2 Test for legacy tokenizer.
9 from nominatim.indexer.place_info import PlaceInfo
10 from nominatim.tokenizer import legacy_tokenizer
11 from nominatim.db import properties
12 from nominatim.errors import UsageError
14 from mock_legacy_word_table import MockLegacyWordTable
16 # Force use of legacy word table
18 def word_table(temp_db_conn):
19 return MockLegacyWordTable(temp_db_conn)
23 def test_config(project_env, tmp_path):
24 module_dir = tmp_path / 'module_src'
26 (module_dir / 'nominatim.so').write_text('TEST nomiantim.so')
28 project_env.lib_dir.module = module_dir
30 sqldir = tmp_path / 'sql'
32 (sqldir / 'tokenizer').mkdir()
34 # Get the original SQL but replace make_standard_name to avoid module use.
35 init_sql = (project_env.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer.sql').read_text()
36 for fn in ('transliteration', 'gettokenstring'):
37 init_sql = re.sub(f'CREATE OR REPLACE FUNCTION {fn}[^;]*;',
38 '', init_sql, re.DOTALL)
40 CREATE OR REPLACE FUNCTION make_standard_name(name TEXT)
41 RETURNS TEXT AS $$ SELECT lower(name); $$ LANGUAGE SQL;
44 # Also load util functions. Some are needed by the tokenizer.
45 init_sql += (project_env.lib_dir.sql / 'functions' / 'utils.sql').read_text()
46 (sqldir / 'tokenizer' / 'legacy_tokenizer.sql').write_text(init_sql)
48 (sqldir / 'words.sql').write_text("SELECT 'a'")
50 shutil.copy(str(project_env.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer_tables.sql'),
51 str(sqldir / 'tokenizer' / 'legacy_tokenizer_tables.sql'))
53 project_env.lib_dir.sql = sqldir
54 project_env.lib_dir.data = sqldir
60 def tokenizer_factory(dsn, tmp_path, property_table):
61 (tmp_path / 'tokenizer').mkdir()
64 return legacy_tokenizer.create(dsn, tmp_path / 'tokenizer')
70 def tokenizer_setup(tokenizer_factory, test_config, monkeypatch, sql_preprocessor):
71 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
72 tok = tokenizer_factory()
73 tok.init_new_db(test_config)
77 def analyzer(tokenizer_factory, test_config, monkeypatch, sql_preprocessor,
78 word_table, temp_db_with_extensions, tmp_path):
79 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
80 monkeypatch.setenv('NOMINATIM_TERM_NORMALIZATION', ':: lower();')
81 tok = tokenizer_factory()
82 tok.init_new_db(test_config)
85 with tok.name_analyzer() as analyzer:
90 def make_standard_name(temp_db_cursor):
91 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION make_standard_name(name TEXT)
92 RETURNS TEXT AS $$ SELECT '#' || lower(name) || '#'; $$ LANGUAGE SQL""")
96 def create_postcode_id(temp_db_cursor):
97 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION create_postcode_id(postcode TEXT)
99 INSERT INTO word (word_token, word, class, type)
100 VALUES (' ' || postcode, postcode, 'place', 'postcode')
105 def test_init_new(tokenizer_factory, test_config, monkeypatch,
106 temp_db_conn, sql_preprocessor):
107 monkeypatch.setenv('NOMINATIM_TERM_NORMALIZATION', 'xxvv')
108 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
110 tok = tokenizer_factory()
111 tok.init_new_db(test_config)
113 assert properties.get_property(temp_db_conn, legacy_tokenizer.DBCFG_NORMALIZATION) == 'xxvv'
115 outfile = test_config.project_dir / 'module' / 'nominatim.so'
117 assert outfile.exists()
118 assert outfile.read_text() == 'TEST nomiantim.so'
119 assert outfile.stat().st_mode == 33261
122 def test_init_module_load_failed(tokenizer_factory, test_config):
123 tok = tokenizer_factory()
