Special phrases documentation updated

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

diff --git a/test/python/test_cli.py b/test/python/test_cli.py
index 2b4240b424bec318a708c940c20d5037bc6155c8..1d89ec699e38de8db1902ae2ecf9ef03a90745e4 100644 (file)
--- a/test/python/test_cli.py
+++ b/test/python/test_cli.py
@@ -9,15 +9,20 @@ from pathlib import Path
 
 import pytest
 
 
 import pytest
 

+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.tools.admin
 import nominatim.tools.check_database
 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.freeze
 import nominatim.tools.refresh

+import nominatim.tools.postcodes
+import nominatim.tokenizer.factory

 
 from mocks import MockParamCapture
 
 
 from mocks import MockParamCapture
 


@@ -40,15 +45,40 @@ def mock_run_legacy(monkeypatch):
     monkeypatch.setattr(nominatim.cli, 'run_legacy_script', mock)
     return mock
 
     monkeypatch.setattr(nominatim.cli, 'run_legacy_script', mock)
     return mock
 

+

 @pytest.fixture
 def mock_func_factory(monkeypatch):
     def get_mock(module, func):
         mock = MockParamCapture()
 @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
 
         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.
     """


@@ -59,9 +89,6 @@ def test_cli_help(capsys):
 
 
 @pytest.mark.parametrize("command,script", [
 
 
 @pytest.mark.parametrize("command,script", [

-                         (('import', '--continue', 'load-data'), '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')
                          ])


@@ -72,6 +99,101 @@ 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')
 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')


@@ -122,7 +244,8 @@ 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(mock_func_factory, 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)")
     bnd_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_boundaries')
     rank_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_by_rank')
     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')


@@ -132,25 +255,36 @@ def test_index_command(mock_func_factory, temp_db_cursor, params, do_bnds, do_ra
     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(temp_db, mock_func_factory, tokenizer_mock, no_replace):
+    func = mock_func_factory(nominatim.clicmd.special_phrases.SPImporter, 'import_phrases')

 
 

-@pytest.mark.parametrize("command,params", [
-                         ('wiki-data', ('setup.php', '--import-wikipedia-articles')),
-                         ('importance', ('update.php', '--recompute-importance')),
-                         ])
-def test_refresh_legacy_command(mock_func_factory, temp_db, command, params):
-    mock_run_legacy = mock_func_factory(nominatim.clicmd.refresh, 'run_legacy_script')
+    if no_replace:
+        call_nominatim('special-phrases', '--import-from-wiki', '--no-replace')
+    else:
+        call_nominatim('special-phrases', '--import-from-wiki')

 
 

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

 
 

-    assert mock_run_legacy.called == 1
-    assert len(mock_run_legacy.last_args) >= len(params)
-    assert mock_run_legacy.last_args[:len(params)] == params
+@pytest.mark.parametrize("no_replace", [(True), (False)])
+def test_special_phrases_csv_command(temp_db, mock_func_factory, tokenizer_mock, 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:
+        call_nominatim('special-phrases', '--import-from-csv', csv_path, '--no-replace')
+    else:
+        call_nominatim('special-phrases', '--import-from-csv', csv_path)
+
+    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(mock_func_factory, temp_db, command, func):
                          ('website', 'setup_website'),
                          ])
 def test_refresh_command(mock_func_factory, temp_db, command, func):


@@ -160,13 +294,24 @@ def test_refresh_command(mock_func_factory, temp_db, command, func):
     assert func_mock.called == 1
 
 
     assert func_mock.called == 1
 
 

-def test_refresh_importance_computed_after_wiki_import(mock_func_factory, temp_db):
-    mock_run_legacy = mock_func_factory(nominatim.clicmd.refresh, 'run_legacy_script')
+def test_refresh_create_functions(mock_func_factory, temp_db, tokenizer_mock):
+    func_mock = mock_func_factory(nominatim.tools.refresh, 'create_functions')
+
+    assert 0 == call_nominatim('refresh', '--functions')
+    assert func_mock.called == 1
+    assert tokenizer_mock.update_sql_functions_called
+
+
+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 0 == call_nominatim('refresh', '--importance', '--wiki-data')
 
 
     assert 0 == call_nominatim('refresh', '--importance', '--wiki-data')
 

-    assert mock_run_legacy.called == 2
-    assert mock_run_legacy.last_args == ('update.php', '--recompute-importance')
+    assert calls == ['import', 'update']

 
 
 def test_serve_command(mock_func_factory):
 
 
 def test_serve_command(mock_func_factory):