]> git.openstreetmap.org Git - nominatim.git/blobdiff - test/python/tokenizer/test_legacy.py
make localisation of results explicit
[nominatim.git] / test / python / tokenizer / test_legacy.py
index f90fc49402a6d235dd74e1081a7c98ace0587e29..57a82b8a0f4aa5af2f1dc6c2377db0fca8d83959 100644 (file)
@@ -1,28 +1,58 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.
 """
 Test for legacy tokenizer.
 """
 import shutil
+import re
 
 import pytest
 
-from nominatim.indexer.place_info import PlaceInfo
+from nominatim.data.place_info import PlaceInfo
 from nominatim.tokenizer import legacy_tokenizer
 from nominatim.db import properties
 from nominatim.errors import UsageError
 
+from mock_legacy_word_table import MockLegacyWordTable
+
+# Force use of legacy word table
+@pytest.fixture
+def word_table(temp_db_conn):
+    return MockLegacyWordTable(temp_db_conn)
+
+
 @pytest.fixture
 def test_config(project_env, tmp_path):
     module_dir = tmp_path / 'module_src'
     module_dir.mkdir()
-    (module_dir / 'nominatim.so').write_text('TEST nomiantim.so')
+    (module_dir / 'nominatim.so').write_text('TEST nominatim.so')
 
     project_env.lib_dir.module = module_dir
 
     sqldir = tmp_path / 'sql'
     sqldir.mkdir()
     (sqldir / 'tokenizer').mkdir()
-    (sqldir / 'tokenizer' / 'legacy_tokenizer.sql').write_text("SELECT 'a'")
+
+    # Get the original SQL but replace make_standard_name to avoid module use.
+    init_sql = (project_env.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer.sql').read_text()
+    for fn in ('transliteration', 'gettokenstring'):
+        init_sql = re.sub(f'CREATE OR REPLACE FUNCTION {fn}[^;]*;',
+                          '', init_sql, re.DOTALL)
+    init_sql += """
+                   CREATE OR REPLACE FUNCTION make_standard_name(name TEXT)
+                   RETURNS TEXT AS $$ SELECT lower(name); $$ LANGUAGE SQL;
+
+                """
+    # Also load util functions. Some are needed by the tokenizer.
+    init_sql += (project_env.lib_dir.sql / 'functions' / 'utils.sql').read_text()
+    (sqldir / 'tokenizer' / 'legacy_tokenizer.sql').write_text(init_sql)
+
     (sqldir / 'words.sql').write_text("SELECT 'a'")
+
     shutil.copy(str(project_env.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer_tables.sql'),
                 str(sqldir / 'tokenizer' / 'legacy_tokenizer_tables.sql'))
 
@@ -52,12 +82,6 @@ def tokenizer_setup(tokenizer_factory, test_config, monkeypatch, sql_preprocesso
 @pytest.fixture
 def analyzer(tokenizer_factory, test_config, monkeypatch, sql_preprocessor,
              word_table, temp_db_with_extensions, tmp_path):
-    sql = tmp_path / 'sql' / 'tokenizer' / 'legacy_tokenizer.sql'
-    sql.write_text("""
-        CREATE OR REPLACE FUNCTION getorcreate_housenumber_id(lookup_word TEXT)
-          RETURNS INTEGER AS $$ SELECT 342; $$ LANGUAGE SQL;
-        """)
-
     monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
     monkeypatch.setenv('NOMINATIM_TERM_NORMALIZATION', ':: lower();')
     tok = tokenizer_factory()
@@ -84,12 +108,6 @@ def create_postcode_id(temp_db_cursor):
                               $$ LANGUAGE SQL""")
 
