test/python/test_tokenizer_factory.py

   1 """
   2 Tests for creating new tokenizers.
   3 """
   4 import pytest
   5
   6 from nominatim.db import properties
   7 from nominatim.tokenizer import factory
   8 from nominatim.errors import UsageError
   9 from dummy_tokenizer import DummyTokenizer
  10
  11 @pytest.fixture
  12 def test_config(def_config, tmp_path, property_table, tokenizer_mock):
  13     def_config.project_dir = tmp_path
  14     return def_config
  15
  16
  17 def test_setup_dummy_tokenizer(temp_db_conn, test_config):
  18     tokenizer = factory.create_tokenizer(test_config)
  19
  20     assert isinstance(tokenizer, DummyTokenizer)
  21     assert tokenizer.init_state == "new"
  22     assert (test_config.project_dir / 'tokenizer').is_dir()
  23
  24     assert properties.get_property(temp_db_conn, 'tokenizer') == 'dummy'
  25
  26
  27 def test_setup_tokenizer_dir_exists(test_config):
  28     (test_config.project_dir / 'tokenizer').mkdir()
  29
  30     tokenizer = factory.create_tokenizer(test_config)
  31
  32     assert isinstance(tokenizer, DummyTokenizer)
  33     assert tokenizer.init_state == "new"
  34
  35
  36 def test_setup_tokenizer_dir_failure(test_config):
  37     (test_config.project_dir / 'tokenizer').write_text("foo")
  38
  39     with pytest.raises(UsageError):
  40         factory.create_tokenizer(test_config)
  41
  42
  43 def test_setup_bad_tokenizer_name(def_config, tmp_path, monkeypatch):
  44     def_config.project_dir = tmp_path
  45     monkeypatch.setenv('NOMINATIM_TOKENIZER', 'dummy')
  46
  47     with pytest.raises(UsageError):
  48         factory.create_tokenizer(def_config)
  49
  50
  51 def test_load_tokenizer(test_config):
  52     factory.create_tokenizer(test_config)
  53
  54     tokenizer = factory.get_tokenizer_for_db(test_config)
  55
  56     assert isinstance(tokenizer, DummyTokenizer)
  57     assert tokenizer.init_state == "loaded"
  58
  59
  60 def test_load_no_tokenizer_dir(test_config):
  61     factory.create_tokenizer(test_config)
  62
  63     test_config.project_dir = test_config.project_dir / 'foo'
  64
  65     with pytest.raises(UsageError):
  66         factory.get_tokenizer_for_db(test_config)
  67
  68
  69 def test_load_missing_propoerty(temp_db_cursor, test_config):
  70     factory.create_tokenizer(test_config)
  71
  72     temp_db_cursor.execute("TRUNCATE TABLE nominatim_properties")
  73
  74     with pytest.raises(UsageError):
  75         factory.get_tokenizer_for_db(test_config)