2 Tests for Legacy ICU tokenizer.
9 from nominatim.tokenizer import icu_tokenizer
10 import nominatim.tokenizer.icu_rule_loader
11 from nominatim.db import properties
12 from nominatim.db.sql_preprocessor import SQLPreprocessor
13 from nominatim.indexer.place_info import PlaceInfo
15 from mock_icu_word_table import MockIcuWordTable
18 def word_table(temp_db_conn):
19 return MockIcuWordTable(temp_db_conn)
23 def test_config(def_config, tmp_path):
24 def_config.project_dir = tmp_path / 'project'
25 def_config.project_dir.mkdir()
27 sqldir = tmp_path / 'sql'
29 (sqldir / 'tokenizer').mkdir()
30 (sqldir / 'tokenizer' / 'icu_tokenizer.sql').write_text("SELECT 'a'")
31 shutil.copy(str(def_config.lib_dir.sql / 'tokenizer' / 'icu_tokenizer_tables.sql'),
32 str(sqldir / 'tokenizer' / 'icu_tokenizer_tables.sql'))
34 def_config.lib_dir.sql = sqldir
40 def tokenizer_factory(dsn, tmp_path, property_table,
41 sql_preprocessor, place_table, word_table):
42 (tmp_path / 'tokenizer').mkdir()
45 return icu_tokenizer.create(dsn, tmp_path / 'tokenizer')
51 def db_prop(temp_db_conn):
52 def _get_db_property(name):
53 return properties.get_property(temp_db_conn, name)
55 return _get_db_property
59 def analyzer(tokenizer_factory, test_config, monkeypatch,
60 temp_db_with_extensions, tmp_path):
61 sql = tmp_path / 'sql' / 'tokenizer' / 'icu_tokenizer.sql'
62 sql.write_text("SELECT 'a';")
64 monkeypatch.setenv('NOMINATIM_TERM_NORMALIZATION', ':: lower();')
65 tok = tokenizer_factory()
66 tok.init_new_db(test_config)
69 def _mk_analyser(norm=("[[:Punctuation:][:Space:]]+ > ' '",), trans=(':: upper()',),
70 variants=('~gasse -> gasse', 'street => st', ),
72 cfgstr = {'normalization': list(norm),
73 'sanitizers': sanitizers,
74 'transliteration': list(trans),
75 'token-analysis': [{'analyzer': 'generic',
76 'variants': [{'words': list(variants)}]}]}
77 (test_config.project_dir / 'icu_tokenizer.yaml').write_text(yaml.dump(cfgstr))
78 tok.loader = nominatim.tokenizer.icu_rule_loader.ICURuleLoader(test_config)
80 return tok.name_analyzer()
85 def sql_functions(temp_db_conn, def_config, src_dir):
86 orig_sql = def_config.lib_dir.sql
87 def_config.lib_dir.sql = src_dir / 'lib-sql'
88 sqlproc = SQLPreprocessor(temp_db_conn, def_config)
89 sqlproc.run_sql_file(temp_db_conn, 'functions/utils.sql')
90 sqlproc.run_sql_file(temp_db_conn, 'tokenizer/icu_tokenizer.sql')
91 def_config.lib_dir.sql = orig_sql
95 def getorcreate_full_word(temp_db_cursor):
96 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION getorcreate_full_word(
97 norm_term TEXT, lookup_terms TEXT[],
99 OUT partial_tokens INT[])
102 partial_terms TEXT[] = '{}'::TEXT[];
107 SELECT min(word_id) INTO full_token
108 FROM word WHERE info->>'word' = norm_term and type = 'W';
110 IF full_token IS NULL THEN
111 full_token := nextval('seq_word');
112 INSERT INTO word (word_id, word_token, type, info)
113 SELECT full_token, lookup_term, 'W',
114 json_build_object('word', norm_term, 'count', 0)
115 FROM unnest(lookup_terms) as lookup_term;
118 FOR term IN SELECT unnest(string_to_array(unnest(lookup_terms), ' ')) LOOP
120 IF NOT (ARRAY[term] <@ partial_terms) THEN
121 partial_terms := partial_terms || term;
125 partial_tokens := '{}'::INT[];
126 FOR term IN SELECT unnest(partial_terms) LOOP
127 SELECT min(word_id), max(info->>'count') INTO term_id, term_count
