X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/81eed0680cd2a8b5c53217ab2996a43ab17f5056..8f3845660f18bdbf2dd42dd2c6db6c7fa5160f3e:/test/python/tokenizer/test_icu.py?ds=sidebyside diff --git a/test/python/tokenizer/test_icu.py b/test/python/tokenizer/test_icu.py index aa1afe16..2a4865db 100644 --- a/test/python/tokenizer/test_icu.py +++ b/test/python/tokenizer/test_icu.py @@ -227,16 +227,20 @@ def test_update_statistics_reverse_only(word_table, tokenizer_factory, test_conf def test_update_statistics(word_table, table_factory, temp_db_cursor, tokenizer_factory, test_config): word_table.add_full_word(1000, 'hello') + word_table.add_full_word(1001, 'bye') table_factory('search_name', - 'place_id BIGINT, name_vector INT[]', - [(12, [1000])]) + 'place_id BIGINT, name_vector INT[], nameaddress_vector INT[]', + [(12, [1000], [1001])]) tok = tokenizer_factory() tok.update_statistics(test_config) assert temp_db_cursor.scalar("""SELECT count(*) FROM word - WHERE type = 'W' and - (info->>'count')::int > 0""") > 0 + WHERE type = 'W' and word_id = 1000 and + (info->>'count')::int > 0""") == 1 + assert temp_db_cursor.scalar("""SELECT count(*) FROM word + WHERE type = 'W' and word_id = 1001 and + (info->>'addr_count')::int > 0""") == 1 def test_normalize_postcode(analyzer): @@ -550,7 +554,7 @@ class TestPlaceAddress: def test_process_place_place(self): info = self.process_address(place='Honu Lulu') - assert eval(info['place']) == self.name_token_set('HONU', 'LULU') + assert eval(info['place']) == self.name_token_set('HONU', 'LULU', '#HONU LULU') def test_process_place_place_extra(self): @@ -570,8 +574,8 @@ class TestPlaceAddress: suburb='Zwickau', street='Hauptstr', full='right behind the church') - city = self.name_token_set('ZWICKAU') - state = self.name_token_set('SACHSEN') + city = self.name_token_set('ZWICKAU', '#ZWICKAU') + state = self.name_token_set('SACHSEN', '#SACHSEN') result = {k: eval(v) for k,v in info['addr'].items()} @@ -583,7 +587,7 @@ class TestPlaceAddress: result = {k: eval(v) for k,v in info['addr'].items()} - assert result == {'city': self.name_token_set('Bruxelles')} + assert result == {'city': self.name_token_set('Bruxelles', '#Bruxelles')} def test_process_place_address_terms_empty(self):