test/python/tools/test_postcodes.py

   1 # SPDX-License-Identifier: GPL-3.0-or-later
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2025 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 Tests for functions to maintain the artificial postcode table.
   9 """
  10 import subprocess
  11
  12 import pytest
  13
  14 from nominatim_db.tools import postcodes
  15 from nominatim_db.data import country_info
  16 import dummy_tokenizer
  17
  18
  19 class MockPostcodeTable:
  20     """ A location_postcode table for testing.
  21     """
  22     def __init__(self, conn):
  23         self.conn = conn
  24         with conn.cursor() as cur:
  25             cur.execute("""CREATE TABLE location_postcode (
  26                                place_id BIGINT,
  27                                parent_place_id BIGINT,
  28                                rank_search SMALLINT,
  29                                rank_address SMALLINT,
  30                                indexed_status SMALLINT,
  31                                indexed_date TIMESTAMP,
  32                                country_code varchar(2),
  33                                postcode TEXT,
  34                                geometry GEOMETRY(Geometry, 4326))""")
  35             cur.execute("""CREATE OR REPLACE FUNCTION token_normalized_postcode(postcode TEXT)
  36                            RETURNS TEXT AS $$ BEGIN RETURN postcode; END; $$ LANGUAGE plpgsql;
  37
  38                            CREATE OR REPLACE FUNCTION get_country_code(place geometry)
  39                            RETURNS TEXT AS $$ BEGIN
  40                            RETURN null;
  41                            END; $$ LANGUAGE plpgsql;
  42                         """)
  43         conn.commit()
  44
  45     def add(self, country, postcode, x, y):
  46         with self.conn.cursor() as cur:
  47             cur.execute("""INSERT INTO location_postcode (place_id, indexed_status,
  48                                                           country_code, postcode,
  49                                                           geometry)
  50                            VALUES (nextval('seq_place'), 1, %s, %s,
  51                                    ST_SetSRID(ST_MakePoint(%s, %s), 4326))""",
  52                         (country, postcode, x, y))
  53         self.conn.commit()
  54
  55     @property
  56     def row_set(self):
  57         with self.conn.cursor() as cur:
  58             cur.execute("""SELECT country_code, postcode,
  59                                   ST_X(geometry), ST_Y(geometry)
  60                            FROM location_postcode""")
  61             return set((tuple(row) for row in cur))
  62
  63
  64 @pytest.fixture
  65 def tokenizer():
  66     return dummy_tokenizer.DummyTokenizer(None)
  67
  68
  69 @pytest.fixture
  70 def postcode_table(def_config, temp_db_conn, placex_table):
  71     country_info.setup_country_config(def_config)
  72     return MockPostcodeTable(temp_db_conn)
  73
  74
  75 @pytest.fixture
  76 def insert_implicit_postcode(placex_table, place_row):
  77     """
  78         Inserts data into the placex and place table
  79         which can then be used to compute one postcode.
  80     """
  81     def _insert_implicit_postcode(osm_id, country, geometry, address):
  82         placex_table.add(osm_id=osm_id, country=country, geom=geometry)
  83         place_row(osm_id=osm_id, geom='SRID=4326;'+geometry, address=address)
  84
  85     return _insert_implicit_postcode
  86
  87
  88 def test_postcodes_empty(dsn, postcode_table, place_table, tokenizer):
  89     postcodes.update_postcodes(dsn, None, tokenizer)
  90
  91     assert not postcode_table.row_set
  92
  93
  94 def test_postcodes_add_new(dsn, postcode_table, insert_implicit_postcode, tokenizer):
  95     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='9486'))
  96     postcode_table.add('yy', '9486', 99, 34)
  97
  98     postcodes.update_postcodes(dsn, None, tokenizer)
  99
 100     assert postcode_table.row_set == {('xx', '9486', 10, 12), }
 101
 102
 103 def test_postcodes_replace_coordinates(dsn, postcode_table, tmp_path,
 104                                        insert_implicit_postcode, tokenizer):
 105     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 106     postcode_table.add('xx', 'AB 4511', 99, 34)
 107
 108     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 109
 110     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
 111
 112
 113 def test_postcodes_replace_coordinates_close(dsn, postcode_table,
 114                                              insert_implicit_postcode, tokenizer):
 115     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 116     postcode_table.add('xx', 'AB 4511', 10, 11.99999)
 117
 118     postcodes.update_postcodes(dsn, None, tokenizer)
 119
 120     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 11.99999)}
 121
 122
 123 def test_postcodes_remove(dsn, postcode_table,
