test/python/tools/test_tiger_data.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 Test for tiger data function
   9 """
  10 import tarfile
  11 from textwrap import dedent
  12
  13 import pytest
  14 import pytest_asyncio  # noqa: F401
  15
  16 from nominatim_db.db.connection import execute_scalar
  17 from nominatim_db.tools import tiger_data, freeze
  18 from nominatim_db.errors import UsageError
  19
  20
  21 class MockTigerTable:
  22
  23     def __init__(self, conn):
  24         self.conn = conn
  25         with conn.cursor() as cur:
  26             cur.execute("""CREATE TABLE tiger (linegeo GEOMETRY,
  27                                                start INTEGER,
  28                                                stop INTEGER,
  29                                                interpol TEXT,
  30                                                token_info JSONB,
  31                                                postcode TEXT)""")
  32
  33             # We need this table to determine if the database is frozen or not
  34             cur.execute("CREATE TABLE place (number INTEGER)")
  35
  36     def count(self):
  37         return execute_scalar(self.conn, "SELECT count(*) FROM tiger")
  38
  39     def row(self):
  40         with self.conn.cursor() as cur:
  41             cur.execute("SELECT * FROM tiger LIMIT 1")
  42             return cur.fetchone()
  43
  44
  45 @pytest.fixture
  46 def tiger_table(def_config, temp_db_conn, sql_preprocessor,
  47                 temp_db_with_extensions, tmp_path):
  48     def_config.lib_dir.sql = tmp_path / 'sql'
  49     def_config.lib_dir.sql.mkdir()
  50
  51     (def_config.lib_dir.sql / 'tiger_import_start.sql').write_text(
  52         """CREATE OR REPLACE FUNCTION tiger_line_import(linegeo GEOMETRY, start INTEGER,
  53                                                         stop INTEGER, interpol TEXT,
  54                                                         token_info JSONB, postcode TEXT)
  55            RETURNS INTEGER AS $$
  56             INSERT INTO tiger VALUES(linegeo, start, stop, interpol, token_info, postcode)
  57             RETURNING 1
  58            $$ LANGUAGE SQL;""")
  59     (def_config.lib_dir.sql / 'tiger_import_finish.sql').write_text(
  60         """DROP FUNCTION tiger_line_import (linegeo GEOMETRY, in_startnumber INTEGER,
  61                                  in_endnumber INTEGER, interpolationtype TEXT,
  62                                  token_info JSONB, in_postcode TEXT);""")
  63
  64     return MockTigerTable(temp_db_conn)
  65
  66
  67 @pytest.fixture
  68 def csv_factory(tmp_path):
  69     def _mk_file(fname, hnr_from=1, hnr_to=9, interpol='odd', street='Main St',
  70                  city='Newtown', state='AL', postcode='12345',
  71                  geometry='LINESTRING(-86.466995 32.428956,-86.466923 32.428933)'):
  72         (tmp_path / (fname + '.csv')).write_text(dedent("""\
  73         from;to;interpolation;street;city;state;postcode;geometry
  74         {};{};{};{};{};{};{};{}
  75         """.format(hnr_from, hnr_to, interpol, street, city, state,
  76                    postcode, geometry)))
  77
  78     return _mk_file
  79
  80
  81 @pytest.mark.parametrize("threads", (1, 5))
  82 @pytest.mark.asyncio
  83 async def test_add_tiger_data(def_config, src_dir, tiger_table, tokenizer_mock, threads):
  84     await tiger_data.add_tiger_data(str(src_dir / 'test' / 'testdb' / 'tiger'),
  85                                     def_config, threads, tokenizer_mock())
  86
  87     assert tiger_table.count() == 6213
  88
  89
  90 @pytest.mark.asyncio
  91 async def test_add_tiger_data_database_frozen(def_config, temp_db_conn, tiger_table, tokenizer_mock,
  92                                               tmp_path):
  93     freeze.drop_update_tables(temp_db_conn)
  94
  95     with pytest.raises(UsageError) as excinfo:
  96         await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
  97
  98         assert "database frozen" in str(excinfo.value)
  99
 100     assert tiger_table.count() == 0
 101
 102
 103 @pytest.mark.asyncio
 104 async def test_add_tiger_data_no_files(def_config, tiger_table, tokenizer_mock,
 105                                        tmp_path):
 106     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 107
 108     assert tiger_table.count() == 0
 109
 110
 111 @pytest.mark.asyncio
 112 async def test_add_tiger_data_bad_file(def_config, tiger_table, tokenizer_mock,
 113                                        tmp_path):
 114     sqlfile = tmp_path / '1010.csv'
 115     sqlfile.write_text("""Random text""")
 116
 117     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 118
 119     assert tiger_table.count() == 0
 120
 121
 122 @pytest.mark.asyncio
 123 async def test_add_tiger_data_hnr_nan(def_config, tiger_table, tokenizer_mock,
 124                                       csv_factory, tmp_path):
 125     csv_factory('file1', hnr_from=99)
 126     csv_factory('file2', hnr_from='L12')
 127     csv_factory('file3', hnr_to='12.4')
 128
 129     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 130
 131     assert tiger_table.count() == 1
 132     assert tiger_table.row().start == 99
 133
 134
 135 @pytest.mark.parametrize("threads", (1, 5))
 136 @pytest.mark.asyncio
 137 async def test_add_tiger_data_tarfile(def_config, tiger_table, tokenizer_mock,
 138                                       tmp_path, src_dir, threads):
 139     tar = tarfile.open(str(tmp_path / 'sample.tar.gz'), "w:gz")
 140     tar.add(str(src_dir / 'test' / 'testdb' / 'tiger' / '01001.csv'))
 141     tar.close()
 142
 143     await tiger_data.add_tiger_data(str(tmp_path / 'sample.tar.gz'), def_config, threads,
 144                                     tokenizer_mock())
 145
 146     assert tiger_table.count() == 6213
 147
 148
 149 @pytest.mark.asyncio
 150 async def test_add_tiger_data_bad_tarfile(def_config, tiger_table, tokenizer_mock,
 151                                           tmp_path):
 152     tarfile = tmp_path / 'sample.tar.gz'
 153     tarfile.write_text("""Random text""")
 154
 155     with pytest.raises(UsageError):
 156         await tiger_data.add_tiger_data(str(tarfile), def_config, 1, tokenizer_mock())
 157
 158
 159 @pytest.mark.asyncio
 160 async def test_add_tiger_data_empty_tarfile(def_config, tiger_table, tokenizer_mock,
 161                                             tmp_path):
 162     tar = tarfile.open(str(tmp_path / 'sample.tar.gz'), "w:gz")
 163     tar.add(__file__)
 164     tar.close()
 165
 166     await tiger_data.add_tiger_data(str(tmp_path / 'sample.tar.gz'), def_config, 1,
 167                                     tokenizer_mock())
 168
 169     assert tiger_table.count() == 0