add unit tests for new sanatizer functions

[nominatim.git] / test / python / test_tools_import_special_phrases.py

diff --git a/test/python/test_tools_import_special_phrases.py b/test/python/test_tools_import_special_phrases.py
index 058e170b9d6d4f6133bfce6a2c508db2b2330db7..f0a34b0896db1102c1a54d7dae991f4ee02673fe 100644 (file)
--- a/test/python/test_tools_import_special_phrases.py
+++ b/test/python/test_tools_import_special_phrases.py
@@ -1,164 +1,300 @@
 """
 """

-    Tests for import special phrases functions
+    Tests for import special phrases methods
+    of the class SPImporter.

 """
 """

+from shutil import copyfile

 import pytest
 import pytest

-from nominatim.tools.special_phrases import _create_place_classtype_indexes, _create_place_classtype_table, _get_wiki_content, _grant_access_to_webuser, _process_amenity
+from nominatim.tools.special_phrases.sp_importer import SPImporter
+from nominatim.tools.special_phrases.sp_wiki_loader import SPWikiLoader
+from nominatim.tools.special_phrases.special_phrase import SpecialPhrase
+from nominatim.errors import UsageError

 
 

-def test_get_wiki_content():
-    assert _get_wiki_content('fr')
+from cursor import CursorForTesting

 
 

-def execute_and_verify_add_word(temp_db_conn, phrase_label, phrase_class, phrase_type):
-    _process_amenity(temp_db_conn, phrase_label, phrase_class, phrase_type, '')
+@pytest.fixture
+def testfile_dir(src_dir):
+    return src_dir / 'test' / 'testfiles'

 
 

-    with temp_db_conn.cursor() as temp_db_cursor:
-        temp_db_cursor.execute(f"""
-            SELECT * FROM word 
-            WHERE word_token=' {phrase_label}'
-            AND word='{phrase_label}'
-            AND class='{phrase_class}'
-            AND type='{phrase_type}'""")
-        return temp_db_cursor.fetchone()

 
 

-def execute_and_verify_add_word_with_operator(temp_db_conn, phrase_label, phrase_class, phrase_type, phrase_operator):
-    _process_amenity(temp_db_conn, phrase_label, phrase_class, phrase_type, phrase_operator)
+@pytest.fixture
+def sp_importer(temp_db_conn, def_config, temp_phplib_dir_with_migration):
+    """
+        Return an instance of SPImporter.
+    """
+    loader = SPWikiLoader(def_config, ['en'])
+    return SPImporter(def_config, temp_phplib_dir_with_migration, temp_db_conn, loader)

 
 

-    with temp_db_conn.cursor() as temp_db_cursor:
-        temp_db_cursor.execute(f"""
-            SELECT * FROM word 
-            WHERE word_token=' {phrase_label}'
-            AND word='{phrase_label}'
-            AND class='{phrase_class}'
-            AND type='{phrase_type}'
-            AND operator='{phrase_operator}'""")
-        return temp_db_cursor.fetchone()

 
 

