]> git.openstreetmap.org Git - nominatim.git/blob - test/python/test_indexing.py
make word count computation part of the import
[nominatim.git] / test / python / test_indexing.py
1 """
2 Tests for running the indexing.
3 """
4 import itertools
5 import pytest
6
7 from nominatim.indexer import indexer
8 from nominatim.tokenizer import factory
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         self.postcode_id = itertools.count(700000)
16
17         self.conn = conn
18         self.conn.set_isolation_level(0)
19         with self.conn.cursor() as cur:
20             cur.execute('CREATE EXTENSION hstore')
21             cur.execute("""CREATE TABLE placex (place_id BIGINT,
22                                                 name HSTORE,
23                                                 class TEXT,
24                                                 type TEXT,
25                                                 linked_place_id BIGINT,
26                                                 rank_address SMALLINT,
27                                                 rank_search SMALLINT,
28                                                 indexed_status SMALLINT,
29                                                 indexed_date TIMESTAMP,
30                                                 partition SMALLINT,
31                                                 admin_level SMALLINT,
32                                                 country_code TEXT,
33                                                 address HSTORE,
34                                                 token_info JSONB,
35                                                 geometry_sector INTEGER)""")
36             cur.execute("""CREATE TABLE location_property_osmline (
37                                place_id BIGINT,
38                                osm_id BIGINT,
39                                address HSTORE,
40                                token_info JSONB,
41                                indexed_status SMALLINT,
42                                indexed_date TIMESTAMP,
43                                geometry_sector INTEGER)""")
44             cur.execute("""CREATE TABLE location_postcode (
45                                place_id BIGINT,
46                                indexed_status SMALLINT,
47                                indexed_date TIMESTAMP,
48                                country_code varchar(2),
49                                postcode TEXT)""")
50             cur.execute("""CREATE OR REPLACE FUNCTION date_update() RETURNS TRIGGER
51                            AS $$
52                            BEGIN
53                              IF NEW.indexed_status = 0 and OLD.indexed_status != 0 THEN
54                                NEW.indexed_date = now();
55                              END IF;
56                              RETURN NEW;
57                            END; $$ LANGUAGE plpgsql;""")
58             cur.execute("DROP TYPE IF EXISTS prepare_update_info CASCADE")
59             cur.execute("""CREATE TYPE prepare_update_info AS (
60                              name HSTORE,
61                              address HSTORE,
62                              rank_address SMALLINT,
63                              country_code TEXT,
64                              class TEXT,
65                              type TEXT,
66                              linked_place_id BIGINT
67                            )""")
68             cur.execute("""CREATE OR REPLACE FUNCTION placex_indexing_prepare(p placex,
69                                                      OUT result prepare_update_info)
70                            AS $$
71                            BEGIN
72                              result.address := p.address;
73                              result.name := p.name;
74                              result.class := p.class;
75                              result.type := p.type;
76                              result.country_code := p.country_code;
77                              result.rank_address := p.rank_address;
78                            END;
79                            $$ LANGUAGE plpgsql STABLE;
80                         """)
81             cur.execute("""CREATE OR REPLACE FUNCTION
82                              get_interpolation_address(in_address HSTORE, wayid BIGINT)
83                            RETURNS HSTORE AS $$
84                            BEGIN
85                              RETURN in_address;
86                            END;
