test/python/test_indexing.py

   1 """
   2 Tests for running the indexing.
   3 """
   4 import itertools
   5 import psycopg2
   6 import pytest
   7
   8 from nominatim.indexer.indexer import Indexer
   9
  10 class IndexerTestDB:
  11
  12     def __init__(self, conn):
  13         self.placex_id = itertools.count(100000)
  14         self.osmline_id = itertools.count(500000)
  15
  16         self.conn = conn
  17         self.conn.set_isolation_level(0)
  18         with self.conn.cursor() as cur:
  19             cur.execute("""CREATE TABLE placex (place_id BIGINT,
  20                                                 class TEXT,
  21                                                 type TEXT,
  22                                                 rank_address SMALLINT,
  23                                                 rank_search SMALLINT,
  24                                                 indexed_status SMALLINT,
  25                                                 indexed_date TIMESTAMP,
  26                                                 partition SMALLINT,
  27                                                 admin_level SMALLINT,
  28                                                 geometry_sector INTEGER)""")
  29             cur.execute("""CREATE TABLE location_property_osmline (
  30                                place_id BIGINT,
  31                                indexed_status SMALLINT,
  32                                indexed_date TIMESTAMP,
  33                                geometry_sector INTEGER)""")
  34             cur.execute("""CREATE OR REPLACE FUNCTION date_update() RETURNS TRIGGER
  35                            AS $$
  36                            BEGIN
  37                              IF NEW.indexed_status = 0 and OLD.indexed_status != 0 THEN
  38                                NEW.indexed_date = now();
  39                              END IF;
  40                              RETURN NEW;
  41                            END; $$ LANGUAGE plpgsql;""")
  42             cur.execute("""CREATE TRIGGER placex_update BEFORE UPDATE ON placex
  43                            FOR EACH ROW EXECUTE PROCEDURE date_update()""")
  44             cur.execute("""CREATE TRIGGER osmline_update BEFORE UPDATE ON location_property_osmline
  45                            FOR EACH ROW EXECUTE PROCEDURE date_update()""")
  46
  47     def scalar(self, query):
  48         with self.conn.cursor() as cur:
  49             cur.execute(query)
  50             return cur.fetchone()[0]
  51
  52     def add_place(self, cls='place', typ='locality',
  53                   rank_search=30, rank_address=30, sector=20):
  54         next_id = next(self.placex_id)
  55         with self.conn.cursor() as cur:
  56             cur.execute("""INSERT INTO placex
  57                               (place_id, class, type, rank_search, rank_address,
  58                                indexed_status, geometry_sector)
  59                               VALUES (%s, %s, %s, %s, %s, 1, %s)""",
  60                         (next_id, cls, typ, rank_search, rank_address, sector))
  61         return next_id
  62
  63     def add_admin(self, **kwargs):
  64         kwargs['cls'] = 'boundary'
  65         kwargs['typ'] = 'administrative'
  66         return self.add_place(**kwargs)
  67
  68     def add_osmline(self, sector=20):
  69         next_id = next(self.osmline_id)
  70         with self.conn.cursor() as cur:
  71             cur.execute("""INSERT INTO location_property_osmline
  72                               (place_id, indexed_status, geometry_sector)
  73                               VALUES (%s, 1, %s)""",
  74                         (next_id, sector))
  75         return next_id
  76
  77     def placex_unindexed(self):
  78         return self.scalar('SELECT count(*) from placex where indexed_status > 0')
  79
  80     def osmline_unindexed(self):
  81         return self.scalar('SELECT count(*) from location_property_osmline where indexed_status > 0')
  82
  83
  84 @pytest.fixture
  85 def test_db(temp_db_conn):
  86     yield IndexerTestDB(temp_db_conn)
  87
  88
  89 @pytest.mark.parametrize("threads", [1, 15])
  90 def test_index_full(test_db, threads):
