Merge pull request #2321 from AntoJvlt/csv-import-special-phrases

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

diff --git a/test/python/test_cli.py b/test/python/test_cli.py
index eb0ee58487b5917112824ad4b0ce2940d5d21f3a..27ea0d5dbaea84958408d3b40b0a08ccd0517171 100644 (file)
--- a/test/python/test_cli.py
+++ b/test/python/test_cli.py
@@ -21,6 +21,8 @@ import nominatim.tools.check_database
 import nominatim.tools.database_import
 import nominatim.tools.freeze
 import nominatim.tools.refresh
 import nominatim.tools.database_import
 import nominatim.tools.freeze
 import nominatim.tools.refresh

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

 
 from mocks import MockParamCapture
 
 
 from mocks import MockParamCapture
 


@@ -48,12 +50,35 @@ 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
 
 

+@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.
     """


@@ -82,10 +107,9 @@ def test_import_bad_file(temp_db):
     assert 1 == call_nominatim('import', '--osm-file', '.')
 
 
     assert 1 == call_nominatim('import', '--osm-file', '.')
 
 

-def test_import_full(temp_db, mock_func_factory):
+def test_import_full(temp_db, mock_func_factory, tokenizer_mock):

     mocks = [
         mock_func_factory(nominatim.tools.database_import, 'setup_database_skeleton'),
     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, 'import_osm_data'),
         mock_func_factory(nominatim.tools.refresh, 'import_wikipedia_articles'),
         mock_func_factory(nominatim.tools.database_import, 'truncate_data_tables'),


@@ -96,20 +120,79 @@ def test_import_full(temp_db, mock_func_factory):
         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.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')
         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')

-    mock_func_factory(nominatim.clicmd.setup, 'run_legacy_script')

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

+    assert tokenizer_mock.finalize_import_called

 
     assert cf_mock.called > 1
 
     for mock in mocks:
 
     assert cf_mock.called > 1
 
     for mock in mocks:

-        assert mock.called == 1
+        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, '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('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')
 
 def test_freeze_command(mock_func_factory, temp_db):
     mock_drop = mock_func_factory(nominatim.tools.freeze, 'drop_update_tables')


@@ -161,7 +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')


@@ -171,30 +255,60 @@ 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
 

-def test_special_phrases_command(temp_db, mock_func_factory):
-    func = mock_func_factory(nominatim.clicmd.special_phrases.SpecialPhrasesImporter, 'import_from_wiki')
+@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')
+
+    if no_replace:
+        call_nominatim('special-phrases', '--import-from-wiki', '--no-replace')
+    else:
+        call_nominatim('special-phrases', '--import-from-wiki')
+
+    assert func.called == 1
+
+@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())

 
 

-    call_nominatim('special-phrases', '--import-from-wiki')
+    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", [
 
     assert func.called == 1
 
 @pytest.mark.parametrize("command,func", [

-                         ('postcodes', 'update_postcodes'),

                          ('word-counts', 'recompute_word_counts'),
                          ('address-levels', 'load_address_levels_from_file'),
                          ('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'),
                          ])
                          ('wiki-data', 'import_wikipedia_articles'),
                          ('importance', 'recompute_importance'),
                          ('website', 'setup_website'),
                          ])

-def test_refresh_command(mock_func_factory, temp_db, command, func):
+def test_refresh_command(mock_func_factory, temp_db, command, func, tokenizer_mock):

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

-def test_refresh_importance_computed_after_wiki_import(monkeypatch, temp_db):
+def test_refresh_postcodes(mock_func_factory, temp_db, tokenizer_mock):
+    func_mock = mock_func_factory(nominatim.tools.postcodes, 'update_postcodes')
+    idx_mock = mock_func_factory(nominatim.indexer.indexer.Indexer, 'index_postcodes')
+
+    assert 0 == call_nominatim('refresh', '--postcodes')
+    assert func_mock.called == 1
+
+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, tokenizer_mock):

     calls = []
     monkeypatch.setattr(nominatim.tools.refresh, 'import_wikipedia_articles',
                         lambda *args, **kwargs: calls.append('import') or 0)
     calls = []
     monkeypatch.setattr(nominatim.tools.refresh, 'import_wikipedia_articles',
                         lambda *args, **kwargs: calls.append('import') or 0)