]> git.openstreetmap.org Git - nominatim.git/blobdiff - test/python/test_cli.py
make sure old data gets deleted on place type change
[nominatim.git] / test / python / test_cli.py
index 334a9e78985773e174b1be9decb4faf37347d1e4..d9e0104014eb74c529cfa6d198847ce415460fa6 100644 (file)
@@ -5,42 +5,24 @@ These tests just check that the various command line parameters route to the
 correct functionionality. They use a lot of monkeypatching to avoid executing
 the actual functions.
 """
-import datetime as dt
-import psycopg2
 import pytest
-import time
 
+import nominatim.db.properties
 import nominatim.cli
 import nominatim.clicmd.api
 import nominatim.clicmd.refresh
 import nominatim.clicmd.admin
+import nominatim.clicmd.setup
 import nominatim.indexer.indexer
+import nominatim.tools.admin
+import nominatim.tools.check_database
+import nominatim.tools.database_import
+import nominatim.tools.freeze
 import nominatim.tools.refresh
-import nominatim.tools.replication
-from nominatim.errors import UsageError
-from nominatim.db import status
-
-def call_nominatim(*args):
-    return nominatim.cli.nominatim(module_dir='build/module',
-                                   osm2pgsql_path='build/osm2pgsql/osm2pgsql',
-                                   phplib_dir='lib-php',
-                                   data_dir='.',
-                                   phpcgi_path='/usr/bin/php-cgi',
-                                   cli_args=args)
-
-class MockParamCapture:
-    """ Mock that records the parameters with which a function was called
-        as well as the number of calls.
-    """
-    def __init__(self, retval=0):
-        self.called = 0
-        self.return_value = retval
-
-    def __call__(self, *args, **kwargs):
-        self.called += 1
-        self.last_args = args
-        self.last_kwargs = kwargs
-        return self.return_value
+import nominatim.tools.postcodes
+import nominatim.tokenizer.factory
+
+from mocks import MockParamCapture
 
 @pytest.fixture
 def mock_run_legacy(monkeypatch):
@@ -49,216 +31,318 @@ def mock_run_legacy(monkeypatch):
     return mock
 
 
-def test_cli_help(capsys):
-    """ Running nominatim tool without arguments prints help.
-    """
-    assert 1 == call_nominatim()
+@pytest.fixture
+def mock_func_factory(monkeypatch):
+    def get_mock(module, func):
+        mock = MockParamCapture()
+        mock.func_name = func
+        monkeypatch.setattr(module, func, mock)
+        return mock
 
-    captured = capsys.readouterr()
-    assert captured.out.startswith('usage:')
+    return get_mock
 
 
-@pytest.mark.parametrize("command,script", [
-                         (('import', '--continue', 'load-data'), 'setup'),
-                         (('freeze',), 'setup'),
-                         (('special-phrases',), 'specialphrases'),
-                         (('add-data', '--tiger-data', 'tiger'), 'setup'),
-                         (('add-data', '--file', 'foo.osm'), 'update'),
-                         (('export',), 'export')
-                         ])
-def test_legacy_commands_simple(mock_run_legacy, command, script):
-    assert 0 == call_nominatim(*command)
 
-    assert mock_run_legacy.called == 1
-    assert mock_run_legacy.last_args[0] == script + '.php'
+class TestCli:
 
+    @pytest.fixture(autouse=True)
+    def setup_cli_call(self, cli_call):
+        self.call_nominatim = cli_call
 
-@pytest.mark.parametrize("params", [('--warm', ),
-                                    ('--warm', '--reverse-only'),
-                                    ('--warm', '--search-only'),
-                                    ('--check-database', )])
-def test_admin_command_legacy(monkeypatch, params):
-    mock_run_legacy = MockParamCapture()
-    monkeypatch.setattr(nominatim.clicmd.admin, 'run_legacy_script', mock_run_legacy)
 