125 with pytest.raises(UsageError):
126 tok.init_new_db(test_config)
129 def test_init_module_custom(tokenizer_factory, test_config,
130 monkeypatch, tmp_path, sql_preprocessor):
131 module_dir = (tmp_path / 'custom').resolve()
133 (module_dir/ 'nominatim.so').write_text('CUSTOM nomiantim.so')
135 monkeypatch.setenv('NOMINATIM_DATABASE_MODULE_PATH', str(module_dir))
136 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
138 tok = tokenizer_factory()
139 tok.init_new_db(test_config)
141 assert not (test_config.project_dir / 'module').exists()
144 def test_init_from_project(tokenizer_setup, tokenizer_factory, test_config):
145 tok = tokenizer_factory()
147 tok.init_from_project(test_config)
149 assert tok.normalization is not None
152 def test_update_sql_functions(sql_preprocessor, temp_db_conn,
153 tokenizer_factory, test_config, table_factory,
154 monkeypatch, temp_db_cursor):
155 monkeypatch.setenv('NOMINATIM_MAX_WORD_FREQUENCY', '1133')
156 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
157 tok = tokenizer_factory()
158 tok.init_new_db(test_config)
161 assert properties.get_property(temp_db_conn, legacy_tokenizer.DBCFG_MAXWORDFREQ) == '1133'
163 table_factory('test', 'txt TEXT')
165 func_file = test_config.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer.sql'
166 func_file.write_text("""INSERT INTO test VALUES ('{{max_word_freq}}'),
167 ('{{modulepath}}')""")
169 tok.update_sql_functions(test_config)
171 test_content = temp_db_cursor.row_set('SELECT * FROM test')
172 assert test_content == set((('1133', ), (str(test_config.project_dir / 'module'), )))
175 def test_finalize_import(tokenizer_factory, temp_db_conn,
176 temp_db_cursor, test_config, monkeypatch,
177 sql_preprocessor_cfg):
178 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
180 func_file = test_config.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer_indices.sql'
181 func_file.write_text("""CREATE FUNCTION test() RETURNS TEXT
182 AS $$ SELECT 'b' $$ LANGUAGE SQL""")
184 tok = tokenizer_factory()
185 tok.init_new_db(test_config)
187 tok.finalize_import(test_config)
189 temp_db_cursor.scalar('SELECT test()') == 'b'
192 def test_migrate_database(tokenizer_factory, test_config, temp_db_conn, monkeypatch):
193 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
194 tok = tokenizer_factory()
195 tok.migrate_database(test_config)
197 assert properties.get_property(temp_db_conn, legacy_tokenizer.DBCFG_MAXWORDFREQ) is not None
198 assert properties.get_property(temp_db_conn, legacy_tokenizer.DBCFG_NORMALIZATION) is not None
200 outfile = test_config.project_dir / 'module' / 'nominatim.so'
202 assert outfile.exists()
203 assert outfile.read_text() == 'TEST nomiantim.so'
204 assert outfile.stat().st_mode == 33261
207 def test_check_database(test_config, tokenizer_factory, monkeypatch,
208 temp_db_cursor, sql_preprocessor_cfg):
209 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
210 tok = tokenizer_factory()
211 tok.init_new_db(test_config)
213 assert tok.check_database(False) is None
216 def test_check_database_no_tokenizer(test_config, tokenizer_factory):
217 tok = tokenizer_factory()
219 assert tok.check_database(False) is not None
222 def test_check_database_bad_setup(test_config, tokenizer_factory, monkeypatch,
223 temp_db_cursor, sql_preprocessor_cfg):
224 monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
225 tok = tokenizer_factory()
226 tok.init_new_db(test_config)
228 # Inject a bad transliteration.
