X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/9397bf54b89c1ae0b41115673d2f38582cd03ed9..5d32ac9ac918929c11602a740c3033bcfdb0f929:/test/python/test_indexing.py diff --git a/test/python/test_indexing.py b/test/python/test_indexing.py index 223a599e..4c9d940d 100644 --- a/test/python/test_indexing.py +++ b/test/python/test_indexing.py @@ -2,10 +2,10 @@ Tests for running the indexing. """ import itertools -import psycopg2 import pytest from nominatim.indexer import indexer +from nominatim.tokenizer import factory class IndexerTestDB: @@ -19,18 +19,25 @@ class IndexerTestDB: with self.conn.cursor() as cur: cur.execute('CREATE EXTENSION hstore') cur.execute("""CREATE TABLE placex (place_id BIGINT, + name HSTORE, class TEXT, type TEXT, + linked_place_id BIGINT, rank_address SMALLINT, rank_search SMALLINT, indexed_status SMALLINT, indexed_date TIMESTAMP, partition SMALLINT, admin_level SMALLINT, + country_code TEXT, address HSTORE, + token_info JSONB, geometry_sector INTEGER)""") cur.execute("""CREATE TABLE location_property_osmline ( place_id BIGINT, + osm_id BIGINT, + address HSTORE, + token_info JSONB, indexed_status SMALLINT, indexed_date TIMESTAMP, geometry_sector INTEGER)""") @@ -48,17 +55,38 @@ class IndexerTestDB: END IF; RETURN NEW; END; $$ LANGUAGE plpgsql;""") - cur.execute("""CREATE OR REPLACE FUNCTION placex_prepare_update(p placex, - OUT name HSTORE, - OUT address HSTORE, - OUT country_feature VARCHAR) + cur.execute("DROP TYPE IF EXISTS prepare_update_info CASCADE") + cur.execute("""CREATE TYPE prepare_update_info AS ( + name HSTORE, + address HSTORE, + rank_address SMALLINT, + country_code TEXT, + class TEXT, + type TEXT, + linked_place_id BIGINT + )""") + cur.execute("""CREATE OR REPLACE FUNCTION placex_indexing_prepare(p placex, + OUT result prepare_update_info) AS $$ BEGIN - address := p.address; - name := p.address; + result.address := p.address; + result.name := p.name; + result.class := p.class; + result.type := p.type; + result.country_code := p.country_code; + result.rank_address := p.rank_address; END; $$ LANGUAGE plpgsql STABLE; """) + cur.execute("""CREATE OR REPLACE FUNCTION + get_interpolation_address(in_address HSTORE, wayid BIGINT) + RETURNS HSTORE AS $$ + BEGIN + RETURN in_address; + END; + $$ LANGUAGE plpgsql STABLE; + """) + for table in ('placex', 'location_property_osmline', 'location_postcode'): cur.execute("""CREATE TRIGGER {0}_update BEFORE UPDATE ON {0} FOR EACH ROW EXECUTE PROCEDURE date_update() @@ -89,9 +117,9 @@ class IndexerTestDB: next_id = next(self.osmline_id) with self.conn.cursor() as cur: cur.execute("""INSERT INTO location_property_osmline - (place_id, indexed_status, geometry_sector) - VALUES (%s, 1, %s)""", - (next_id, sector)) + (place_id, osm_id, indexed_status, geometry_sector) + VALUES (%s, %s, 1, %s)""", + (next_id, next_id, sector)) return next_id def add_postcode(self, country, postcode): @@ -107,7 +135,8 @@ class IndexerTestDB: return self.scalar('SELECT count(*) from placex where indexed_status > 0') def osmline_unindexed(self): - return self.scalar('SELECT count(*) from location_property_osmline where indexed_status > 0') + return self.scalar("""SELECT count(*) from location_property_osmline + WHERE indexed_status > 0""") @pytest.fixture @@ -115,122 +144,133 @@ def test_db(temp_db_conn): yield IndexerTestDB(temp_db_conn) +@pytest.fixture +def test_tokenizer(tokenizer_mock, def_config, tmp_path): + def_config.project_dir = tmp_path + return factory.create_tokenizer(def_config) + + @pytest.mark.parametrize("threads", [1, 15]) -def test_index_all_by_rank(test_db, threads): +def test_index_all_by_rank(test_db, threads, test_tokenizer): for rank in range(31): test_db.add_place(rank_address=rank, rank_search=rank) test_db.add_osmline() - assert 31 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 31 + assert test_db.osmline_unindexed() == 1 - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads) idx.index_by_rank(0, 30) - assert 0 == test_db.placex_unindexed() - assert 0 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 0 + assert test_db.osmline_unindexed() == 0 - assert 0 == test_db.scalar("""SELECT count(*) from placex - WHERE indexed_status = 0 and indexed_date is null""") + assert test_db.scalar("""SELECT count(*) from placex + WHERE indexed_status = 0 and indexed_date is null""") == 0 # ranks come in order of rank address - assert 0 == test_db.scalar(""" + assert test_db.scalar(""" SELECT count(*) FROM placex p WHERE rank_address > 0 AND indexed_date >= (SELECT min(indexed_date) FROM placex o - WHERE p.rank_address < o.rank_address)""") + WHERE p.rank_address < o.rank_address)""") == 0 # placex rank < 30 objects come before interpolations - assert 0 == test_db.scalar( + assert test_db.scalar( """SELECT count(*) FROM placex WHERE rank_address < 30 - AND indexed_date > (SELECT min(indexed_date) FROM location_property_osmline)""") + AND indexed_date > + (SELECT min(indexed_date) FROM location_property_osmline)""") == 0 # placex rank = 30 objects come after interpolations - assert 0 == test_db.scalar( + assert test_db.scalar( """SELECT count(*) FROM placex WHERE rank_address = 30 - AND indexed_date < (SELECT max(indexed_date) FROM location_property_osmline)""") + AND indexed_date < + (SELECT max(indexed_date) FROM location_property_osmline)""") == 