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) 2024 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
  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 class MockTigerTable:
  21
  22     def __init__(self, conn):
  23         self.conn = conn
  24         with conn.cursor() as cur:
  25             cur.execute("""CREATE TABLE tiger (linegeo GEOMETRY,
  26                                                start INTEGER,
  27                                                stop INTEGER,
  28                                                interpol TEXT,
  29                                                token_info JSONB,
  30                                                postcode TEXT)""")
  31
  32             # We need this table to determine if the database is frozen or not
  33             cur.execute("CREATE TABLE place (number INTEGER)")
  34
  35     def count(self):
  36         return execute_scalar(self.conn, "SELECT count(*) FROM tiger")
  37
  38     def row(self):
  39         with self.conn.cursor() as cur:
  40             cur.execute("SELECT * FROM tiger LIMIT 1")
  41             return cur.fetchone()
  42
  43 @pytest.fixture
  44 def tiger_table(def_config, temp_db_conn, sql_preprocessor,
  45                 temp_db_with_extensions, tmp_path):
  46     def_config.lib_dir.sql = tmp_path / 'sql'
  47     def_config.lib_dir.sql.mkdir()
  48
  49     (def_config.lib_dir.sql / 'tiger_import_start.sql').write_text(
  50         """CREATE OR REPLACE FUNCTION tiger_line_import(linegeo GEOMETRY, start INTEGER,
  51                                                         stop INTEGER, interpol TEXT,
  52                                                         token_info JSONB, postcode TEXT)
  53            RETURNS INTEGER AS $$
  54             INSERT INTO tiger VALUES(linegeo, start, stop, interpol, token_info, postcode)
  55             RETURNING 1
  56            $$ LANGUAGE SQL;""")
  57     (def_config.lib_dir.sql / 'tiger_import_finish.sql').write_text(
  58         """DROP FUNCTION tiger_line_import (linegeo GEOMETRY, in_startnumber INTEGER,
  59                                  in_endnumber INTEGER, interpolationtype TEXT,
  60                                  token_info JSONB, in_postcode TEXT);""")
  61
  62     return MockTigerTable(temp_db_conn)
  63
  64
  65 @pytest.fixture
  66 def csv_factory(tmp_path):
  67     def _mk_file(fname, hnr_from=1, hnr_to=9, interpol='odd', street='Main St',
  68                  city='Newtown', state='AL', postcode='12345',
  69                  geometry='LINESTRING(-86.466995 32.428956,-86.466923 32.428933)'):
  70         (tmp_path / (fname + '.csv')).write_text(dedent("""\
  71         from;to;interpolation;street;city;state;postcode;geometry
  72         {};{};{};{};{};{};{};{}
  73         """.format(hnr_from, hnr_to, interpol, street, city, state,
  74                    postcode, geometry)))
  75
  76     return _mk_file
  77
  78
  79 @pytest.mark.parametrize("threads", (1, 5))
  80 @pytest.mark.asyncio
  81 async def test_add_tiger_data(def_config, src_dir, tiger_table, tokenizer_mock, threads):
  82     await tiger_data.add_tiger_data(str(src_dir / 'test' / 'testdb' / 'tiger'),
  83                                     def_config, threads, tokenizer_mock())
  84
  85     assert tiger_table.count() == 6213
  86
  87
  88 @pytest.mark.asyncio
  89 async def test_add_tiger_data_database_frozen(def_config, temp_db_conn, tiger_table, tokenizer_mock,
  90                                  tmp_path):
  91     freeze.drop_update_tables(temp_db_conn)
  92
  93     with pytest.raises(UsageError) as excinfo:
  94         await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
  95
  96         assert "database frozen" in str(excinfo.value)
  97
  98     assert tiger_table.count() == 0
  99
 100
 101 @pytest.mark.asyncio
 102 async def test_add_tiger_data_no_files(def_config, tiger_table, tokenizer_mock,
 103                                  tmp_path):
 104     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 105
 106     assert tiger_table.count() == 0
 107
 108
 109 @pytest.mark.asyncio
 110 async def test_add_tiger_data_bad_file(def_config, tiger_table, tokenizer_mock,
 111                                  tmp_path):
 112     sqlfile = tmp_path / '1010.csv'
 113     sqlfile.write_text("""Random text""")
 114
 115     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 116
 117     assert tiger_table.count() == 0
 118
 119
 120 @pytest.mark.asyncio
 121 async def test_add_tiger_data_hnr_nan(def_config, tiger_table, tokenizer_mock,
 122                                 csv_factory, tmp_path):
 123     csv_factory('file1', hnr_from=99)
 124     csv_factory('file2', hnr_from='L12')
 125     csv_factory('file3', hnr_to='12.4')
 126
 127     await tiger_data.add_tiger_data(str(tmp_path), def_config, 1, tokenizer_mock())
 128
 129     assert tiger_table.count() == 1
 130     assert tiger_table.row().start == 99
 131
 132
 133 @pytest.mark.parametrize("threads", (1, 5))
 134 @pytest.mark.asyncio
 135 async def test_add_tiger_data_tarfile(def_config, tiger_table, tokenizer_mock,
 136                                 tmp_path, src_dir, threads):
 137     tar = tarfile.open(str(tmp_path / 'sample.tar.gz'), "w:gz")
 138     tar.add(str(src_dir / 'test' / 'testdb' / 'tiger' / '01001.csv'))
 139     tar.close()
 140
 141     await tiger_data.add_tiger_data(str(tmp_path / 'sample.tar.gz'), def_config, threads,
 142                                     tokenizer_mock())
 143
 144     assert tiger_table.count() == 6213
 145
 146
 147 @pytest.mark.asyncio
 148 async def test_add_tiger_data_bad_tarfile(def_config, tiger_table, tokenizer_mock,
 149                                     tmp_path):
 150     tarfile = tmp_path / 'sample.tar.gz'
 151     tarfile.write_text("""Random text""")
 152
 153     with pytest.raises(UsageError):
 154         await tiger_data.add_tiger_data(str(tarfile), def_config, 1, tokenizer_mock())
 155
 156
 157 @pytest.mark.asyncio
 158 async def test_add_tiger_data_empty_tarfile(def_config, tiger_table, tokenizer_mock,
 159                                       tmp_path):
 160     tar = tarfile.open(str(tmp_path / 'sample.tar.gz'), "w:gz")
 161     tar.add(__file__)
 162     tar.close()
 163
 164     await tiger_data.add_tiger_data(str(tmp_path / 'sample.tar.gz'), def_config, 1,
 165                                     tokenizer_mock())
 166
 167     assert tiger_table.count() == 0