move index creation for word table to tokenizer

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

diff --git a/test/python/test_cli.py b/test/python/test_cli.py
index a8efd35670e63e0cc2744a2da097bd7db13f55f3..a286995611ddf524c54ae5c1657b3e3e15bf55c2 100644 (file)
--- a/test/python/test_cli.py
+++ b/test/python/test_cli.py
@@ -5,44 +5,39 @@ 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.
 """

-import datetime as dt
-import psycopg2
+from pathlib import Path
+

 import pytest
 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.cli
 import nominatim.clicmd.api
 import nominatim.clicmd.refresh
 import nominatim.clicmd.admin

+import nominatim.clicmd.setup

 import nominatim.indexer.indexer
 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.refresh

-import nominatim.tools.replication
-from nominatim.errors import UsageError
-from nominatim.db import status
+import nominatim.tools.postcodes
+import nominatim.tokenizer.factory
+
+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',
 
 def call_nominatim(*args):
     return nominatim.cli.nominatim(module_dir='build/module',
                                    osm2pgsql_path='build/osm2pgsql/osm2pgsql',

-                                   phplib_dir='lib-php',
-                                   data_dir='.',
+                                   phplib_dir=str(SRC_DIR / 'lib-php'),
+                                   data_dir=str(SRC_DIR / 'data'),

                                    phpcgi_path='/usr/bin/php-cgi',
                                    phpcgi_path='/usr/bin/php-cgi',

-                                   sqllib_dir='lib-sql',
-                                   config_dir='settings',
+                                   sqllib_dir=str(SRC_DIR / 'lib-sql'),
+                                   config_dir=str(SRC_DIR / 'settings'),

                                    cli_args=args)
 
                                    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

 
 @pytest.fixture
 def mock_run_legacy(monkeypatch):
 
 @pytest.fixture
 def mock_run_legacy(monkeypatch):


@@ -51,6 +46,39 @@ def mock_run_legacy(monkeypatch):
     return mock
 
 
     return mock
 
 

+@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
+
+    return get_mock
+
+
+@pytest.fixture
+def tokenizer_mock(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)
+
+    return tok
+
+

 def test_cli_help(capsys):
     """ Running nominatim tool without arguments prints help.
     """
 def test_cli_help(capsys):
     """ Running nominatim tool without arguments prints help.
     """


@@ -61,10 +89,6 @@ def test_cli_help(capsys):
 
 
 @pytest.mark.parametrize("command,script", [
 
 
 @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')
                          ])
                          (('add-data', '--file', 'foo.osm'), 'update'),
                          (('export',), 'export')
                          ])


@@ -75,26 +99,137 @@ def test_legacy_commands_simple(mock_run_legacy, command, script):
     assert mock_run_legacy.last_args[0] == script + '.php'
 
 
     assert mock_run_legacy.last_args[0] == script + '.php'
 
 

+def test_import_missing_file(temp_db):
+    assert 1 == call_nominatim('import', '--osm-file', 'sfsafegweweggdgw.reh.erh')
+
+
+def test_import_bad_file(temp_db):
+    assert 1 == call_nominatim('import', '--osm-file', '.')
+
+
+def test_import_full(temp_db, mock_func_factory, tokenizer_mock):
+    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, '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')
+    ]
+
+    cf_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')
+
+    assert 0 == call_nominatim('import', '--osm-file', __file__)
+    assert tokenizer_mock.finalize_import_called
+
+    assert cf_mock.called > 1
+
+    for mock in mocks:
+        assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+
+def test_import_continue_load_data(temp_db, mock_func_factory, tokenizer_mock):
+    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, '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 0 == call_nominatim('import', '--continue', 'load-data')
+    assert 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(temp_db, mock_func_factory, placex_table,
+                                  temp_db_conn, tokenizer_mock):
+    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('import', '--continue', 'indexing')
+
+    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 0 == call_nominatim('import', '--continue', 'indexing')
+    assert temp_db_conn.index_exists('idx_placex_pendingsector')
+
+
+def test_import_continue_postprocess(temp_db, mock_func_factory, tokenizer_mock):
+    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('import', '--continue', 'db-postprocess')
+
+    assert tokenizer_mock.finalize_import_called
+
+    for mock in mocks:
+        assert mock.called == 1, "Mock '{}' not called".format(mock.func_name)
+
+
+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')
+
+    assert mock_drop.called == 1
+    assert mock_flatnode.called == 1
+
+

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

 
     assert 0 == call_nominatim('admin', *params)
 
     assert mock_run_legacy.called == 1
 
 
     assert 0 == call_nominatim('admin', *params)
 
     assert mock_run_legacy.called == 1
 

+

 @pytest.mark.parametrize("func, params", [('analyse_indexing', ('--analyse-indexing', ))])
 @pytest.mark.parametrize("func, params", [('analyse_indexing', ('--analyse-indexing', ))])

