US TIGER data 2021 released

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

diff --git a/test/python/test_cli.py b/test/python/test_cli.py
index 38bbaefee82688bbad8194ee5623884222af6b03..c83ee3ddfba64e5fc880886275ed866cc2259e38 100644 (file)
--- a/test/python/test_cli.py
+++ b/test/python/test_cli.py
@@ -5,8 +5,6 @@ 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.
 """
 correct functionionality. They use a lot of monkeypatching to avoid executing
 the actual functions.
 """

-from pathlib import Path
-

 import pytest
 
 import nominatim.db.properties
 import pytest
 
 import nominatim.db.properties


@@ -17,27 +15,17 @@ import nominatim.clicmd.admin
 import nominatim.clicmd.setup
 import nominatim.indexer.indexer
 import nominatim.tools.admin
 import nominatim.clicmd.setup
 import nominatim.indexer.indexer
 import nominatim.tools.admin

+import nominatim.tools.add_osm_data

 import nominatim.tools.check_database
 import nominatim.tools.database_import
 import nominatim.tools.check_database
 import nominatim.tools.database_import

+import nominatim.tools.country_info

 import nominatim.tools.freeze
 import nominatim.tools.refresh
 import nominatim.tools.postcodes
 import nominatim.tools.freeze
 import nominatim.tools.refresh
 import nominatim.tools.postcodes

+import nominatim.tokenizer.factory

 
 from mocks import MockParamCapture
 
 
 from mocks import MockParamCapture
 

-SRC_DIR = (Path(__file__) / '..' / '..' / '..').resolve()
-
-def call_nominatim(*args):
-    return nominatim.cli.nominatim(module_dir='build/module',
-                                   osm2pgsql_path='build/osm2pgsql/osm2pgsql',
-                                   phplib_dir=str(SRC_DIR / 'lib-php'),
-                                   data_dir=str(SRC_DIR / 'data'),
-                                   phpcgi_path='/usr/bin/php-cgi',
-                                   sqllib_dir=str(SRC_DIR / 'lib-sql'),
-                                   config_dir=str(SRC_DIR / 'settings'),
-                                   cli_args=args)
-
-

 @pytest.fixture
 def mock_run_legacy(monkeypatch):
     mock = MockParamCapture()
 @pytest.fixture
 def mock_run_legacy(monkeypatch):
     mock = MockParamCapture()


@@ -49,185 +37,321 @@ def mock_run_legacy(monkeypatch):
 def mock_func_factory(monkeypatch):
     def get_mock(module, func):
         mock = MockParamCapture()
 def mock_func_factory(monkeypatch):
     def get_mock(module, func):
         mock = MockParamCapture()

+        mock.func_name = func

         monkeypatch.setattr(module, func, mock)
         return mock
 
     return get_mock
 
 
         monkeypatch.setattr(module, func, mock)
         return mock
 
     return get_mock
 
 

-def test_cli_help(capsys):
-    """ Running nominatim tool without arguments prints help.
-    """
-    assert 1 == call_nominatim()

 
 

-    captured = capsys.readouterr()
-    assert captured.out.startswith('usage:')
+class TestCli:
+
+    @pytest.fixture(autouse=True)
+    def setup_cli_call(self, cli_call):
+        self.call_nominatim = cli_call

 
 
 
 

-@pytest.mark.parametrize("command,script", [
-                         (('add-data', '--file', 'foo.osm'), 'update'),
-                         (('export',), 'export')
-                         ])
-def test_legacy_commands_simple(mock_run_legacy, command, script):
-    assert 0 == call_nominatim(*command)
+    def test_cli_help(self, capsys):
+        """ Running nominatim tool without arguments prints help.
+        """
+        assert self.call_nominatim() == 1

 
 

-    assert mock_run_legacy.called == 1
-    assert mock_run_legacy.last_args[0] == script + '.php'
+        captured = capsys.readouterr()
+        assert captured.out.startswith('usage:')

 
 
 
 

-def test_import_missing_file(temp_db):
-    assert 1 == call_nominatim('import', '--osm-file', 'sfsafegweweggdgw.reh.erh')
+    @pytest.mark.parametrize("command,script", [
+                             (('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'

 
 

-def test_import_bad_file(temp_db):
-    assert 1 == call_nominatim('import', '--osm-file', '.')