-    assert 0 == call_nominatim('admin', *params)
+    def test_cli_help(self, capsys):
+        """ Running nominatim tool without arguments prints help.
+        """
+        assert self.call_nominatim() == 1
 
-    assert mock_run_legacy.called == 1
+        captured = capsys.readouterr()
+        assert captured.out.startswith('usage:')
 
-@pytest.mark.parametrize("func, params", [('analyse_indexing', ('--analyse-indexing', ))])
-def test_admin_command_tool(monkeypatch, func, params):
-    mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.tools.admin, func, mock)
 
-    assert 0 == call_nominatim('admin', *params)
-    assert mock.called == 1
+    @pytest.mark.parametrize("command,script", [
+                             (('add-data', '--file', 'foo.osm'), 'update'),
+                             (('export',), 'export')
+                             ])
+    def test_legacy_commands_simple(self, mock_run_legacy, command, script):
+        assert self.call_nominatim(*command) == 0
+
+        assert mock_run_legacy.called == 1
+        assert mock_run_legacy.last_args[0] == script + '.php'
+
+
+    @pytest.mark.parametrize("params", [('--warm', ),
+                                        ('--warm', '--reverse-only'),
+                                        ('--warm', '--search-only')])
+    def test_admin_command_legacy(self, mock_func_factory, params):
+        mock_run_legacy = mock_func_factory(nominatim.clicmd.admin, 'run_legacy_script')
+
+        assert self.call_nominatim('admin', *params) == 0
+
+        assert mock_run_legacy.called == 1
+
+
+    def test_admin_command_check_database(self, mock_func_factory):
+        mock = mock_func_factory(nominatim.tools.check_database, 'check_database')
+
+        assert self.call_nominatim('admin', '--check-database') == 0
+        assert mock.called == 1
+
+
+    @pytest.mark.parametrize("name,oid", [('file', 'foo.osm'), ('diff', 'foo.osc'),
+                                          ('node', 12), ('way', 8), ('relation', 32)])
+    def test_add_data_command(self, mock_run_legacy, name, oid):
+        assert self.call_nominatim('add-data', '--' + name, str(oid)) == 0
+
+        assert mock_run_legacy.called == 1
+        assert mock_run_legacy.last_args == ('update.php', '--import-' + name, oid)
+
+
+    def test_serve_command(self, mock_func_factory):
+        func = mock_func_factory(nominatim.cli, 'run_php_server')
+
+        self.call_nominatim('serve')
+
+        assert func.called == 1
+
+
+    @pytest.mark.parametrize("params", [('search', '--query', 'new'),
+                                        ('reverse', '--lat', '0', '--lon', '0'),
+                                        ('lookup', '--id', 'N1'),
+                                        ('details', '--node', '1'),
+                                        ('details', '--way', '1'),
+                                        ('details', '--relation', '1'),
+                                        ('details', '--place_id', '10001'),
+                                        ('status',)])
+    def test_api_commands_simple(self, mock_func_factory, params):
+        mock_run_api = mock_func_factory(nominatim.clicmd.api, 'run_api_script')
+
+        assert self.call_nominatim(*params) == 0
+
+        assert mock_run_api.called == 1
+        assert mock_run_api.last_args[0] == params[0]
+
+
+
+class TestCliWithDb:
+
+    @pytest.fixture(autouse=True)
+    def setup_cli_call(self, cli_call, temp_db):
+        self.call_nominatim = cli_call
+
+
+    @pytest.fixture(autouse=True)
+    def setup_tokenizer_mock(self, monkeypatch):
+        class DummyTokenizer:
+            def __init__(self, *args, **kwargs):
+                self.update_sql_functions_called = False
+                self.finalize_import_called = False
+
+            def update_sql_functions(self, *args):
+                self.update_sql_functions_called = True
+
+            def finalize_import(self, *args):
+                self.finalize_import_called = True
+
+        tok = DummyTokenizer()
+        monkeypatch.setattr(nominatim.tokenizer.factory, 'get_tokenizer_for_db',
+                            lambda *args: tok)
+        monkeypatch.setattr(nominatim.tokenizer.factory, 'create_tokenizer',
+                            lambda *args: tok)
+
+        self.tokenizer_mock = tok
+
+
+    def test_import_missing_file(self):
+        assert self.call_nominatim('import', '--osm-file', 'sfsafegwedgw.reh.erh') == 1
+
+
+    def test_import_bad_file(self):
+        assert self.call_nominatim('import', '--osm-file', '.') == 1
+
+
+    def test_import_full(self, mock_func_factory):
+        mocks = [
+            mock_func_factory(nominatim.tools.database_import, 'setup_database_skeleton'),
+            mock_func_factory(nominatim.tools.database_import, 'import_osm_data'),
+            mock_func_factory(nominatim.tools.refresh, 'import_wikipedia_articles'),
+            mock_func_factory(nominatim.tools.database_import, 'truncate_data_tables'),
+            mock_func_factory(nominatim.tools.database_import, 'load_data'),
+            mock_func_factory(nominatim.tools.database_import, 'create_tables'),
+            mock_func_factory(nominatim.tools.database_import, 'create_table_triggers'),
+            mock_func_factory(nominatim.tools.database_import, 'create_partition_tables'),
+            mock_func_factory(nominatim.tools.database_import, 'create_search_indices'),
+            mock_func_factory(nominatim.tools.database_import, 'create_country_names'),
+            mock_func_factory(nominatim.tools.refresh, 'load_address_levels_from_file'),
+            mock_func_factory(nominatim.tools.postcodes, 'update_postcodes'),
+            mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_full'),
+            mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+            mock_func_factory(nominatim.db.properties, 'set_property')
+        ]
+
+        cf_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')
+
+        assert self.call_nominatim('import', '--osm-file', __file__) == 0
+        assert self.tokenizer_mock.finalize_import_called
 