 
-@pytest.fixture
-def make_keywords(temp_db_cursor, temp_db_with_extensions):
-    temp_db_cursor.execute(
-        """CREATE OR REPLACE FUNCTION make_keywords(names HSTORE)
-           RETURNS INTEGER[] AS $$ SELECT ARRAY[1, 2, 3] $$ LANGUAGE SQL""")
-
 def test_init_new(tokenizer_factory, test_config, monkeypatch,
                   temp_db_conn, sql_preprocessor):
     monkeypatch.setenv('NOMINATIM_TERM_NORMALIZATION', 'xxvv')
@@ -103,7 +121,7 @@ def test_init_new(tokenizer_factory, test_config, monkeypatch,
     outfile = test_config.project_dir / 'module' / 'nominatim.so'
 
     assert outfile.exists()
-    assert outfile.read_text() == 'TEST nomiantim.so'
+    assert outfile.read_text() == 'TEST nominatim.so'
     assert outfile.stat().st_mode == 33261
 
 
@@ -160,6 +178,23 @@ def test_update_sql_functions(sql_preprocessor, temp_db_conn,
     assert test_content == set((('1133', ), (str(test_config.project_dir / 'module'), )))
 
 
+def test_finalize_import(tokenizer_factory, temp_db_conn,
+                         temp_db_cursor, test_config, monkeypatch,
+                         sql_preprocessor_cfg):
+    monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
+
+    func_file = test_config.lib_dir.sql / 'tokenizer' / 'legacy_tokenizer_indices.sql'
+    func_file.write_text("""CREATE FUNCTION test() RETURNS TEXT
+                            AS $$ SELECT 'b'::text $$ LANGUAGE SQL""")
+
+    tok = tokenizer_factory()
+    tok.init_new_db(test_config)
+
+    tok.finalize_import(test_config)
+
+    temp_db_cursor.scalar('SELECT test()') == 'b'
+
+
 def test_migrate_database(tokenizer_factory, test_config, temp_db_conn, monkeypatch):
     monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
     tok = tokenizer_factory()
@@ -171,10 +206,64 @@ def test_migrate_database(tokenizer_factory, test_config, temp_db_conn, monkeypa
     outfile = test_config.project_dir / 'module' / 'nominatim.so'
 
     assert outfile.exists()
-    assert outfile.read_text() == 'TEST nomiantim.so'
+    assert outfile.read_text() == 'TEST nominatim.so'
     assert outfile.stat().st_mode == 33261
 
 
+def test_check_database(test_config, tokenizer_factory, monkeypatch,
+                        temp_db_cursor, sql_preprocessor_cfg):
+    monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
+    tok = tokenizer_factory()
+    tok.init_new_db(test_config)
+
+    assert tok.check_database(False) is None
+
+
+def test_check_database_no_tokenizer(test_config, tokenizer_factory):
+    tok = tokenizer_factory()
+
+    assert tok.check_database(False) is not None
+
+
+def test_check_database_bad_setup(test_config, tokenizer_factory, monkeypatch,
+                                  temp_db_cursor, sql_preprocessor_cfg):
+    monkeypatch.setattr(legacy_tokenizer, '_check_module', lambda m, c: None)
+    tok = tokenizer_factory()
+    tok.init_new_db(test_config)
+
+    # Inject a bad transliteration.
+    temp_db_cursor.execute("""CREATE OR REPLACE FUNCTION make_standard_name(name TEXT)
+                              RETURNS TEXT AS $$ SELECT 'garbage'::text; $$ LANGUAGE SQL""")
+
+    assert tok.check_database(False) is not None
+
+
+def test_update_statistics_reverse_only(word_table, tokenizer_factory):
+    tok = tokenizer_factory()
+    tok.update_statistics()
+
+
+def test_update_statistics(word_table, table_factory, temp_db_cursor, tokenizer_factory):
+    word_table.add_full_word(1000, 'hello')
+    table_factory('search_name',
+                  'place_id BIGINT, name_vector INT[]',
+                  [(12, [1000])])
+    tok = tokenizer_factory()
+
+    tok.update_statistics()
+
+    assert temp_db_cursor.scalar("""SELECT count(*) FROM word
+                                    WHERE word_token like ' %' and
+                                          search_name_count > 0""") > 0
+
+
+def test_update_word_tokens(tokenizer_factory):
+    tok = tokenizer_factory()
+
+    # This is a noop and should just pass.
+    tok.update_word_tokens()
+
+
 def test_normalize(analyzer):
     assert analyzer.normalize('TEsT') == 'test'
 