-def test_process_amenity_with_near_operator(temp_db_conn, word_table, amenity_operator_funcs):
-    phrase_label = 'label'
-    phrase_class = 'class'
-    phrase_type = 'type'
+@pytest.fixture
+def temp_phplib_dir_with_migration(src_dir, tmp_path):
+    """
+        Return temporary phpdir with migration subdirectory and
+        PhraseSettingsToJson.php script inside.
+    """
+    migration_file = (src_dir / 'lib-php' / 'migration' / 'PhraseSettingsToJson.php').resolve()
+
+    phpdir = tmp_path / 'tempphp'
+    phpdir.mkdir()
+
+    (phpdir / 'migration').mkdir()
+    migration_dest_path = (phpdir / 'migration' / 'PhraseSettingsToJson.php').resolve()
+    copyfile(str(migration_file), str(migration_dest_path))
+
+    return phpdir
+
+
+@pytest.fixture
+def xml_wiki_content(src_dir):
+    """
+        return the content of the static xml test file.
+    """
+    xml_test_content = src_dir / 'test' / 'testdata' / 'special_phrases_test_content.txt'
+    return xml_test_content.read_text()
+
+
+@pytest.fixture
+def default_phrases(table_factory):
+    table_factory('place_classtype_testclasstypetable_to_delete')
+    table_factory('place_classtype_testclasstypetable_to_keep')
+
+
+def test_fetch_existing_place_classtype_tables(sp_importer, table_factory):
+    """
+        Check for the fetch_existing_place_classtype_tables() method.
+        It should return the table just created.
+    """
+    table_factory('place_classtype_testclasstypetable')
+
+    sp_importer._fetch_existing_place_classtype_tables()
+    contained_table = sp_importer.table_phrases_to_delete.pop()
+    assert contained_table == 'place_classtype_testclasstypetable'
+
+def test_check_sanity_class(sp_importer):
+    """
+        Check for _check_sanity() method.
+        If a wrong class or type is given, an UsageError should raise.
+        If a good class and type are given, nothing special happens.
+    """
+
+    assert not sp_importer._check_sanity(SpecialPhrase('en', '', 'type', ''))
+    assert not sp_importer._check_sanity(SpecialPhrase('en', 'class', '', ''))
+
+    assert sp_importer._check_sanity(SpecialPhrase('en', 'class', 'type', ''))
+
+def test_load_white_and_black_lists(sp_importer):
+    """
+        Test that _load_white_and_black_lists() well return
+        black list and white list and that they are of dict type.
+    """
+    black_list, white_list = sp_importer._load_white_and_black_lists()
+
+    assert isinstance(black_list, dict) and isinstance(white_list, dict)
+
+def test_convert_php_settings(sp_importer, testfile_dir, tmp_path):
+    """
+        Test that _convert_php_settings_if_needed() convert the given
+        php file to a json file.
+    """
+    php_file = (testfile_dir / 'phrase_settings.php').resolve()
+
+    temp_settings = (tmp_path / 'phrase_settings.php').resolve()
+    copyfile(php_file, temp_settings)
+    sp_importer._convert_php_settings_if_needed(temp_settings)
+
+    assert (tmp_path / 'phrase_settings.json').is_file()
+
+def test_convert_settings_wrong_file(sp_importer):
+    """
+        Test that _convert_php_settings_if_needed() raise an exception
+        if the given file is not a valid file.
+    """
+    with pytest.raises(UsageError, match='random_file is not a valid file.'):
+        sp_importer._convert_php_settings_if_needed('random_file')
+
+def test_convert_settings_json_already_exist(sp_importer, testfile_dir):
+    """
+        Test that if we give to '_convert_php_settings_if_needed' a php file path
+        and that a the corresponding json file already exists, it is returned.
+    """
+    php_file = (testfile_dir / 'phrase_settings.php').resolve()
+    json_file = (testfile_dir / 'phrase_settings.json').resolve()
+
+    returned = sp_importer._convert_php_settings_if_needed(php_file)
+
+    assert returned == json_file

 
 

-    assert execute_and_verify_add_word(temp_db_conn, phrase_label, phrase_class, phrase_type)
-    assert execute_and_verify_add_word_with_operator(temp_db_conn, phrase_label, phrase_class, phrase_type, 'near')
-    assert execute_and_verify_add_word_with_operator(temp_db_conn, phrase_label, phrase_class, phrase_type, 'in')
-
-def index_exists(db_connect, index):
-        """ Check that an index with the given name exists in the database.
-        """
-        with db_connect.cursor() as cur:
-            cur.execute("""SELECT tablename FROM pg_indexes
-                           WHERE indexname = %s and schemaname = 'public'""", (index, ))
-            if cur.rowcount == 0:
-                return False
-        return True
-
-def test_create_place_classtype_indexes(temp_db_conn):
+def test_convert_settings_giving_json(sp_importer, testfile_dir):
+    """
+        Test that if we give to '_convert_php_settings_if_needed' a json file path
+        the same path is directly returned
+    """
+    json_file = (testfile_dir / 'phrase_settings.json').resolve()
+
+    returned = sp_importer._convert_php_settings_if_needed(json_file)
+
+    assert returned == json_file
+
+def test_create_place_classtype_indexes(temp_db_with_extensions, temp_db_conn,
+                                        table_factory, sp_importer):
+    """
+        Test that _create_place_classtype_indexes() create the
+        place_id index and centroid index on the right place_class_type table.
+    """