229 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION make_standard_name(name TEXT)
230 RETURNS TEXT AS $$ SELECT 'garbage'; $$ LANGUAGE SQL""")
232 assert tok.check_database(False) is not None
235 def test_update_statistics_reverse_only(word_table, tokenizer_factory):
236 tok = tokenizer_factory()
237 tok.update_statistics()
240 def test_update_statistics(word_table, table_factory, temp_db_cursor, tokenizer_factory):
241 word_table.add_full_word(1000, 'hello')
242 table_factory('search_name',
243 'place_id BIGINT, name_vector INT[]',
245 tok = tokenizer_factory()
247 tok.update_statistics()
249 assert temp_db_cursor.scalar("""SELECT count(*) FROM word
250 WHERE word_token like ' %' and
251 search_name_count > 0""") > 0
254 def test_normalize(analyzer):
255 assert analyzer.normalize('TEsT') == 'test'
258 def test_update_postcodes_from_db_empty(analyzer, table_factory, word_table,
260 table_factory('location_postcode', 'postcode TEXT',
261 content=(('1234',), ('12 34',), ('AB23',), ('1234',)))
263 analyzer.update_postcodes_from_db()
265 assert word_table.get_postcodes() == {'1234', '12 34', 'AB23'}
268 def test_update_postcodes_from_db_add_and_remove(analyzer, table_factory, word_table,
270 table_factory('location_postcode', 'postcode TEXT',
271 content=(('1234',), ('45BC', ), ('XX45', )))
272 word_table.add_postcode(' 1234', '1234')
273 word_table.add_postcode(' 5678', '5678')
275 analyzer.update_postcodes_from_db()
277 assert word_table.get_postcodes() == {'1234', '45BC', 'XX45'}
280 def test_update_special_phrase_empty_table(analyzer, word_table, make_standard_name):
281 analyzer.update_special_phrases([
282 ("König bei", "amenity", "royal", "near"),
283 ("Könige", "amenity", "royal", "-"),
284 ("könige", "amenity", "royal", "-"),
285 ("strasse", "highway", "primary", "in")
288 assert word_table.get_special() \
289 == set(((' #könig bei#', 'könig bei', 'amenity', 'royal', 'near'),
290 (' #könige#', 'könige', 'amenity', 'royal', None),
291 (' #strasse#', 'strasse', 'highway', 'primary', 'in')))
294 def test_update_special_phrase_delete_all(analyzer, word_table, make_standard_name):
295 word_table.add_special(' #foo#', 'foo', 'amenity', 'prison', 'in')
296 word_table.add_special(' #bar#', 'bar', 'highway', 'road', None)
298 assert word_table.count_special() == 2
300 analyzer.update_special_phrases([], True)
302 assert word_table.count_special() == 0
305 def test_update_special_phrases_no_replace(analyzer, word_table, make_standard_name):
306 word_table.add_special(' #foo#', 'foo', 'amenity', 'prison', 'in')
307 word_table.add_special(' #bar#', 'bar', 'highway', 'road', None)
309 assert word_table.count_special() == 2
311 analyzer.update_special_phrases([], False)
313 assert word_table.count_special() == 2
316 def test_update_special_phrase_modify(analyzer, word_table, make_standard_name):
317 word_table.add_special(' #foo#', 'foo', 'amenity', 'prison', 'in')
318 word_table.add_special(' #bar#', 'bar', 'highway', 'road', None)
320 assert word_table.count_special() == 2
322 analyzer.update_special_phrases([
323 ('prison', 'amenity', 'prison', 'in'),
324 ('bar', 'highway', 'road', '-'),
325 ('garden', 'leisure', 'garden', 'near')
328 assert word_table.get_special() \
329 == set(((' #prison#', 'prison', 'amenity', 'prison', 'in'),
330 (' #bar#', 'bar', 'highway', 'road', None),
331 (' #garden#', 'garden', 'leisure', 'garden', 'near')))
334 def test_add_country_names(analyzer, word_table, make_standard_name):
335 analyzer.add_country_names('de', {'name': 'Germany',
336 'name:de': 'Deutschland',
