test/python/test_tools_postcodes.py

   1 """
   2 Tests for functions to maintain the artificial postcode table.
   3 """
   4 import subprocess
   5
   6 import pytest
   7
   8 from nominatim.tools import postcodes
   9 import dummy_tokenizer
  10
  11 class MockPostcodeTable:
  12     """ A location_postcode table for testing.
  13     """
  14     def __init__(self, conn):
  15         self.conn = conn
  16         with conn.cursor() as cur:
  17             cur.execute("""CREATE TABLE location_postcode (
  18                                place_id BIGINT,
  19                                parent_place_id BIGINT,
  20                                rank_search SMALLINT,
  21                                rank_address SMALLINT,
  22                                indexed_status SMALLINT,
  23                                indexed_date TIMESTAMP,
  24                                country_code varchar(2),
  25                                postcode TEXT,
  26                                geometry GEOMETRY(Geometry, 4326))""")
  27             cur.execute("""CREATE OR REPLACE FUNCTION token_normalized_postcode(postcode TEXT)
  28                            RETURNS TEXT AS $$ BEGIN RETURN postcode; END; $$ LANGUAGE plpgsql;
  29
  30                            CREATE OR REPLACE FUNCTION get_country_code(place geometry)
  31                            RETURNS TEXT AS $$ BEGIN
  32                            RETURN (SELECT country_code FROM placex WHERE geometry = place LIMIT 1);
  33                            END; $$ LANGUAGE plpgsql;
  34                         """)
  35         conn.commit()
  36
  37     def add(self, country, postcode, x, y):
  38         with self.conn.cursor() as cur:
  39             cur.execute("""INSERT INTO location_postcode (place_id, indexed_status,
  40                                                           country_code, postcode,
  41                                                           geometry)
  42                            VALUES (nextval('seq_place'), 1, %s, %s,
  43                                    'SRID=4326;POINT(%s %s)')""",
  44                         (country, postcode, x, y))
  45         self.conn.commit()
  46
  47
  48     @property
  49     def row_set(self):
  50         with self.conn.cursor() as cur:
  51             cur.execute("""SELECT country_code, postcode,
  52                                   ST_X(geometry), ST_Y(geometry)
  53                            FROM location_postcode""")
  54             return set((tuple(row) for row in cur))
  55
  56
  57 @pytest.fixture
  58 def tokenizer():
  59     return dummy_tokenizer.DummyTokenizer(None, None)
  60
  61 @pytest.fixture
  62 def postcode_table(temp_db_conn, placex_table, word_table):
  63     return MockPostcodeTable(temp_db_conn)
  64
  65
  66 def test_postcodes_empty(dsn, postcode_table, place_table,
  67                          tmp_path, tokenizer):
  68     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
  69
  70     assert not postcode_table.row_set
  71
  72
  73 def test_postcodes_add_new(dsn, postcode_table, placex_table, place_row,
  74                            tmp_path, tokenizer):
  75     placex_table.add(country='xx', geom='POINT(10 12)')
  76     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='9486'))
  77     postcode_table.add('yy', '9486', 99, 34)
  78
  79     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
  80
  81     assert postcode_table.row_set == {('xx', '9486', 10, 12), }
  82
  83
  84 def test_postcodes_replace_coordinates(dsn, placex_table, postcode_table,
  85                                        place_row, tmp_path, tokenizer):
  86     placex_table.add(country='xx', geom='POINT(10 12)')
  87     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
  88     postcode_table.add('xx', 'AB 4511', 99, 34)
  89
  90     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
  91
  92     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
  93
  94
  95 def test_postcodes_replace_coordinates_close(dsn, placex_table, postcode_table,
  96                                              place_row, tmp_path, tokenizer):
  97     placex_table.add(country='xx', geom='POINT(10 12)')
  98     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
  99     postcode_table.add('xx', 'AB 4511', 10, 11.99999)
 100
 101     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 102
 103     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 11.99999)}