 
 

+    @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')

 
 

-def test_import_full(temp_db, mock_func_factory):
-    mocks = [
-        mock_func_factory(nominatim.tools.database_import, 'setup_database_skeleton'),
-        mock_func_factory(nominatim.tools.database_import, 'install_module'),
-        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, 'import_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 self.call_nominatim('admin', *params) == 0

 
 

-    cf_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')
+        assert mock_run_legacy.called == 1

 
 

-    assert 0 == call_nominatim('import', '--osm-file', __file__)

 
 

-    assert cf_mock.called > 1
+    def test_admin_command_check_database(self, mock_func_factory):
+        mock = mock_func_factory(nominatim.tools.check_database, 'check_database')

 
 

-    for mock in mocks:
+        assert self.call_nominatim('admin', '--check-database') == 0

         assert mock.called == 1
 
         assert mock.called == 1
 

-def test_freeze_command(mock_func_factory, temp_db):
-    mock_drop = mock_func_factory(nominatim.tools.freeze, 'drop_update_tables')
-    mock_flatnode = mock_func_factory(nominatim.tools.freeze, 'drop_flatnode_file')

 
 

-    assert 0 == call_nominatim('freeze')
+    @pytest.mark.parametrize("name,oid", [('file', 'foo.osm'), ('diff', 'foo.osc')])
+    def test_add_data_file_command(self, mock_func_factory, name, oid):
+        mock_run_legacy = mock_func_factory(nominatim.tools.add_osm_data, 'add_data_from_file')
+        assert self.call_nominatim('add-data', '--' + name, str(oid)) == 0
+
+        assert mock_run_legacy.called == 1
+
+
+    @pytest.mark.parametrize("name,oid", [('node', 12), ('way', 8), ('relation', 32)])
+    def test_add_data_object_command(self, mock_func_factory, name, oid):
+        mock_run_legacy = mock_func_factory(nominatim.tools.add_osm_data, 'add_osm_object')
+        assert self.call_nominatim('add-data', '--' + name, str(oid)) == 0
+
+        assert mock_run_legacy.called == 1
+
+
+    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

 
 

-    assert mock_drop.called == 1
-    assert mock_flatnode.called == 1
+            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)

 
 

-@pytest.mark.parametrize("params", [('--warm', ),
-                                    ('--warm', '--reverse-only'),
-                                    ('--warm', '--search-only')])
-def test_admin_command_legacy(mock_func_factory, params):
-    mock_run_legacy = mock_func_factory(nominatim.clicmd.admin, 'run_legacy_script')
+        self.tokenizer_mock = tok

 
 

-    assert 0 == call_nominatim('admin', *params)

 
 

-    assert mock_run_legacy.called == 1
+    def test_import_missing_file(self):
+        assert self.call_nominatim('import', '--osm-file', 'sfsafegwedgw.reh.erh') == 1

 
 
 
 

-@pytest.mark.parametrize("func, params", [('analyse_indexing', ('--analyse-indexing', ))])
-def test_admin_command_tool(temp_db, mock_func_factory, func, params):
-    mock = mock_func_factory(nominatim.tools.admin, func)
+    def test_import_bad_file(self):
+        assert self.call_nominatim('import', '--osm-file', '.') == 1

 
 

-    assert 0 == call_nominatim('admin', *params)
-    assert mock.called == 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.country_info, 'setup_country_tables'),
+            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.country_info, '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')
+        ]

 
 

-def test_admin_command_check_database(mock_func_factory):
-    mock = mock_func_factory(nominatim.tools.check_database, 'check_database')
+        cf_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')

 
 