-@pytest.mark.parametrize("name,oid", [('file', 'foo.osm'), ('diff', 'foo.osc'),
-                                      ('node', 12), ('way', 8), ('relation', 32)])
-def test_add_data_command(mock_run_legacy, name, oid):
-    assert 0 == call_nominatim('add-data', '--' + name, str(oid))
+        assert cf_mock.called > 1
 
-    assert mock_run_legacy.called == 1
-    assert mock_run_legacy.last_args == ('update.php', '--import-' + name, oid)
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
 
 
-@pytest.mark.parametrize("params,do_bnds,do_ranks", [
-                          ([], 1, 1),
-                          (['--boundaries-only'], 1, 0),
-                          (['--no-boundaries'], 0, 1),
-                          (['--boundaries-only', '--no-boundaries'], 0, 0)])
-def test_index_command(monkeypatch, temp_db_cursor, params, do_bnds, do_ranks):
-    temp_db_cursor.execute("CREATE TABLE import_status (indexed bool)")
-    bnd_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_boundaries', bnd_mock)
-    rank_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_by_rank', rank_mock)
+    def test_import_continue_load_data(self, mock_func_factory):
+        mocks = [
+            mock_func_factory(nominatim.tools.database_import, 'truncate_data_tables'),
+            mock_func_factory(nominatim.tools.database_import, 'load_data'),
+            mock_func_factory(nominatim.tools.database_import, 'create_search_indices'),
+            mock_func_factory(nominatim.tools.database_import, 'create_country_names'),
+            mock_func_factory(nominatim.tools.postcodes, 'update_postcodes'),
+            mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_full'),
+            mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+            mock_func_factory(nominatim.db.properties, 'set_property')
+        ]
 
-    assert 0 == call_nominatim('index', *params)
+        assert self.call_nominatim('import', '--continue', 'load-data') == 0
+        assert self.tokenizer_mock.finalize_import_called
 