128 FROM word WHERE word_token = term and type = 'w';
130 IF term_id IS NULL THEN
131 term_id := nextval('seq_word');
133 INSERT INTO word (word_id, word_token, type, info)
134 VALUES (term_id, term, 'w', json_build_object('count', term_count));
137 IF NOT (ARRAY[term_id] <@ partial_tokens) THEN
138 partial_tokens := partial_tokens || term_id;
148 def getorcreate_hnr_id(temp_db_cursor):
149 temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION getorcreate_hnr_id(lookup_term TEXT)
150 RETURNS INTEGER AS $$
151 SELECT -nextval('seq_word')::INTEGER; $$ LANGUAGE SQL""")
154 def test_init_new(tokenizer_factory, test_config, db_prop):
155 tok = tokenizer_factory()
156 tok.init_new_db(test_config)
158 assert db_prop(nominatim.tokenizer.icu_rule_loader.DBCFG_IMPORT_NORM_RULES) \
159 .startswith(':: lower ();')
162 def test_init_word_table(tokenizer_factory, test_config, place_row, temp_db_cursor):
163 place_row(names={'name' : 'Test Area', 'ref' : '52'})
164 place_row(names={'name' : 'No Area'})
165 place_row(names={'name' : 'Holzstrasse'})
167 tok = tokenizer_factory()
168 tok.init_new_db(test_config)
170 assert temp_db_cursor.table_exists('word')
173 def test_init_from_project(test_config, tokenizer_factory):
174 tok = tokenizer_factory()
175 tok.init_new_db(test_config)
177 tok = tokenizer_factory()
178 tok.init_from_project(test_config)
180 assert tok.loader is not None
183 def test_update_sql_functions(db_prop, temp_db_cursor,
184 tokenizer_factory, test_config, table_factory,
186 tok = tokenizer_factory()
187 tok.init_new_db(test_config)
189 table_factory('test', 'txt TEXT')
191 func_file = test_config.lib_dir.sql / 'tokenizer' / 'icu_tokenizer.sql'
192 func_file.write_text("""INSERT INTO test VALUES (1133)""")
194 tok.update_sql_functions(test_config)
196 test_content = temp_db_cursor.row_set('SELECT * FROM test')
197 assert test_content == set((('1133', ), ))
200 def test_normalize_postcode(analyzer):
201 with analyzer() as anl:
202 anl.normalize_postcode('123') == '123'
203 anl.normalize_postcode('ab-34 ') == 'AB-34'
204 anl.normalize_postcode('38 Б') == '38 Б'
207 def test_update_postcodes_from_db_empty(analyzer, table_factory, word_table):
208 table_factory('location_postcode', 'postcode TEXT',
209 content=(('1234',), ('12 34',), ('AB23',), ('1234',)))
211 with analyzer() as anl:
212 anl.update_postcodes_from_db()
214 assert word_table.count() == 3
215 assert word_table.get_postcodes() == {'1234', '12 34', 'AB23'}
218 def test_update_postcodes_from_db_add_and_remove(analyzer, table_factory, word_table):
219 table_factory('location_postcode', 'postcode TEXT',
220 content=(('1234',), ('45BC', ), ('XX45', )))
221 word_table.add_postcode(' 1234', '1234')
222 word_table.add_postcode(' 5678', '5678')
224 with analyzer() as anl:
225 anl.update_postcodes_from_db()
227 assert word_table.count() == 3
228 assert word_table.get_postcodes() == {'1234', '45BC', 'XX45'}
231 def test_update_special_phrase_empty_table(analyzer, word_table):
232 with analyzer() as anl:
233 anl.update_special_phrases([
234 ("König bei", "amenity", "royal", "near"),
235 ("Könige ", "amenity", "royal", "-"),
236 ("street", "highway", "primary", "in")
239 assert word_table.get_special() \
240 == {('KÖNIG BEI', 'König bei', 'amenity', 'royal', 'near'),
241 ('KÖNIGE', 'Könige', 'amenity', 'royal', None),
242 ('STREET', 'street', 'highway', 'primary', 'in')}