 124                           insert_implicit_postcode, tokenizer):
 125     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 126     postcode_table.add('xx', 'badname', 10, 12)
 127
 128     postcodes.update_postcodes(dsn, None, tokenizer)
 129
 130     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
 131
 132
 133 def test_postcodes_ignore_empty_country(dsn, postcode_table,
 134                                         insert_implicit_postcode, tokenizer):
 135     insert_implicit_postcode(1, None, 'POINT(10 12)', dict(postcode='AB 4511'))
 136     postcodes.update_postcodes(dsn, None, tokenizer)
 137     assert not postcode_table.row_set
 138
 139
 140 def test_postcodes_remove_all(dsn, postcode_table, place_table, tokenizer):
 141     postcode_table.add('ch', '5613', 10, 12)
 142     postcodes.update_postcodes(dsn, None, tokenizer)
 143
 144     assert not postcode_table.row_set
 145
 146
 147 def test_postcodes_multi_country(dsn, postcode_table,
 148                                  insert_implicit_postcode, tokenizer):
 149     insert_implicit_postcode(1, 'de', 'POINT(10 12)', dict(postcode='54451'))
 150     insert_implicit_postcode(2, 'cc', 'POINT(100 56)', dict(postcode='DD23 T'))
 151     insert_implicit_postcode(3, 'de', 'POINT(10.3 11.0)', dict(postcode='54452'))
 152     insert_implicit_postcode(4, 'cc', 'POINT(10.3 11.0)', dict(postcode='54452'))
 153
 154     postcodes.update_postcodes(dsn, None, tokenizer)
 155
 156     assert postcode_table.row_set == {('de', '54451', 10, 12),
 157                                       ('de', '54452', 10.3, 11.0),
 158                                       ('cc', '54452', 10.3, 11.0),
 159                                       ('cc', 'DD23 T', 100, 56)}
 160
 161
 162 @pytest.mark.parametrize("gzipped", [True, False])
 163 def test_postcodes_extern(dsn, postcode_table, tmp_path,
 164                           insert_implicit_postcode, tokenizer, gzipped):
 165     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 166
 167     extfile = tmp_path / 'xx_postcodes.csv'
 168     extfile.write_text("postcode,lat,lon\nAB 4511,-4,-1\nCD 4511,-5, -10")
 169
 170     if gzipped:
 171         subprocess.run(['gzip', str(extfile)])
 172         assert not extfile.is_file()
 173
 174     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 175
 176     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12),
 177                                       ('xx', 'CD 4511', -10, -5)}
 178
 179
 180 def test_postcodes_extern_bad_column(dsn, postcode_table, tmp_path,
 181                                      insert_implicit_postcode, tokenizer):
 182     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 183
 184     extfile = tmp_path / 'xx_postcodes.csv'
 185     extfile.write_text("postode,lat,lon\nAB 4511,-4,-1\nCD 4511,-5, -10")
 186
 187     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 188
 189     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
 190
 191
 192 def test_postcodes_extern_bad_number(dsn, insert_implicit_postcode,
 193                                      postcode_table, tmp_path, tokenizer):
 194     insert_implicit_postcode(1, 'xx', 'POINT(10 12)', dict(postcode='AB 4511'))
 195
 196     extfile = tmp_path / 'xx_postcodes.csv'
 197     extfile.write_text("postcode,lat,lon\nXX 4511,-4,NaN\nCD 4511,-5, -10\n34,200,0")
 198
 199     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 200
 201     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12),
 202                                       ('xx', 'CD 4511', -10, -5)}
 203
 204
 205 def test_can_compute(dsn, table_factory):
 206     assert not postcodes.can_compute(dsn)
 207     table_factory('place')
 208     assert postcodes.can_compute(dsn)
 209
 210
 211 def test_no_placex_entry(dsn, temp_db_cursor, place_row, postcode_table, tokenizer):
 212     # Rewrite the get_country_code function to verify its execution.
 213     temp_db_cursor.execute("""
 214         CREATE OR REPLACE FUNCTION get_country_code(place geometry)
 215         RETURNS TEXT AS $$ BEGIN
 216         RETURN 'yy';
 217         END; $$ LANGUAGE plpgsql;
 218     """)
 219     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 220     postcodes.update_postcodes(dsn, None, tokenizer)
 221
 222     assert postcode_table.row_set == {('yy', 'AB 4511', 10, 12)}
 223
 224
 225 def test_discard_badly_formatted_postcodes(dsn, temp_db_cursor, place_row,
 226                                            postcode_table, tokenizer):
 227     # Rewrite the get_country_code function to verify its execution.
 228     temp_db_cursor.execute("""
 229         CREATE OR REPLACE FUNCTION get_country_code(place geometry)
 230         RETURNS TEXT AS $$ BEGIN
 231         RETURN 'fr';
 232         END; $$ LANGUAGE plpgsql;
 233     """)
 234     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 235     postcodes.update_postcodes(dsn, None, tokenizer)
 236
 237     assert not postcode_table.row_set