@@ -186,7 +275,6 @@ def test_update_postcodes_from_db_empty(analyzer, table_factory, word_table,
 
     analyzer.update_postcodes_from_db()
 
-    assert word_table.count() == 3
     assert word_table.get_postcodes() == {'1234', '12 34', 'AB23'}
 
 
@@ -199,7 +287,6 @@ def test_update_postcodes_from_db_add_and_remove(analyzer, table_factory, word_t
 
     analyzer.update_postcodes_from_db()
 
-    assert word_table.count() == 3
     assert word_table.get_postcodes() == {'1234', '45BC', 'XX45'}
 
 
@@ -281,12 +368,6 @@ def test_add_more_country_names(analyzer, word_table, make_standard_name):
                    ('it', ' #it#')}
 
 
-def test_process_place_names(analyzer, make_keywords):
-    info = analyzer.process_place(PlaceInfo({'name' : {'name' : 'Soft bAr', 'ref': '34'}}))
-
-    assert info['names'] == '{1,2,3}'
-
-
 @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])
 def test_process_place_postcode(analyzer, create_postcode_id, word_table, pcode):
     analyzer.process_place(PlaceInfo({'address': {'postcode' : pcode}}))
@@ -337,3 +418,174 @@ class TestHousenumberName:
                                                    'streetnumber' : '99a'}}))
 