245 def test_update_special_phrase_delete_all(analyzer, word_table):
246 word_table.add_special('FOO', 'foo', 'amenity', 'prison', 'in')
247 word_table.add_special('BAR', 'bar', 'highway', 'road', None)
249 assert word_table.count_special() == 2
251 with analyzer() as anl:
252 anl.update_special_phrases([], True)
254 assert word_table.count_special() == 0
257 def test_update_special_phrases_no_replace(analyzer, word_table):
258 word_table.add_special('FOO', 'foo', 'amenity', 'prison', 'in')
259 word_table.add_special('BAR', 'bar', 'highway', 'road', None)
261 assert word_table.count_special() == 2
263 with analyzer() as anl:
264 anl.update_special_phrases([], False)
266 assert word_table.count_special() == 2
269 def test_update_special_phrase_modify(analyzer, word_table):
270 word_table.add_special('FOO', 'foo', 'amenity', 'prison', 'in')
271 word_table.add_special('BAR', 'bar', 'highway', 'road', None)
273 assert word_table.count_special() == 2
275 with analyzer() as anl:
276 anl.update_special_phrases([
277 ('prison', 'amenity', 'prison', 'in'),
278 ('bar', 'highway', 'road', '-'),
279 ('garden', 'leisure', 'garden', 'near')
282 assert word_table.get_special() \
283 == {('PRISON', 'prison', 'amenity', 'prison', 'in'),
284 ('BAR', 'bar', 'highway', 'road', None),
285 ('GARDEN', 'garden', 'leisure', 'garden', 'near')}
288 def test_add_country_names_new(analyzer, word_table):
289 with analyzer() as anl:
290 anl.add_country_names('es', {'name': 'Espagña', 'name:en': 'Spain'})
292 assert word_table.get_country() == {('es', 'ESPAGÑA'), ('es', 'SPAIN')}
295 def test_add_country_names_extend(analyzer, word_table):
296 word_table.add_country('ch', 'SCHWEIZ')
298 with analyzer() as anl:
299 anl.add_country_names('ch', {'name': 'Schweiz', 'name:fr': 'Suisse'})
301 assert word_table.get_country() == {('ch', 'SCHWEIZ'), ('ch', 'SUISSE')}
304 class TestPlaceNames:
306 @pytest.fixture(autouse=True)
307 def setup(self, analyzer, sql_functions):
308 sanitizers = [{'step': 'split-name-list'},
309 {'step': 'strip-brace-terms'}]
310 with analyzer(sanitizers=sanitizers) as anl:
315 def expect_name_terms(self, info, *expected_terms):
316 tokens = self.analyzer.get_word_token_info(expected_terms)
318 assert token[2] is not None, "No token for {0}".format(token)
320 assert eval(info['names']) == set((t[2] for t in tokens))
323 def process_named_place(self, names):
324 return self.analyzer.process_place(PlaceInfo({'name': names}))
327 def test_simple_names(self):
328 info = self.process_named_place({'name': 'Soft bAr', 'ref': '34'})
330 self.expect_name_terms(info, '#Soft bAr', '#34', 'Soft', 'bAr', '34')
333 @pytest.mark.parametrize('sep', [',' , ';'])
334 def test_names_with_separator(self, sep):
335 info = self.process_named_place({'name': sep.join(('New York', 'Big Apple'))})
337 self.expect_name_terms(info, '#New York', '#Big Apple',
338 'new', 'york', 'big', 'apple')
341 def test_full_names_with_bracket(self):
342 info = self.process_named_place({'name': 'Houseboat (left)'})
344 self.expect_name_terms(info, '#Houseboat (left)', '#Houseboat',
348 def test_country_name(self, word_table):
349 place = PlaceInfo({'name' : {'name': 'Norge'},
350 'country_code': 'no',
353 'type': 'administrative'})
355 info = self.analyzer.process_place(place)
357 self.expect_name_terms(info, '#norge', 'norge')