-    assert bnd_mock.called == do_bnds
-    assert rank_mock.called == do_ranks
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
 
 
-@pytest.mark.parametrize("command,params", [
-                         ('wiki-data', ('setup.php', '--import-wikipedia-articles')),
-                         ('importance', ('update.php', '--recompute-importance')),
-                         ('website', ('setup.php', '--setup-website')),
-                         ])
-def test_refresh_legacy_command(monkeypatch, temp_db, command, params):
-    mock_run_legacy = MockParamCapture()
-    monkeypatch.setattr(nominatim.clicmd.refresh, 'run_legacy_script', mock_run_legacy)
+    def test_import_continue_indexing(self, mock_func_factory, placex_table,
+                                      temp_db_conn):
+        mocks = [
+            mock_func_factory(nominatim.tools.database_import, 'create_search_indices'),
+            mock_func_factory(nominatim.tools.database_import, 'create_country_names'),
+            mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_full'),
+            mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+            mock_func_factory(nominatim.db.properties, 'set_property')
+        ]
 
-    assert 0 == call_nominatim('refresh', '--' + command)
+        assert self.call_nominatim('import', '--continue', 'indexing') == 0
 
-    assert mock_run_legacy.called == 1
-    assert len(mock_run_legacy.last_args) >= len(params)
-    assert mock_run_legacy.last_args[:len(params)] == params
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
 
-@pytest.mark.parametrize("command,func", [
-                         ('postcodes', 'update_postcodes'),
-                         ('word-counts', 'recompute_word_counts'),
-                         ('address-levels', 'load_address_levels_from_file'),
-                         ('functions', 'create_functions'),
-                         ])
-def test_refresh_command(monkeypatch, temp_db, command, func):
-    func_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.tools.refresh, func, func_mock)
+        assert temp_db_conn.index_exists('idx_placex_pendingsector')
 
-    assert 0 == call_nominatim('refresh', '--' + command)
-    assert func_mock.called == 1
+        # Calling it again still works for the index
+        assert self.call_nominatim('import', '--continue', 'indexing') == 0
+        assert temp_db_conn.index_exists('idx_placex_pendingsector')
 
 
-def test_refresh_importance_computed_after_wiki_import(monkeypatch, temp_db):
-    mock_run_legacy = MockParamCapture()
-    monkeypatch.setattr(nominatim.clicmd.refresh, 'run_legacy_script', mock_run_legacy)
+    def test_import_continue_postprocess(self, mock_func_factory):
+        mocks = [
+            mock_func_factory(nominatim.tools.database_import, 'create_search_indices'),
+            mock_func_factory(nominatim.tools.database_import, 'create_country_names'),
+            mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+            mock_func_factory(nominatim.db.properties, 'set_property')
+        ]
 
-    assert 0 == call_nominatim('refresh', '--importance', '--wiki-data')
+        assert self.call_nominatim('import', '--continue', 'db-postprocess') == 0
 
-    assert mock_run_legacy.called == 2
-    assert mock_run_legacy.last_args == ('update.php', '--recompute-importance')
+        assert self.tokenizer_mock.finalize_import_called
 
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
 
-@pytest.mark.parametrize("params,func", [
-                         (('--init', '--no-update-functions'), 'init_replication'),
-                         (('--check-for-updates',), 'check_for_updates')
-                         ])
-def test_replication_command(monkeypatch, temp_db, params, func):
-    func_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.tools.replication, func, func_mock)
 
-    assert 0 == call_nominatim('replication', *params)
-    assert func_mock.called == 1
+    def test_freeze_command(self, mock_func_factory):
+        mock_drop = mock_func_factory(nominatim.tools.freeze, 'drop_update_tables')
+        mock_flatnode = mock_func_factory(nominatim.tools.freeze, 'drop_flatnode_file')
 