87                            $$ LANGUAGE plpgsql STABLE;
88                         """)
89
90             for table in ('placex', 'location_property_osmline', 'location_postcode'):
91                 cur.execute("""CREATE TRIGGER {0}_update BEFORE UPDATE ON {0}
92                                FOR EACH ROW EXECUTE PROCEDURE date_update()
93                             """.format(table))
94
95     def scalar(self, query):
96         with self.conn.cursor() as cur:
97             cur.execute(query)
98             return cur.fetchone()[0]
99
100     def add_place(self, cls='place', typ='locality',
101                   rank_search=30, rank_address=30, sector=20):
102         next_id = next(self.placex_id)
103         with self.conn.cursor() as cur:
104             cur.execute("""INSERT INTO placex
105                               (place_id, class, type, rank_search, rank_address,
106                                indexed_status, geometry_sector)
107                               VALUES (%s, %s, %s, %s, %s, 1, %s)""",
108                         (next_id, cls, typ, rank_search, rank_address, sector))
109         return next_id
110
111     def add_admin(self, **kwargs):
112         kwargs['cls'] = 'boundary'
113         kwargs['typ'] = 'administrative'
114         return self.add_place(**kwargs)
115
116     def add_osmline(self, sector=20):
117         next_id = next(self.osmline_id)
118         with self.conn.cursor() as cur:
119             cur.execute("""INSERT INTO location_property_osmline
120                               (place_id, osm_id, indexed_status, geometry_sector)
121                               VALUES (%s, %s, 1, %s)""",
122                         (next_id, next_id, sector))
123         return next_id
124
125     def add_postcode(self, country, postcode):
126         next_id = next(self.postcode_id)
127         with self.conn.cursor() as cur:
128             cur.execute("""INSERT INTO location_postcode
129                             (place_id, indexed_status, country_code, postcode)
130                             VALUES (%s, 1, %s, %s)""",
131                         (next_id, country, postcode))
132         return next_id
133
134     def placex_unindexed(self):
135         return self.scalar('SELECT count(*) from placex where indexed_status > 0')
136
137     def osmline_unindexed(self):
138         return self.scalar("""SELECT count(*) from location_property_osmline
139                               WHERE indexed_status > 0""")
140
141
142 @pytest.fixture
143 def test_db(temp_db_conn):
144     yield IndexerTestDB(temp_db_conn)
145
146
147 @pytest.fixture
148 def test_tokenizer(tokenizer_mock, def_config, tmp_path):
149     def_config.project_dir = tmp_path
150     return factory.create_tokenizer(def_config)
151
152
153 @pytest.mark.parametrize("threads", [1, 15])
154 def test_index_all_by_rank(test_db, threads, test_tokenizer):
155     for rank in range(31):
156         test_db.add_place(rank_address=rank, rank_search=rank)
157     test_db.add_osmline()
158
159     assert test_db.placex_unindexed() == 31
160     assert test_db.osmline_unindexed() == 1
161
162     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads)
163     idx.index_by_rank(0, 30)
164
165     assert test_db.placex_unindexed() == 0
166     assert test_db.osmline_unindexed() == 0
167
168     assert test_db.scalar("""SELECT count(*) from placex
169                              WHERE indexed_status = 0 and indexed_date is null""") == 0
170     # ranks come in order of rank address
171     assert test_db.scalar("""
172         SELECT count(*) FROM placex p WHERE rank_address > 0
173           AND indexed_date >= (SELECT min(indexed_date) FROM placex o
174                                WHERE p.rank_address < o.rank_address)""") == 0
175     # placex rank < 30 objects come before interpolations
176     assert test_db.scalar(
177         """SELECT count(*) FROM placex WHERE rank_address < 30
178              AND indexed_date >
179                    (SELECT min(indexed_date) FROM location_property_osmline)""") == 0
180     # placex rank = 30 objects come after interpolations
181     assert test_db.scalar(
182         """SELECT count(*) FROM placex WHERE rank_address = 30
183              AND indexed_date <
184                    (SELECT max(indexed_date) FROM location_property_osmline)""") == 0
185     # rank 0 comes after rank 29 and before rank 30
186     assert test_db.scalar(
187         """SELECT count(*) FROM placex WHERE rank_address < 30
188              AND indexed_date >