     phrase_class = 'class'
     phrase_type = 'type'
     phrase_class = 'class'
     phrase_type = 'type'

-    table_name = f'place_classtype_{phrase_class}_{phrase_type}'
-
-    with temp_db_conn.cursor() as temp_db_cursor:
-        temp_db_cursor.execute("CREATE EXTENSION postgis;")
-        temp_db_cursor.execute(f'CREATE TABLE {table_name}(place_id BIGINT, centroid GEOMETRY)')
+    table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type)

 
 

-    _create_place_classtype_indexes(temp_db_conn, '', phrase_class, phrase_type)
+    table_factory(table_name, 'place_id BIGINT, centroid GEOMETRY')

 
 

-    centroid_index_exists = index_exists(temp_db_conn, f'idx_place_classtype_{phrase_class}_{phrase_type}_centroid')
-    place_id_index_exists = index_exists(temp_db_conn, f'idx_place_classtype_{phrase_class}_{phrase_type}_place_id')
+    sp_importer._create_place_classtype_indexes('', phrase_class, phrase_type)

 
 

-    assert centroid_index_exists and place_id_index_exists
+    assert check_placeid_and_centroid_indexes(temp_db_conn, phrase_class, phrase_type)

 
 

-def test_create_place_classtype_table(temp_db_conn, placex_table):
+def test_create_place_classtype_table(temp_db_conn, placex_table, sp_importer):
+    """
+        Test that _create_place_classtype_table() create
+        the right place_classtype table.
+    """

     phrase_class = 'class'
     phrase_type = 'type'
     phrase_class = 'class'
     phrase_type = 'type'

-    _create_place_classtype_table(temp_db_conn, '', phrase_class, phrase_type)
+    sp_importer._create_place_classtype_table('', phrase_class, phrase_type)

 
 

-    with temp_db_conn.cursor() as temp_db_cursor:
-        temp_db_cursor.execute(f"""
-            SELECT *
-            FROM information_schema.tables
-            WHERE table_type='BASE TABLE'
-            AND table_name='place_classtype_{phrase_class}_{phrase_type}'""")
-        result = temp_db_cursor.fetchone()
-    assert result
-
-def test_grant_access_to_web_user(temp_db_conn, def_config):
+    assert check_table_exist(temp_db_conn, phrase_class, phrase_type)
+
+def test_grant_access_to_web_user(temp_db_conn, table_factory, def_config, sp_importer):
+    """
+        Test that _grant_access_to_webuser() give
+        right access to the web user.
+    """

     phrase_class = 'class'
     phrase_type = 'type'
     phrase_class = 'class'
     phrase_type = 'type'

-    table_name = f'place_classtype_{phrase_class}_{phrase_type}'
+    table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type)

 
 