358 assert word_table.get_country() == {('no', 'NORGE')}
361 class TestPlaceAddress:
363 @pytest.fixture(autouse=True)
364 def setup(self, analyzer, sql_functions):
365 with analyzer(trans=(":: upper()", "'🜵' > ' '")) as anl:
370 def process_address(self, **kwargs):
371 return self.analyzer.process_place(PlaceInfo({'address': kwargs}))
374 def name_token_set(self, *expected_terms):
375 tokens = self.analyzer.get_word_token_info(expected_terms)
377 assert token[2] is not None, "No token for {0}".format(token)
379 return set((t[2] for t in tokens))
382 @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])
383 def test_process_place_postcode(self, word_table, pcode):
384 self.process_address(postcode=pcode)
386 assert word_table.get_postcodes() == {pcode, }
389 @pytest.mark.parametrize('pcode', ['12:23', 'ab;cd;f', '123;836'])
390 def test_process_place_bad_postcode(self, word_table, pcode):
391 self.process_address(postcode=pcode)
393 assert not word_table.get_postcodes()
396 @pytest.mark.parametrize('hnr', ['123a', '1', '101'])
397 def test_process_place_housenumbers_simple(self, hnr, getorcreate_hnr_id):
398 info = self.process_address(housenumber=hnr)
400 assert info['hnr'] == hnr.upper()
401 assert info['hnr_tokens'] == "{-1}"
404 def test_process_place_housenumbers_lists(self, getorcreate_hnr_id):
405 info = self.process_address(conscriptionnumber='1; 2;3')
407 assert set(info['hnr'].split(';')) == set(('1', '2', '3'))
408 assert info['hnr_tokens'] == "{-1,-2,-3}"
411 def test_process_place_housenumbers_duplicates(self, getorcreate_hnr_id):
412 info = self.process_address(housenumber='134',
413 conscriptionnumber='134',
416 assert set(info['hnr'].split(';')) == set(('134', '99A'))
417 assert info['hnr_tokens'] == "{-1,-2}"
420 def test_process_place_housenumbers_cached(self, getorcreate_hnr_id):
421 info = self.process_address(housenumber="45")
422 assert info['hnr_tokens'] == "{-1}"
424 info = self.process_address(housenumber="46")
425 assert info['hnr_tokens'] == "{-2}"
427 info = self.process_address(housenumber="41;45")
428 assert eval(info['hnr_tokens']) == {-1, -3}
430 info = self.process_address(housenumber="41")
431 assert eval(info['hnr_tokens']) == {-3}
434 def test_process_place_street(self):
435 info = self.process_address(street='Grand Road')
437 assert eval(info['street']) == self.name_token_set('GRAND', 'ROAD')
440 def test_process_place_street_empty(self):
441 info = self.process_address(street='🜵')
443 assert 'street' not in info
446 def test_process_place_place(self):
447 info = self.process_address(place='Honu Lulu')
449 assert eval(info['place']) == self.name_token_set('HONU', 'LULU')
452 def test_process_place_place_empty(self):
453 info = self.process_address(place='🜵')
455 assert 'place' not in info
458 def test_process_place_address_terms(self):
459 info = self.process_address(country='de', city='Zwickau', state='Sachsen',
460 suburb='Zwickau', street='Hauptstr',
461 full='right behind the church')
463 city = self.name_token_set('ZWICKAU')
464 state = self.name_token_set('SACHSEN')
466 result = {k: eval(v) for k,v in info['addr'].items()}
468 assert result == {'city': city, 'suburb': city, 'state': state}
471 def test_process_place_address_terms_empty(self):
472 info = self.process_address(country='de', city=' ', street='Hauptstr',
473 full='right behind the church')
475 assert 'addr' not in info