-def test_admin_command_tool(temp_db, monkeypatch, func, params):
-    mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.tools.admin, func, mock)
+def test_admin_command_tool(temp_db, mock_func_factory, func, params):
+    mock = mock_func_factory(nominatim.tools.admin, func)

 
     assert 0 == call_nominatim('admin', *params)
     assert mock.called == 1
 
 
     assert 0 == call_nominatim('admin', *params)
     assert mock.called == 1
 

+
+def test_admin_command_check_database(mock_func_factory):
+    mock = mock_func_factory(nominatim.tools.check_database, 'check_database')
+
+    assert 0 == call_nominatim('admin', '--check-database')
+    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(mock_run_legacy, name, oid):
 @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):


@@ -109,138 +244,61 @@ def test_add_data_command(mock_run_legacy, name, oid):
                           (['--boundaries-only'], 1, 0),
                           (['--no-boundaries'], 0, 1),
                           (['--boundaries-only', '--no-boundaries'], 0, 0)])
                           (['--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):
+def test_index_command(mock_func_factory, temp_db_cursor, tokenizer_mock,
+                       params, do_bnds, do_ranks):

     temp_db_cursor.execute("CREATE TABLE import_status (indexed bool)")
     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)
+    bnd_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_boundaries')
+    rank_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_by_rank')

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

+def test_special_phrases_command(temp_db, mock_func_factory, tokenizer_mock):
+    func = mock_func_factory(nominatim.clicmd.special_phrases.SpecialPhrasesImporter, 'import_from_wiki')

 
 

-@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)
+    call_nominatim('special-phrases', '--import-from-wiki')

 
 

-    assert 0 == call_nominatim('refresh', '--' + command)
-
-    assert mock_run_legacy.called == 1
-    assert len(mock_run_legacy.last_args) >= len(params)
-    assert mock_run_legacy.last_args[:len(params)] == params
+    assert func.called == 1

 
 @pytest.mark.parametrize("command,func", [
                          ('postcodes', 'update_postcodes'),
                          ('word-counts', 'recompute_word_counts'),
                          ('address-levels', 'load_address_levels_from_file'),
 
 @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(monkeypatch, temp_db, command, func):
-    func_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.tools.refresh, func, func_mock)
+def test_refresh_command(mock_func_factory, temp_db, command, func):
+    func_mock = mock_func_factory(nominatim.tools.refresh, func)

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

-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)
-
-    assert 0 == call_nominatim('refresh', '--importance', '--wiki-data')
-
-    assert mock_run_legacy.called == 2
-    assert mock_run_legacy.last_args == ('update.php', '--recompute-importance')
+def test_refresh_create_functions(mock_func_factory, temp_db, tokenizer_mock):
+    func_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')

 
 

-
-@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 0 == call_nominatim('refresh', '--functions')

     assert func_mock.called == 1
     assert func_mock.called == 1

+    assert tokenizer_mock.update_sql_functions_called

 
 
 
 

-def test_replication_update_bad_interval(monkeypatch, temp_db):
-    monkeypatch.setenv('NOMINATIM_REPLICATION_UPDATE_INTERVAL', 'xx')
-
-    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')
-
-    assert call_nominatim('replication') == 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)
-
-    assert retval == call_nominatim('replication', '--once', '--no-index')
-
-
-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())
-
-    index_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_boundaries', index_mock)
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_by_rank', index_mock)
-
-    with pytest.raises(IndexError):
-        call_nominatim('replication')
-
-    assert index_mock.called == 4
-
-
-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())
-
-    index_mock = MockParamCapture()
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_boundaries', index_mock)
-    monkeypatch.setattr(nominatim.indexer.indexer.Indexer, 'index_by_rank', index_mock)
-
-    sleep_mock = MockParamCapture()
-    monkeypatch.setattr(time, 'sleep', sleep_mock)
+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'))

 
 

-    with pytest.raises(IndexError):
-        call_nominatim('replication')
+    assert 0 == call_nominatim('refresh', '--importance', '--wiki-data')

 
 

-    assert index_mock.called == 2
-    assert sleep_mock.called == 1
-    assert sleep_mock.last_args[0] == 60
+    assert calls == ['import', 'update']

 
 
 
 

-def test_serve_command(monkeypatch):
-    func = MockParamCapture()
-    monkeypatch.setattr(nominatim.cli, 'run_php_server', func)
+def test_serve_command(mock_func_factory):
+    func = mock_func_factory(nominatim.cli, 'run_php_server')

 
     call_nominatim('serve')
 
 
     call_nominatim('serve')
 


@@ -256,9 +314,8 @@ def test_serve_command(monkeypatch):
                          ('details', '--place_id', '10001'),
                          ('status',)
                          ])
                          ('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_api_commands_simple(mock_func_factory, params):
+    mock_run_api = mock_func_factory(nominatim.clicmd.api, 'run_api_script')

 
     assert 0 == call_nominatim(*params)
 
 
     assert 0 == call_nominatim(*params)