         assert set(info['hnr'].split(';')) == set(('134', '99a'))
+
+
+class TestPlaceNames:
+
+    @pytest.fixture(autouse=True)
+    def setup(self, analyzer):
+        self.analyzer = analyzer
+
+
+    def expect_name_terms(self, info, *expected_terms):
+        tokens = self.analyzer.get_word_token_info(list(expected_terms))
+        for token in tokens:
+            assert token[2] is not None, "No token for {0}".format(token)
+
+        assert eval(info['names']) == set((t[2] for t in tokens)),\
+               f"Expected: {tokens}\nGot: {info['names']}"
+
+
+    def process_named_place(self, names):
+        return self.analyzer.process_place(PlaceInfo({'name': names}))
+
+
+    def test_simple_names(self):
+        info = self.process_named_place({'name': 'Soft bAr', 'ref': '34'})
+
+        self.expect_name_terms(info, '#Soft bAr', '#34', 'Soft', 'bAr', '34')
+
+
+    @pytest.mark.parametrize('sep', [',' , ';'])
+    def test_names_with_separator(self, sep):
+        info = self.process_named_place({'name': sep.join(('New York', 'Big Apple'))})
+
+        self.expect_name_terms(info, '#New York', '#Big Apple',
+                                     'new', 'york', 'big', 'apple')
+
+
+    def test_full_names_with_bracket(self):
+        info = self.process_named_place({'name': 'Houseboat (left)'})
+
+        self.expect_name_terms(info, '#Houseboat (left)', '#Houseboat',
+                                     'houseboat', '(left)')
+
+
+    def test_country_name(self, word_table):
+        place = PlaceInfo({'name' : {'name': 'Norge'},
+                           'country_code': 'no',
+                           'rank_address': 4,
+                           'class': 'boundary',
+                           'type': 'administrative'})
+
+        info = self.analyzer.process_place(place)
+
+        self.expect_name_terms(info, '#norge', 'norge')
+        assert word_table.get_country() == {('no', ' norge')}
+
+
+class TestPlaceAddress:
+
+    @pytest.fixture(autouse=True)
+    def setup(self, analyzer):
+        self.analyzer = analyzer
+
+
+    @pytest.fixture
+    def getorcreate_hnr_id(self, temp_db_cursor):
+        temp_db_cursor.execute("""CREATE SEQUENCE seq_hnr start 1;
+                                  CREATE OR REPLACE FUNCTION getorcreate_housenumber_id(lookup_word TEXT)
+                                  RETURNS INTEGER AS $$
+                                  SELECT -nextval('seq_hnr')::INTEGER; $$ LANGUAGE SQL""")
+
+    def process_address(self, **kwargs):
+        return self.analyzer.process_place(PlaceInfo({'address': kwargs}))
+
+
+    def name_token_set(self, *expected_terms):
+        tokens = self.analyzer.get_word_token_info(list(expected_terms))
+        for token in tokens:
+            assert token[2] is not None, "No token for {0}".format(token)
+
+        return set((t[2] for t in tokens))
+
+
+    @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])
+    def test_process_place_postcode(self, word_table, pcode):
+        self.process_address(postcode=pcode)
+
+        assert word_table.get_postcodes() == {pcode, }
+
+
+    @pytest.mark.parametrize('pcode', ['12:23', 'ab;cd;f', '123;836'])
+    def test_process_place_bad_postcode(self, word_table, pcode):
+        self.process_address(postcode=pcode)
+
+        assert not word_table.get_postcodes()
+
+
+    @pytest.mark.parametrize('hnr', ['123a', '0', '101'])
+    def test_process_place_housenumbers_simple(self, hnr, getorcreate_hnr_id):
+        info = self.process_address(housenumber=hnr)
+
+        assert info['hnr'] == hnr.lower()
+        assert info['hnr_tokens'] == "{-1}"
+
+
+    def test_process_place_housenumbers_lists(self, getorcreate_hnr_id):
+        info = self.process_address(conscriptionnumber='1; 2;3')
+
+        assert set(info['hnr'].split(';')) == set(('1', '2', '3'))
+        assert info['hnr_tokens'] == "{-1,-2,-3}"
+
+
+    def test_process_place_housenumbers_duplicates(self, getorcreate_hnr_id):
+        info = self.process_address(housenumber='134',
+                                    conscriptionnumber='134',
+                                    streetnumber='99A')
+
+        assert set(info['hnr'].split(';')) == set(('134', '99a'))
+        assert info['hnr_tokens'] == "{-1,-2}"
+
+
+    def test_process_place_street(self):
+        # legacy tokenizer only indexes known names
+        self.analyzer.process_place(PlaceInfo({'name': {'name' : 'Grand Road'}}))
+        info = self.process_address(street='Grand Road')
+
+        assert eval(info['street']) == self.name_token_set('#Grand Road')
+
+
+    def test_process_place_street_empty(self):
+        info = self.process_address(street='🜵')
+
+        assert 'street' not in info
+
+
+    def test_process_place_place(self):
+        self.analyzer.process_place(PlaceInfo({'name': {'name' : 'Honu Lulu'}}))
+        info = self.process_address(place='Honu Lulu')
+
+        assert eval(info['place_search']) == self.name_token_set('#Honu Lulu',
+                                                                 'Honu', 'Lulu')
+        assert eval(info['place_match']) == self.name_token_set('#Honu Lulu')
+
+
+    def test_process_place_place_empty(self):
+        info = self.process_address(place='🜵')
+
+        assert 'place' not in info
+
+
+    def test_process_place_address_terms(self):
+        for name in ('Zwickau', 'Haupstraße', 'Sachsen'):
+            self.analyzer.process_place(PlaceInfo({'name': {'name' : name}}))
+        info = self.process_address(country='de', city='Zwickau', state='Sachsen',
+                                    suburb='Zwickau', street='Hauptstr',
+                                    full='right behind the church')
+
+        city = self.name_token_set('ZWICKAU')
+        state = self.name_token_set('SACHSEN')
+
+        print(info)
+        result = {k: eval(v[0]) for k,v in info['addr'].items()}
+
+        assert result == {'city': city, 'suburb': city, 'state': state}
+
+
+    def test_process_place_address_terms_empty(self):
+        info = self.process_address(country='de', city=' ', street='Hauptstr',
+                                    full='right behind the church')
+
+        assert 'addr' not in info
+