+        assert self.call_nominatim('freeze') == 0
 
-def test_replication_update_bad_interval(monkeypatch, temp_db):
-    monkeypatch.setenv('NOMINATIM_REPLICATION_UPDATE_INTERVAL', 'xx')
+        assert mock_drop.called == 1
+        assert mock_flatnode.called == 1
 
-    assert call_nominatim('replication') == 1
 
 
-def test_replication_update_bad_interval_for_geofabrik(monkeypatch, temp_db):
-    monkeypatch.setenv('NOMINATIM_REPLICATION_URL',
-                       'https://download.geofabrik.de/europe/ireland-and-northern-ireland-updates')
+    @pytest.mark.parametrize("func, params", [('analyse_indexing', ('--analyse-indexing', ))])
+    def test_admin_command_tool(self, mock_func_factory, func, params):
+        mock = mock_func_factory(nominatim.tools.admin, func)
 
-    assert call_nominatim('replication') == 1
+        assert self.call_nominatim('admin', *params) == 0
+        assert mock.called == 1
 
 
-@pytest.mark.parametrize("state, retval", [
-                         (nominatim.tools.replication.UpdateState.UP_TO_DATE, 0),
-                         (nominatim.tools.replication.UpdateState.NO_CHANGES, 3)
-                         ])
-def test_replication_update_once_no_index(monkeypatch, temp_db, temp_db_conn,
-                                          status_table, state, retval):
-    status.set_status(temp_db_conn, date=dt.datetime.now(dt.timezone.utc), seq=1)
-    func_mock = MockParamCapture(retval=state)
-    monkeypatch.setattr(nominatim.tools.replication, 'update', func_mock)
+    @pytest.mark.parametrize("params,do_bnds,do_ranks", [
+                              ([], 1, 1),
+                              (['--boundaries-only'], 1, 0),
+                              (['--no-boundaries'], 0, 1),
+                              (['--boundaries-only', '--no-boundaries'], 0, 0)])
+    def test_index_command(self, mock_func_factory, table_factory,
+                           params, do_bnds, do_ranks):
+        table_factory('import_status', 'indexed bool')
+        bnd_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_boundaries')
+        rank_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_by_rank')
 
-    assert retval == call_nominatim('replication', '--once', '--no-index')
+        assert self.call_nominatim('index', *params) == 0
 
+        assert bnd_mock.called == do_bnds
+        assert rank_mock.called == do_ranks
 
-def test_replication_update_continuous(monkeypatch, temp_db_conn, status_table):
-    status.set_status(temp_db_conn, date=dt.datetime.now(dt.timezone.utc), seq=1)
-    states = [nominatim.tools.replication.UpdateState.UP_TO_DATE,
-              nominatim.tools.replication.UpdateState.UP_TO_DATE]
-    monkeypatch.setattr(nominatim.tools.replication, 'update',
-                        lambda *args, **kwargs: states.pop())
+    @pytest.mark.parametrize("no_replace", [(True), (False)])
+    def test_special_phrases_wiki_command(self, mock_func_factory, no_replace):
+        func = mock_func_factory(nominatim.clicmd.special_phrases.SPImporter, 'import_phrases')
 
-    index_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_boundaries', index_mock)
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_by_rank', index_mock)
+        if no_replace:
+            self.call_nominatim('special-phrases', '--import-from-wiki', '--no-replace')
+        else:
+            self.call_nominatim('special-phrases', '--import-from-wiki')
 
-    with pytest.raises(IndexError):
-        call_nominatim('replication')
+        assert func.called == 1
 
-    assert index_mock.called == 4
+    @pytest.mark.parametrize("no_replace", [(True), (False)])
+    def test_special_phrases_csv_command(self, src_dir, mock_func_factory, no_replace):
+        func = mock_func_factory(nominatim.clicmd.special_phrases.SPImporter, 'import_phrases')
+        testdata = src_dir / 'test' / 'testdb'
+        csv_path = str((testdata / 'full_en_phrases_test.csv').resolve())
 