337 'short_name': 'germany'})
339 assert word_table.get_country() \
340 == {('de', ' #germany#'),
341 ('de', ' #deutschland#')}
344 def test_add_more_country_names(analyzer, word_table, make_standard_name):
345 word_table.add_country('fr', ' #france#')
346 word_table.add_country('it', ' #italy#')
347 word_table.add_country('it', ' #itala#')
349 analyzer.add_country_names('it', {'name': 'Italy', 'ref': 'IT'})
351 assert word_table.get_country() \
352 == {('fr', ' #france#'),
358 @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])
359 def test_process_place_postcode(analyzer, create_postcode_id, word_table, pcode):
360 analyzer.process_place(PlaceInfo({'address': {'postcode' : pcode}}))
362 assert word_table.get_postcodes() == {pcode, }
365 @pytest.mark.parametrize('pcode', ['12:23', 'ab;cd;f', '123;836'])
366 def test_process_place_bad_postcode(analyzer, create_postcode_id, word_table, pcode):
367 analyzer.process_place(PlaceInfo({'address': {'postcode' : pcode}}))
369 assert not word_table.get_postcodes()
372 class TestHousenumberName:
375 @pytest.fixture(autouse=True)
376 def setup_create_housenumbers(temp_db_cursor):
377 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION create_housenumbers(
379 OUT tokens TEXT, OUT normtext TEXT)
381 SELECT housenumbers::TEXT, array_to_string(housenumbers, ';')
386 @pytest.mark.parametrize('hnr', ['123a', '1', '101'])
387 def test_process_place_housenumbers_simple(analyzer, hnr):
388 info = analyzer.process_place(PlaceInfo({'address': {'housenumber' : hnr}}))
390 assert info['hnr'] == hnr
391 assert info['hnr_tokens'].startswith("{")
395 def test_process_place_housenumbers_lists(analyzer):
396 info = analyzer.process_place(PlaceInfo({'address': {'conscriptionnumber' : '1; 2;3'}}))
398 assert set(info['hnr'].split(';')) == set(('1', '2', '3'))
402 def test_process_place_housenumbers_duplicates(analyzer):
403 info = analyzer.process_place(PlaceInfo({'address': {'housenumber' : '134',
404 'conscriptionnumber' : '134',
405 'streetnumber' : '99a'}}))
407 assert set(info['hnr'].split(';')) == set(('134', '99a'))
410 class TestPlaceNames:
412 @pytest.fixture(autouse=True)
413 def setup(self, analyzer):
414 self.analyzer = analyzer
417 def expect_name_terms(self, info, *expected_terms):
418 tokens = self.analyzer.get_word_token_info(list(expected_terms))
420 assert token[2] is not None, "No token for {0}".format(token)
422 assert eval(info['names']) == set((t[2] for t in tokens)),\
423 f"Expected: {tokens}\nGot: {info['names']}"
426 def process_named_place(self, names):
427 return self.analyzer.process_place(PlaceInfo({'name': names}))
430 def test_simple_names(self):
431 info = self.process_named_place({'name': 'Soft bAr', 'ref': '34'})
433 self.expect_name_terms(info, '#Soft bAr', '#34', 'Soft', 'bAr', '34')
436 @pytest.mark.parametrize('sep', [',' , ';'])
437 def test_names_with_separator(self, sep):
438 info = self.process_named_place({'name': sep.join(('New York', 'Big Apple'))})
440 self.expect_name_terms(info, '#New York', '#Big Apple',
441 'new', 'york', 'big', 'apple')
444 def test_full_names_with_bracket(self):
445 info = self.process_named_place({'name': 'Houseboat (left)'})
447 self.expect_name_terms(info, '#Houseboat (left)', '#Houseboat',
448 'houseboat', '(left)')
451 def test_country_name(self, word_table):
452 place = PlaceInfo({'name' : {'name': 'Norge'},
453 'country_code': 'no',
456 'type': 'administrative'})
458 info = self.analyzer.process_place(place)
460 self.expect_name_terms(info, '#norge', 'norge')