0 # rank 0 comes after rank 29 and before rank 30 - assert 0 == test_db.scalar( + assert test_db.scalar( """SELECT count(*) FROM placex WHERE rank_address < 30 - AND indexed_date > (SELECT min(indexed_date) FROM placex WHERE rank_address = 0)""") - assert 0 == test_db.scalar( + AND indexed_date > + (SELECT min(indexed_date) FROM placex WHERE rank_address = 0)""") == 0 + assert test_db.scalar( """SELECT count(*) FROM placex WHERE rank_address = 30 - AND indexed_date < (SELECT max(indexed_date) FROM placex WHERE rank_address = 0)""") + AND indexed_date < + (SELECT max(indexed_date) FROM placex WHERE rank_address = 0)""") == 0 @pytest.mark.parametrize("threads", [1, 15]) -def test_index_partial_without_30(test_db, threads): +def test_index_partial_without_30(test_db, threads, test_tokenizer): for rank in range(31): test_db.add_place(rank_address=rank, rank_search=rank) test_db.add_osmline() - assert 31 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 31 + assert test_db.osmline_unindexed() == 1 - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', + test_tokenizer, threads) idx.index_by_rank(4, 15) - assert 19 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 19 + assert test_db.osmline_unindexed() == 1 - assert 0 == test_db.scalar(""" + assert test_db.scalar(""" SELECT count(*) FROM placex - WHERE indexed_status = 0 AND not rank_address between 4 and 15""") + WHERE indexed_status = 0 AND not rank_address between 4 and 15""") == 0 @pytest.mark.parametrize("threads", [1, 15]) -def test_index_partial_with_30(test_db, threads): +def test_index_partial_with_30(test_db, threads, test_tokenizer): for rank in range(31): test_db.add_place(rank_address=rank, rank_search=rank) test_db.add_osmline() - assert 31 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 31 + assert test_db.osmline_unindexed() == 1 - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads) idx.index_by_rank(28, 30) - assert 27 == test_db.placex_unindexed() - assert 0 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 27 + assert test_db.osmline_unindexed() == 0 - assert 0 == test_db.scalar(""" + assert test_db.scalar(""" SELECT count(*) FROM placex - WHERE indexed_status = 0 AND rank_address between 1 and 27""") + WHERE indexed_status = 0 AND rank_address between 1 and 27""") == 0 @pytest.mark.parametrize("threads", [1, 15]) -def test_index_boundaries(test_db, threads): +def test_index_boundaries(test_db, threads, test_tokenizer): for rank in range(4, 10): test_db.add_admin(rank_address=rank, rank_search=rank) for rank in range(31): test_db.add_place(rank_address=rank, rank_search=rank) test_db.add_osmline() - assert 37 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 37 + assert test_db.osmline_unindexed() == 1 - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads) idx.index_boundaries(0, 30) - assert 31 == test_db.placex_unindexed() - assert 1 == test_db.osmline_unindexed() + assert test_db.placex_unindexed() == 31 + assert test_db.osmline_unindexed() == 1 - assert 0 == test_db.scalar(""" + assert test_db.scalar(""" SELECT count(*) FROM placex - WHERE indexed_status = 0 AND class != 'boundary'""") + WHERE indexed_status = 0 AND class != 'boundary'""") == 0 @pytest.mark.parametrize("threads", [1, 15]) -def test_index_postcodes(test_db, threads): +def test_index_postcodes(test_db, threads, test_tokenizer): for postcode in range(1000): test_db.add_postcode('de', postcode) for postcode in range(32000, 33000): test_db.add_postcode('us', postcode) - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads) idx.index_postcodes() - assert 0 == test_db.scalar("""SELECT count(*) FROM location_postcode - WHERE indexed_status != 0""") + assert test_db.scalar("""SELECT count(*) FROM location_postcode + WHERE indexed_status != 0""") == 0 @pytest.mark.parametrize("analyse", [True, False]) -def test_index_full(test_db, analyse): +def test_index_full(test_db, analyse, test_tokenizer): for rank in range(4, 10): test_db.add_admin(rank_address=rank, rank_search=rank) for rank in range(31): @@ -239,23 +279,23 @@ def test_index_full(test_db, analyse): for postcode in range(1000): test_db.add_postcode('de', postcode) - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', 4) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, 4) idx.index_full(analyse=analyse) - assert 0 == test_db.placex_unindexed() - assert 0 == test_db.osmline_unindexed() - assert 0 == test_db.scalar("""SELECT count(*) FROM location_postcode - WHERE indexed_status != 0""") + assert test_db.placex_unindexed() == 0 + assert test_db.osmline_unindexed() == 0 + assert test_db.scalar("""SELECT count(*) FROM location_postcode + WHERE indexed_status != 0""") == 0 @pytest.mark.parametrize("threads", [1, 15]) -def test_index_reopen_connection(test_db, threads, monkeypatch): +def test_index_reopen_connection(test_db, threads, monkeypatch, test_tokenizer): monkeypatch.setattr(indexer.WorkerPool, "REOPEN_CONNECTIONS_AFTER", 15) for _ in range(1000): test_db.add_place(rank_address=30, rank_search=30) - idx = indexer.Indexer('dbname=test_nominatim_python_unittest', threads) + idx = indexer.Indexer('dbname=test_nominatim_python_unittest', test_tokenizer, threads) idx.index_by_rank(28, 30) - assert 0 == test_db.placex_unindexed() + assert test_db.placex_unindexed() == 0