+        if no_replace:
+            self.call_nominatim('special-phrases', '--import-from-csv', csv_path, '--no-replace')
+        else:
+            self.call_nominatim('special-phrases', '--import-from-csv', csv_path)
 
-def test_replication_update_continuous_no_change(monkeypatch, temp_db_conn, status_table):
-    status.set_status(temp_db_conn, date=dt.datetime.now(dt.timezone.utc), seq=1)
-    states = [nominatim.tools.replication.UpdateState.NO_CHANGES,
-              nominatim.tools.replication.UpdateState.UP_TO_DATE]
-    monkeypatch.setattr(nominatim.tools.replication, 'update',
-                        lambda *args, **kwargs: states.pop())
+        assert func.called == 1
 
-    index_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_boundaries', index_mock)
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_by_rank', index_mock)
+    @pytest.mark.parametrize("command,func", [
+                             ('word-counts', 'recompute_word_counts'),
+                             ('address-levels', 'load_address_levels_from_file'),
+                             ('wiki-data', 'import_wikipedia_articles'),
+                             ('importance', 'recompute_importance'),
+                             ('website', 'setup_website'),
+                             ])
+    def test_refresh_command(self, mock_func_factory, command, func):
+        func_mock = mock_func_factory(nominatim.tools.refresh, func)
 
-    sleep_mock = MockParamCapture()
-    monkeypatch.setattr(time, 'sleep', sleep_mock)
+        assert self.call_nominatim('refresh', '--' + command) == 0
+        assert func_mock.called == 1
 
-    with pytest.raises(IndexError):
-        call_nominatim('replication')
 
-    assert index_mock.called == 2
-    assert sleep_mock.called == 1
-    assert sleep_mock.last_args[0] == 60
+    def test_refresh_postcodes(self, mock_func_factory, place_table):
+        func_mock = mock_func_factory(nominatim.tools.postcodes, 'update_postcodes')
+        idx_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_postcodes')
 
+        assert self.call_nominatim('refresh', '--postcodes') == 0
+        assert func_mock.called == 1
+        assert idx_mock.called == 1
 
-def test_serve_command(monkeypatch):
-    func = MockParamCapture()
-    monkeypatch.setattr(nominatim.cli, 'run_php_server', func)
+    def test_refresh_create_functions(self, mock_func_factory):
+        func_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')
 
-    call_nominatim('serve')
+        assert self.call_nominatim('refresh', '--functions') == 0
+        assert func_mock.called == 1
+        assert self.tokenizer_mock.update_sql_functions_called
 
-    assert func.called == 1
 
-@pytest.mark.parametrize("params", [
-                         ('search', '--query', 'new'),
-                         ('reverse', '--lat', '0', '--lon', '0'),
-                         ('lookup', '--id', 'N1'),
-                         ('details', '--node', '1'),
-                         ('details', '--way', '1'),
-                         ('details', '--relation', '1'),
-                         ('details', '--place_id', '10001'),
-                         ('status',)
-                         ])
-def test_api_commands_simple(monkeypatch, params):
-    mock_run_api = MockParamCapture()
-    monkeypatch.setattr(nominatim.clicmd.api, 'run_api_script', mock_run_api)
+    def test_refresh_importance_computed_after_wiki_import(self, monkeypatch):
+        calls = []
+        monkeypatch.setattr(nominatim.tools.refresh, 'import_wikipedia_articles',
+                            lambda *args, **kwargs: calls.append('import') or 0)
+        monkeypatch.setattr(nominatim.tools.refresh, 'recompute_importance',
+                            lambda *args, **kwargs: calls.append('update'))
 
-    assert 0 == call_nominatim(*params)
+        assert self.call_nominatim('refresh', '--importance', '--wiki-data') == 0
 
-    assert mock_run_api.called == 1
-    assert mock_run_api.last_args[0] == params[0]
+        assert calls == ['import', 'update']