lambda u : OSM_NODE_DATA)
  27
  28     with pytest.raises(UsageError, match="Failed to reach replication service"):
  29         nominatim.tools.replication.init_replication(temp_db_conn, 'https://test.io')
  30
  31
  32 def test_init_replication_success(monkeypatch, status_table, place_row, temp_db_conn, temp_db_cursor):
  33     place_row(osm_type='N', osm_id=100)
  34
  35     monkeypatch.setattr(nominatim.db.status, "get_url", lambda u : OSM_NODE_DATA)
  36     monkeypatch.setattr(nominatim.tools.replication.ReplicationServer,
  37                         "timestamp_to_sequence",
  38                         lambda self, date: 234)
  39
  40     nominatim.tools.replication.init_replication(temp_db_conn, 'https://test.io')
  41
  42     expected_date = dt.datetime.strptime('2006-01-27T19:09:10', status.ISODATE_FORMAT)\
  43                         .replace(tzinfo=dt.timezone.utc)
  44
  45     assert temp_db_cursor.row_set("SELECT * FROM import_status") \
  46              == {(expected_date, 234, True)}
  47
  48
  49 ### checking for updates
  50
  51 def test_check_for_updates_empty_status_table(status_table, temp_db_conn):
  52     assert nominatim.tools.replication.check_for_updates(temp_db_conn, 'https://test.io') == 254
  53
  54
  55 def test_check_for_updates_seq_not_set(status_table, temp_db_conn):
  56     status.set_status(temp_db_conn, dt.datetime.now(dt.timezone.utc))
  57
  58     assert nominatim.tools.replication.check_for_updates(temp_db_conn, 'https://test.io') == 254
  59
  60
  61 def test_check_for_updates_no_state(monkeypatch, status_table, temp_db_conn):
  62     status.set_status(temp_db_conn, dt.datetime.now(dt.timezone.utc), seq=345)
  63
  64     monkeypatch.setattr(nominatim.tools.replication.ReplicationServer,
  65                         "get_state_info", lambda self: None)
  66
  67     assert nominatim.tools.replication.check_for_updates(temp_db_conn, 'https://test.io') == 253
  68
  69
  70 @pytest.mark.parametrize("server_sequence,result", [(344, 2), (345, 2), (346, 0)])
  71 def test_check_for_updates_no_new_data(monkeypatch, status_table, temp_db_conn,
  72                                        server_sequence, result):
  73     date = dt.datetime.now(dt.timezone.utc)
  74     status.set_status(temp_db_conn, date, seq=345)
  75
  76     monkeypatch.setattr(nominatim.tools.replication.ReplicationServer,
  77                         "get_state_info",
  78                         lambda self: OsmosisState(server_sequence, date))
  79
  80     assert nominatim.tools.replication.check_for_updates(temp_db_conn, 'https://test.io') == result
  81
  82
  83 ### updating
  84
  85 @pytest.fixture
  86 def update_options(tmpdir):
  87     return dict(base_url='https://test.io',
  88                    indexed_only=False,
  89                    update_interval=3600,
  90                    import_file=tmpdir / 'foo.osm',
  91                    max_diff_size=1)
  92
  93 def test_update_empty_status_table(status_table, temp_db_conn):
  94     with pytest.raises(UsageError):
  95         nominatim.tools.replication.update(temp_db_conn, {})
  96
  97
  98 def test_update_already_indexed(status_table, temp_db_conn):
  99     status.set_status(temp_db_conn, dt.datetime.now(dt.timezone.utc), seq=34, indexed=False)
 100
 101     assert nominatim.tools.replication.update(temp_db_conn, dict(indexed_only=True)) \
 102              == nominatim.tools.replication.UpdateState.MORE_PENDING
 103
 104
 105 def test_update_no_data_no_sleep(monkeypatch, status_table, temp_db_conn, update_options):
 106     date = dt.datetime.now(dt.timezone.utc) - dt.timedelta(days=1)
 107     status.set_status(temp_db_conn, date, seq=34)
 108
 109     monkeypatch.setattr(nominatim.tools.replication.ReplicationServer,
 110                         "apply_diffs",
 111                         lambda *args, **kwargs: None)
 112
 113     sleeptime = []
 114     monkeypatch.setattr(time, 'sleep', lambda s: sleeptime.append(s))
 115
 116     assert nominatim.tools.replication.update(temp_db_conn, update_options) \
 117              == nominatim.tools.replication.UpdateState.NO_CHANGES
 118
 119     assert not sleeptime
 120
 121
 122 def test_update_no_data_sleep(monkeypatch, status_table, temp_db_conn, update_options):
 123     date = dt.datetime.now(dt.timezone.utc) - dt.timedelta(minutes=30)
 124     status.set_status(temp_db_conn, date, seq=34)
 125
 126     monkeypatch.setattr(nominatim.tools.replication.ReplicationServer,
 127                         "apply_diffs",
 128                         lambda *args, **kwargs: None)
 129
 130     sleeptime = []
 131     monkeypatch.setattr(time, 'sleep', lambda s: sleeptime.append(s))
 132
 133     assert nominatim.tools.replication.update(temp_db_conn, update_options) \
 134              == nominatim.tools.replication.UpdateState.NO_CHANGES
 135
 136     assert len(sleeptime) == 1
 137     assert sleeptime[0] < 3600
 138     assert sleeptime[0] > 0