 104
 105
 106 def test_postcodes_remove(dsn, placex_table, postcode_table,
 107                           place_row, tmp_path, tokenizer):
 108     placex_table.add(country='xx', geom='POINT(10 12)')
 109     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 110     postcode_table.add('xx', 'badname', 10, 12)
 111
 112     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 113
 114     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
 115
 116
 117 def test_postcodes_ignore_empty_country(dsn, placex_table, postcode_table,
 118                                         place_row, tmp_path, tokenizer):
 119     placex_table.add(country=None, geom='POINT(10 12)')
 120     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 121     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 122
 123     assert not postcode_table.row_set
 124
 125
 126 def test_postcodes_remove_all(dsn, postcode_table, place_table,
 127                               tmp_path, tokenizer):
 128     postcode_table.add('ch', '5613', 10, 12)
 129     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 130
 131     assert not postcode_table.row_set
 132
 133
 134 def test_postcodes_multi_country(dsn, placex_table, postcode_table,
 135                                  place_row, tmp_path, tokenizer):
 136     placex_table.add(country='de', geom='POINT(10 12)')
 137     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='54451'))
 138
 139     placex_table.add(country='cc', geom='POINT(100 56)')
 140     place_row(geom='SRID=4326;POINT(100 56)', address=dict(postcode='DD23 T'))
 141
 142     placex_table.add(country='de', geom='POINT(10.3 11.0)')
 143     place_row(geom='SRID=4326;POINT(10.3 11.0)', address=dict(postcode='54452'))
 144
 145     placex_table.add(country='cc', geom='POINT(10.3 10.0)')
 146     place_row(geom='SRID=4326;POINT(10.3 10.0)', address=dict(postcode='54452'))
 147
 148     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 149
 150     assert postcode_table.row_set == {('de', '54451', 10, 12),
 151                                       ('de', '54452', 10.3, 11.0),
 152                                       ('cc', '54452', 10.3, 10.0),
 153                                       ('cc', 'DD23 T', 100, 56)}
 154
 155
 156 @pytest.mark.parametrize("gzipped", [True, False])
 157 def test_postcodes_extern(dsn, placex_table, postcode_table, tmp_path,
 158                           place_row, tokenizer, gzipped):
 159     placex_table.add(country='xx', geom='POINT(10 12)')
 160     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 161
 162     extfile = tmp_path / 'xx_postcodes.csv'
 163     extfile.write_text("postcode,lat,lon\nAB 4511,-4,-1\nCD 4511,-5, -10")
 164
 165     if gzipped:
 166         subprocess.run(['gzip', str(extfile)])
 167         assert not extfile.is_file()
 168
 169     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 170
 171     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12),
 172                                       ('xx', 'CD 4511', -10, -5)}
 173
 174
 175 def test_postcodes_extern_bad_column(dsn, placex_table, postcode_table,
 176                                      place_row, tmp_path, tokenizer):
 177     placex_table.add(country='xx', geom='POINT(10 12)')
 178     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 179
 180     extfile = tmp_path / 'xx_postcodes.csv'
 181     extfile.write_text("postode,lat,lon\nAB 4511,-4,-1\nCD 4511,-5, -10")
 182
 183     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 184
 185     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12)}
 186
 187
 188 def test_postcodes_extern_bad_number(dsn, placex_table, postcode_table,
 189                                      place_row, tmp_path, tokenizer):
 190     placex_table.add(country='xx', geom='POINT(10 12)')
 191     place_row(geom='SRID=4326;POINT(10 12)', address=dict(postcode='AB 4511'))
 192
 193     extfile = tmp_path / 'xx_postcodes.csv'
 194     extfile.write_text("postcode,lat,lon\nXX 4511,-4,NaN\nCD 4511,-5, -10\n34,200,0")
 195
 196     postcodes.update_postcodes(dsn, tmp_path, tokenizer)
 197
 198     assert postcode_table.row_set == {('xx', 'AB 4511', 10, 12),
 199                                       ('xx', 'CD 4511', -10, -5)}
 200
 201 def test_can_compute(dsn, temp_db_cursor):
 202     assert not postcodes.can_compute(dsn)
 203     temp_db_cursor.execute('CREATE TABLE place()')
 204     assert postcodes.can_compute(dsn)