461 assert word_table.get_country() == {('no', ' norge')}
464 class TestPlaceAddress:
466 @pytest.fixture(autouse=True)
467 def setup(self, analyzer):
468 self.analyzer = analyzer
472 def getorcreate_hnr_id(self, temp_db_cursor):
473 temp_db_cursor.execute("""CREATE SEQUENCE seq_hnr start 1;
474 CREATE OR REPLACE FUNCTION getorcreate_housenumber_id(lookup_word TEXT)
475 RETURNS INTEGER AS $$
476 SELECT -nextval('seq_hnr')::INTEGER; $$ LANGUAGE SQL""")
478 def process_address(self, **kwargs):
479 return self.analyzer.process_place(PlaceInfo({'address': kwargs}))
482 def name_token_set(self, *expected_terms):
483 tokens = self.analyzer.get_word_token_info(list(expected_terms))
485 assert token[2] is not None, "No token for {0}".format(token)
487 return set((t[2] for t in tokens))
490 @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])
491 def test_process_place_postcode(self, word_table, pcode):
492 self.process_address(postcode=pcode)
494 assert word_table.get_postcodes() == {pcode, }
497 @pytest.mark.parametrize('pcode', ['12:23', 'ab;cd;f', '123;836'])
498 def test_process_place_bad_postcode(self, word_table, pcode):
499 self.process_address(postcode=pcode)
501 assert not word_table.get_postcodes()
504 @pytest.mark.parametrize('hnr', ['123a', '0', '101'])
505 def test_process_place_housenumbers_simple(self, hnr, getorcreate_hnr_id):
506 info = self.process_address(housenumber=hnr)
508 assert info['hnr'] == hnr.lower()
509 assert info['hnr_tokens'] == "{-1}"
512 def test_process_place_housenumbers_lists(self, getorcreate_hnr_id):
513 info = self.process_address(conscriptionnumber='1; 2;3')
515 assert set(info['hnr'].split(';')) == set(('1', '2', '3'))
516 assert info['hnr_tokens'] == "{-1,-2,-3}"
519 def test_process_place_housenumbers_duplicates(self, getorcreate_hnr_id):
520 info = self.process_address(housenumber='134',
521 conscriptionnumber='134',
524 assert set(info['hnr'].split(';')) == set(('134', '99a'))
525 assert info['hnr_tokens'] == "{-1,-2}"
528 def test_process_place_street(self):
529 # legacy tokenizer only indexes known names
530 self.analyzer.process_place(PlaceInfo({'name': {'name' : 'Grand Road'}}))
531 info = self.process_address(street='Grand Road')
533 assert eval(info['street']) == self.name_token_set('#Grand Road')
536 def test_process_place_street_empty(self):
537 info = self.process_address(street='🜵')
539 assert 'street' not in info
542 def test_process_place_place(self):
543 self.analyzer.process_place(PlaceInfo({'name': {'name' : 'Honu Lulu'}}))
544 info = self.process_address(place='Honu Lulu')
546 assert eval(info['place_search']) == self.name_token_set('#Honu Lulu',
548 assert eval(info['place_match']) == self.name_token_set('#Honu Lulu')
551 def test_process_place_place_empty(self):
552 info = self.process_address(place='🜵')
554 assert 'place' not in info
557 def test_process_place_address_terms(self):
558 for name in ('Zwickau', 'Haupstraße', 'Sachsen'):
559 self.analyzer.process_place(PlaceInfo({'name': {'name' : name}}))
560 info = self.process_address(country='de', city='Zwickau', state='Sachsen',
561 suburb='Zwickau', street='Hauptstr',
562 full='right behind the church')
564 city = self.name_token_set('ZWICKAU')
565 state = self.name_token_set('SACHSEN')
568 result = {k: eval(v[0]) for k,v in info['addr'].items()}
570 assert result == {'city': city, 'suburb': city, 'state': state}
573 def test_process_place_address_terms_empty(self):
574 info = self.process_address(country='de', city=' ', street='Hauptstr',
575 full='right behind the church')
577 assert 'addr' not in info