189                    (SELECT min(indexed_date) FROM placex WHERE rank_address = 0)""") == 0
190     assert test_db.scalar(
191         """SELECT count(*) FROM placex WHERE rank_address = 30
192              AND indexed_date <
193                    (SELECT max(indexed_date) FROM placex WHERE rank_address = 0)""") == 0
194
195
196 @pytest.mark.parametrize("threads", [1, 15])
197 def test_index_partial_without_30(test_db, threads, test_tokenizer):
198     for rank in range(31):
199         test_db.add_place(rank_address=rank, rank_search=rank)
200     test_db.add_osmline()
201
202     assert test_db.placex_unindexed() == 31
203     assert test_db.osmline_unindexed() == 1
204
205     idx = indexer.Indexer('dbname=test_nominatim_python_unittest',
206                           test_tokenizer, threads)
207     idx.index_by_rank(4, 15)
208
209     assert test_db.placex_unindexed() == 19
210     assert test_db.osmline_unindexed() == 1
211
212     assert test_db.scalar("""
213                     SELECT count(*) FROM placex
214                       WHERE indexed_status = 0 AND not rank_address between 4 and 15""") == 0
215
216
217 @pytest.mark.parametrize("threads", [1, 15])
218 def test_index_partial_with_30(test_db, threads, test_tokenizer):
219     for rank in range(31):
220         test_db.add_place(rank_address=rank, rank_search=rank)
221     test_db.add_osmline()
222
223     assert test_db.placex_unindexed() == 31
224     assert test_db.osmline_unindexed() == 1
225
226     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads)
227     idx.index_by_rank(28, 30)
228
229     assert test_db.placex_unindexed() == 27
230     assert test_db.osmline_unindexed() == 0
231
232     assert test_db.scalar("""
233                     SELECT count(*) FROM placex
234                       WHERE indexed_status = 0 AND rank_address between 1 and 27""") == 0
235
236 @pytest.mark.parametrize("threads", [1, 15])
237 def test_index_boundaries(test_db, threads, test_tokenizer):
238     for rank in range(4, 10):
239         test_db.add_admin(rank_address=rank, rank_search=rank)
240     for rank in range(31):
241         test_db.add_place(rank_address=rank, rank_search=rank)
242     test_db.add_osmline()
243
244     assert test_db.placex_unindexed() == 37
245     assert test_db.osmline_unindexed() == 1
246
247     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads)
248     idx.index_boundaries(0, 30)
249
250     assert test_db.placex_unindexed() == 31
251     assert test_db.osmline_unindexed() == 1
252
253     assert test_db.scalar("""
254                     SELECT count(*) FROM placex
255                       WHERE indexed_status = 0 AND class != 'boundary'""") == 0
256
257
258 @pytest.mark.parametrize("threads", [1, 15])
259 def test_index_postcodes(test_db, threads, test_tokenizer):
260     for postcode in range(1000):
261         test_db.add_postcode('de', postcode)
262     for postcode in range(32000, 33000):
263         test_db.add_postcode('us', postcode)
264
265     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads)
266     idx.index_postcodes()
267
268     assert test_db.scalar("""SELECT count(*) FROM location_postcode
269                                   WHERE indexed_status != 0""") == 0
270
271
272 @pytest.mark.parametrize("analyse", [True, False])
273 def test_index_full(test_db, analyse, test_tokenizer):
274     for rank in range(4, 10):
275         test_db.add_admin(rank_address=rank, rank_search=rank)
276     for rank in range(31):
277         test_db.add_place(rank_address=rank, rank_search=rank)
278     test_db.add_osmline()
279     for postcode in range(1000):
280         test_db.add_postcode('de', postcode)
281
282     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, 4)
283     idx.index_full(analyse=analyse)
284
285     assert test_db.placex_unindexed() == 0
286     assert test_db.osmline_unindexed() == 0
287     assert test_db.scalar("""SELECT count(*) FROM location_postcode
288                              WHERE indexed_status != 0""") == 0
289
290
291 @pytest.mark.parametrize("threads", [1, 15])
292 def test_index_reopen_connection(test_db, threads, monkeypatch, test_tokenizer):
293     monkeypatch.setattr(indexer.WorkerPool, "REOPEN_CONNECTIONS_AFTER", 15)
294
295     for _ in range(1000):
296         test_db.add_place(rank_address=30, rank_search=30)
297
298     idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads)
299     idx.index_by_rank(28, 30)
300
301     assert test_db.placex_unindexed() == 0