-    assert 0 == call_nominatim('admin', '--check-database')
-    assert mock.called == 1
+        assert self.call_nominatim('import', '--osm-file', __file__) == 0
+        assert self.tokenizer_mock.finalize_import_called

 
 

+        assert cf_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(mock_run_legacy, name, oid):
-    assert 0 == call_nominatim('add-data', '--' + name, str(oid))
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+
+    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.country_info, '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 self.call_nominatim('import', '--continue', 'load-data') == 0
+        assert self.tokenizer_mock.finalize_import_called
+
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+
+    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.country_info, '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 self.call_nominatim('import', '--continue', 'indexing') == 0
+
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+        assert temp_db_conn.index_exists('idx_placex_pendingsector')
+
+        # 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_import_continue_postprocess(self, mock_func_factory):
+        mocks = [
+            mock_func_factory(nominatim.tools.database_import, 'create_search_indices'),
+            mock_func_factory(nominatim.tools.country_info, 'create_country_names'),
+            mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+            mock_func_factory(nominatim.db.properties, 'set_property')
+        ]
+
+        assert self.call_nominatim('import', '--continue', 'db-postprocess') == 0
+
+        assert self.tokenizer_mock.finalize_import_called
+
+        for mock in mocks:
+            assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+
+    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
+
+        assert mock_drop.called == 1
+        assert mock_flatnode.called == 1
+
+
+
+    @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 self.call_nominatim('admin', *params) == 0
+        assert mock.called == 1

 
 

-    assert mock_run_legacy.called == 1
-    assert mock_run_legacy.last_args == ('update.php', '--import-' + name, oid)

 
 

+    @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')

 
 

-@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(mock_func_factory, temp_db_cursor, params, do_bnds, do_ranks):
-    temp_db_cursor.execute("CREATE TABLE 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 self.call_nominatim('index', *params) == 0

 
 

-    assert 0 == call_nominatim('index', *params)
+        assert bnd_mock.called == do_bnds
+        assert rank_mock.called == do_ranks

 
 

-    assert bnd_mock.called == do_bnds
-    assert rank_mock.called == do_ranks
+    @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')

 
 

-def test_special_phrases_command(temp_db, mock_func_factory):
-    func = mock_func_factory(nominatim.clicmd.special_phrases.SpecialPhrasesImporter, 'import_from_wiki')
+        if no_replace:
+            self.call_nominatim('special-phrases', '--import-from-wiki', '--no-replace')
+        else:
+            self.call_nominatim('special-phrases', '--import-from-wiki')

 
 

-    call_nominatim('special-phrases', '--import-from-wiki')
+        assert func.called == 1

 
 

-    assert func.called == 1
+    @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())

 
 

-@pytest.mark.parametrize("command,func", [
-                         ('postcodes', 'update_postcodes'),
-                         ('word-counts', 'recompute_word_counts'),
-                         ('address-levels', 'load_address_levels_from_file'),
-                         ('functions', 'create_functions'),
-                         ('wiki-data', 'import_wikipedia_articles'),
-                         ('importance', 'recompute_importance'),
-                         ('website', 'setup_website'),
-                         ])
-def test_refresh_command(mock_func_factory, temp_db, command, func):
-    func_mock = mock_func_factory(nominatim.tools.refresh, func)
+        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)

 
 

-    assert 0 == call_nominatim('refresh', '--' + command)
-    assert func_mock.called == 1
+        assert func.called == 1

 
 

+    @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)

 
 

-def test_refresh_importance_computed_after_wiki_import(monkeypatch, temp_db):
-    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 self.call_nominatim('refresh', '--' + command) == 0
+        assert func_mock.called == 1

 
 

-    assert 0 == call_nominatim('refresh', '--importance', '--wiki-data')

 
 

-    assert calls == ['import', 'update']
+    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(mock_func_factory):
-    func = mock_func_factory(nominatim.cli, 'run_php_server')
+    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(mock_func_factory, params):
-    mock_run_api = mock_func_factory(nominatim.clicmd.api, 'run_api_script')
+    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']