-    with temp_db_conn.cursor() as temp_db_cursor:
-        temp_db_cursor.execute(f'CREATE TABLE {table_name}()')
+    table_factory(table_name)
+
+    sp_importer._grant_access_to_webuser(phrase_class, phrase_type)
+
+    assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, phrase_class, phrase_type)
+
+def test_create_place_classtype_table_and_indexes(
+        temp_db_conn, def_config, placex_table,
+        sp_importer):
+    """
+        Test that _create_place_classtype_table_and_indexes()
+        create the right place_classtype tables and place_id indexes
+        and centroid indexes and grant access to the web user
+        for the given set of pairs.
+    """
+    pairs = set([('class1', 'type1'), ('class2', 'type2')])
+
+    sp_importer._create_place_classtype_table_and_indexes(pairs)
+
+    for pair in pairs:
+        assert check_table_exist(temp_db_conn, pair[0], pair[1])
+        assert check_placeid_and_centroid_indexes(temp_db_conn, pair[0], pair[1])
+        assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, pair[0], pair[1])
+
+def test_remove_non_existent_tables_from_db(sp_importer, default_phrases,
+                                            temp_db_conn):
+    """
+        Check for the remove_non_existent_phrases_from_db() method.
+
+        It should removed entries from the word table which are contained
+        in the words_phrases_to_delete set and not those also contained
+        in the words_phrases_still_exist set.
+
+        place_classtype tables contained in table_phrases_to_delete should
+        be deleted.
+    """
+    sp_importer.table_phrases_to_delete = {
+        'place_classtype_testclasstypetable_to_delete'
+    }

 
 

-    _grant_access_to_webuser(temp_db_conn, def_config, phrase_class, phrase_type)
+    query_tables = """
+        SELECT table_name
+        FROM information_schema.tables
+        WHERE table_schema='public'
+        AND table_name like 'place_classtype_%';
+    """
+
+    sp_importer._remove_non_existent_tables_from_db()
+
+    # Changes are not committed yet. Use temp_db_conn for checking results.
+    with temp_db_conn.cursor(cursor_factory=CursorForTesting) as cur:
+        assert cur.row_set(query_tables) \
+                 == {('place_classtype_testclasstypetable_to_keep', )}
+
+
+@pytest.mark.parametrize("should_replace", [(True), (False)])
+def test_import_phrases(monkeypatch, temp_db_conn, def_config, sp_importer,
+                        placex_table, table_factory, tokenizer_mock,
+                        xml_wiki_content, should_replace):
+    """
+        Check that the main import_phrases() method is well executed.
+        It should create the place_classtype table, the place_id and centroid indexes,
+        grand access to the web user and executing the SQL functions for amenities.
+        It should also update the database well by deleting or preserving existing entries
+        of the database.
+    """
+    #Add some data to the database before execution in order to test
+    #what is deleted and what is preserved.
+    table_factory('place_classtype_amenity_animal_shelter')
+    table_factory('place_classtype_wrongclass_wrongtype')
+
+    monkeypatch.setattr('nominatim.tools.special_phrases.sp_wiki_loader.SPWikiLoader._get_wiki_content',
+                        lambda self, lang: xml_wiki_content)
+
+    tokenizer = tokenizer_mock()
+    sp_importer.import_phrases(tokenizer, should_replace)
+
+    assert len(tokenizer.analyser_cache['special_phrases']) == 18
+
+    class_test = 'aerialway'
+    type_test = 'zip_line'
+
+    assert check_table_exist(temp_db_conn, class_test, type_test)
+    assert check_placeid_and_centroid_indexes(temp_db_conn, class_test, type_test)
+    assert check_grant_access(temp_db_conn, def_config.DATABASE_WEBUSER, class_test, type_test)
+    assert check_table_exist(temp_db_conn, 'amenity', 'animal_shelter')
+    if should_replace:
+        assert not check_table_exist(temp_db_conn, 'wrong_class', 'wrong_type')
+
+    assert temp_db_conn.table_exists('place_classtype_amenity_animal_shelter')
+    if should_replace:
+        assert not temp_db_conn.table_exists('place_classtype_wrongclass_wrongtype')
+
+def check_table_exist(temp_db_conn, phrase_class, phrase_type):
+    """
+        Verify that the place_classtype table exists for the given
+        phrase_class and phrase_type.
+    """
+    return temp_db_conn.table_exists('place_classtype_{}_{}'.format(phrase_class, phrase_type))
+
+
+def check_grant_access(temp_db_conn, user, phrase_class, phrase_type):
+    """
+        Check that the web user has been granted right access to the
+        place_classtype table of the given phrase_class and phrase_type.
+    """
+    table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type)

 
     with temp_db_conn.cursor() as temp_db_cursor:
 
     with temp_db_conn.cursor() as temp_db_cursor:

-        temp_db_cursor.execute(f"""
+        temp_db_cursor.execute("""

                 SELECT * FROM information_schema.role_table_grants
                 SELECT * FROM information_schema.role_table_grants

-                WHERE table_name='{table_name}' 
-                AND grantee='{def_config.DATABASE_WEBUSER}' 
-                AND privilege_type='SELECT'""")
-        result = temp_db_cursor.fetchone()
-    assert result
+                WHERE table_name='{}'
+                AND grantee='{}'
+                AND privilege_type='SELECT'""".format(table_name, user))
+        return temp_db_cursor.fetchone()

 
 

-@pytest.fixture
-def amenity_operator_funcs(temp_db_cursor):                        
-    temp_db_cursor.execute(f"""
-        CREATE OR REPLACE FUNCTION make_standard_name(name TEXT) RETURNS TEXT
-        AS $$
-        DECLARE
-        o TEXT;
-        BEGIN
-        RETURN name; --Basically return the same name for the tests
-        END;
-        $$
-        LANGUAGE plpgsql IMMUTABLE;
-
-        CREATE SEQUENCE seq_word start 1;
-
-        CREATE OR REPLACE FUNCTION getorcreate_amenity(lookup_word TEXT,
-                                                    lookup_class text, lookup_type text)
-        RETURNS INTEGER
-        AS $$
-        DECLARE
-        lookup_token TEXT;
-        return_word_id INTEGER;
-        BEGIN
-        lookup_token := ' '||trim(lookup_word);
-        SELECT min(word_id) FROM word
-        WHERE word_token = lookup_token and word = lookup_word
-                and class = lookup_class and type = lookup_type
-        INTO return_word_id;
-        IF return_word_id IS NULL THEN
-            return_word_id := nextval('seq_word');
-            INSERT INTO word VALUES (return_word_id, lookup_token, lookup_word,
-                                    lookup_class, lookup_type, null, 0);
-        END IF;
-        RETURN return_word_id;
-        END;
-        $$
-        LANGUAGE plpgsql;
-        
-        CREATE OR REPLACE FUNCTION getorcreate_amenityoperator(lookup_word TEXT,
-                                                        lookup_class text,
-                                                        lookup_type text,
-                                                        op text)
-        RETURNS INTEGER
-        AS $$
-        DECLARE
-        lookup_token TEXT;
-        return_word_id INTEGER;
-        BEGIN
-        lookup_token := ' '||trim(lookup_word);
-        SELECT min(word_id) FROM word
-        WHERE word_token = lookup_token and word = lookup_word
-                and class = lookup_class and type = lookup_type and operator = op
-        INTO return_word_id;
-        IF return_word_id IS NULL THEN
-            return_word_id := nextval('seq_word');
-            INSERT INTO word VALUES (return_word_id, lookup_token, lookup_word,
-                                    lookup_class, lookup_type, null, 0, op);
-        END IF;
-        RETURN return_word_id;
-        END;
-        $$
-        LANGUAGE plpgsql;""")
+def check_placeid_and_centroid_indexes(temp_db_conn, phrase_class, phrase_type):
+    """
+        Check that the place_id index and centroid index exist for the
+        place_classtype table of the given phrase_class and phrase_type.
+    """
+    index_prefix = 'idx_place_classtype_{}_{}_'.format(phrase_class, phrase_type)
+
+    return (
+        temp_db_conn.index_exists(index_prefix + 'centroid')
+        and
+        temp_db_conn.index_exists(index_prefix + 'place_id')
+    )