  91     for rank in range(31):
  92         test_db.add_place(rank_address=rank, rank_search=rank)
  93     test_db.add_osmline()
  94
  95     assert 31 == test_db.placex_unindexed()
  96     assert 1 == test_db.osmline_unindexed()
  97
  98     idx = Indexer('dbname=test_nominatim_python_unittest', threads)
  99     idx.index_by_rank(0, 30)
 100
 101     assert 0 == test_db.placex_unindexed()
 102     assert 0 == test_db.osmline_unindexed()
 103
 104     assert 0 == test_db.scalar("""SELECT count(*) from placex
 105                                WHERE indexed_status = 0 and indexed_date is null""")
 106     # ranks come in order of rank address
 107     assert 0 == test_db.scalar("""
 108         SELECT count(*) FROM placex p WHERE rank_address > 0
 109           AND indexed_date >= (SELECT min(indexed_date) FROM placex o
 110                                WHERE p.rank_address < o.rank_address)""")
 111     # placex rank < 30 objects come before interpolations
 112     assert 0 == test_db.scalar(
 113         """SELECT count(*) FROM placex WHERE rank_address < 30
 114              AND indexed_date > (SELECT min(indexed_date) FROM location_property_osmline)""")
 115     # placex rank = 30 objects come after interpolations
 116     assert 0 == test_db.scalar(
 117         """SELECT count(*) FROM placex WHERE rank_address = 30
 118              AND indexed_date < (SELECT max(indexed_date) FROM location_property_osmline)""")
 119     # rank 0 comes after rank 29 and before rank 30
 120     assert 0 == test_db.scalar(
 121         """SELECT count(*) FROM placex WHERE rank_address < 30
 122              AND indexed_date > (SELECT min(indexed_date) FROM placex WHERE rank_address = 0)""")
 123     assert 0 == test_db.scalar(
 124         """SELECT count(*) FROM placex WHERE rank_address = 30
 125              AND indexed_date < (SELECT max(indexed_date) FROM placex WHERE rank_address = 0)""")
 126
 127
 128 @pytest.mark.parametrize("threads", [1, 15])
 129 def test_index_partial_without_30(test_db, threads):
 130     for rank in range(31):
 131         test_db.add_place(rank_address=rank, rank_search=rank)
 132     test_db.add_osmline()
 133
 134     assert 31 == test_db.placex_unindexed()
 135     assert 1 == test_db.osmline_unindexed()
 136
 137     idx = Indexer('dbname=test_nominatim_python_unittest', threads)
 138     idx.index_by_rank(4, 15)
 139
 140     assert 19 == test_db.placex_unindexed()
 141     assert 1 == test_db.osmline_unindexed()
 142
 143     assert 0 == test_db.scalar("""
 144                     SELECT count(*) FROM placex
 145                       WHERE indexed_status = 0 AND not rank_address between 4 and 15""")
 146
 147
 148 @pytest.mark.parametrize("threads", [1, 15])
 149 def test_index_partial_with_30(test_db, threads):
 150     for rank in range(31):
 151         test_db.add_place(rank_address=rank, rank_search=rank)
 152     test_db.add_osmline()
 153
 154     assert 31 == test_db.placex_unindexed()
 155     assert 1 == test_db.osmline_unindexed()
 156
 157     idx = Indexer('dbname=test_nominatim_python_unittest', threads)
 158     idx.index_by_rank(28, 30)
 159
 160     assert 27 == test_db.placex_unindexed()
 161     assert 0 == test_db.osmline_unindexed()
 162
 163     assert 0 == test_db.scalar("""
 164                     SELECT count(*) FROM placex
 165                       WHERE indexed_status = 0 AND rank_address between 1 and 27""")
 166
 167 @pytest.mark.parametrize("threads", [1, 15])
 168 def test_index_boundaries(test_db, threads):
 169     for rank in range(4, 10):
 170         test_db.add_admin(rank_address=rank, rank_search=rank)
 171     for rank in range(31):
 172         test_db.add_place(rank_address=rank, rank_search=rank)
 173     test_db.add_osmline()
 174
 175     assert 37 == test_db.placex_unindexed()
 176     assert 1 == test_db.osmline_unindexed()
 177
 178     idx = Indexer('dbname=test_nominatim_python_unittest', threads)
 179     idx.index_boundaries(0, 30)
 180
 181     assert 31 == test_db.placex_unindexed()
 182     assert 1 == test_db.osmline_unindexed()
 183
 184     assert 0 == test_db.scalar("""
 185                     SELECT count(*) FROM placex
 186                       WHERE indexed_status = 0 AND class != 'boundary'""")