move setup function to python

There are still back-calls to PHP for some of the sub-steps.
These needs some larger refactoring to be moved to Python.

test/python/conftest.py

@@ -43,6 +43,11 @@ class _TestingCursor(psycopg2.extras.DictCursor):
                              WHERE tablename = %s""", (table, ))
         return num == 1
 
+    def table_rows(self, table):
+        """ Return the number of rows in the given table.
+        """
+        return self.scalar('SELECT count(*) FROM ' + table)
+
 
 @pytest.fixture
 def temp_db(monkeypatch):
@@ -109,8 +114,12 @@ def temp_db_cursor(temp_db):
 
 @pytest.fixture
 def table_factory(temp_db_cursor):
-    def mk_table(name, definition='id INT'):
+    def mk_table(name, definition='id INT', content=None):
         temp_db_cursor.execute('CREATE TABLE {} ({})'.format(name, definition))
+        if content is not None:
+            if not isinstance(content, str):
+                content = '),('.join([str(x) for x in content])
+            temp_db_cursor.execute("INSERT INTO {} VALUES ({})".format(name, content))
 
     return mk_table
 
@@ -174,7 +183,7 @@ def place_row(place_table, temp_db_cursor):
         temp_db_cursor.execute("INSERT INTO place VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)",
                                (osm_id or next(idseq), osm_type, cls, typ, names,
                                 admin_level, address, extratags,
-                                geom or 'SRID=4326;POINT(0 0 )'))
+                                geom or 'SRID=4326;POINT(0 0)'))
 
     return _insert
 
@@ -184,7 +193,7 @@ def placex_table(temp_db_with_extensions, temp_db_conn):
     """
     with temp_db_conn.cursor() as cur:
         cur.execute("""CREATE TABLE placex (
-                           place_id BIGINT NOT NULL,
+                           place_id BIGINT,
                            parent_place_id BIGINT,
                            linked_place_id BIGINT,
                            importance FLOAT,
@@ -207,8 +216,43 @@ def placex_table(temp_db_with_extensions, temp_db_conn):
                            country_code varchar(2),
                            housenumber TEXT,
                            postcode TEXT,
-                           centroid GEOMETRY(Geometry, 4326))
-                           """)
+                           centroid GEOMETRY(Geometry, 4326))""")
+    temp_db_conn.commit()
+
+
+@pytest.fixture
+def osmline_table(temp_db_with_extensions, temp_db_conn):
+    with temp_db_conn.cursor() as cur:
+        cur.execute("""CREATE TABLE location_property_osmline (
+                           place_id BIGINT,
+                           osm_id BIGINT,
+                           parent_place_id BIGINT,
+                           geometry_sector INTEGER,
+                           indexed_date TIMESTAMP,
+                           startnumber INTEGER,
+                           endnumber INTEGER,
+                           partition SMALLINT,
+                           indexed_status SMALLINT,
+                           linegeo GEOMETRY,
+                           interpolationtype TEXT,
+                           address HSTORE,
+                           postcode TEXT,
+                           country_code VARCHAR(2))""")
+    temp_db_conn.commit()
+
+
+@pytest.fixture
+def word_table(temp_db, temp_db_conn):
+    with temp_db_conn.cursor() as cur:
+        cur.execute("""CREATE TABLE word (
+                           word_id INTEGER,
+                           word_token text,
+                           word text,
+                           class text,
+                           type text,
+                           country_code varchar(2),
+                           search_name_count INTEGER,
+                           operator TEXT)""")
     temp_db_conn.commit()
 
 

test/python/test_cli.py

@@ -13,9 +13,11 @@ 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
 
@@ -61,7 +63,6 @@ def test_cli_help(capsys):
 
 
 @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'),
@@ -74,6 +75,36 @@ def test_legacy_commands_simple(mock_run_legacy, command, script):
     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):
+    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.indexer.indexer.Indexer, 'index_full'),
+        mock_func_factory(nominatim.tools.refresh, 'setup_website'),
+    ]
+
+    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 cf_mock.called > 1
+
+    for mock in mocks:
+        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')

test/python/test_tools_check_database.py

@@ -63,6 +63,10 @@ def test_check_database_indexes_bad(temp_db_conn, def_config):
     assert chkdb.check_database_indexes(temp_db_conn, def_config) == chkdb.CheckState.FAIL
 
 
+def test_check_database_indexes_valid(temp_db_conn, def_config):
+    assert chkdb.check_database_index_valid(temp_db_conn, def_config) == chkdb.CheckState.OK
+
+
 def test_check_tiger_table_disabled(temp_db_conn, def_config, monkeypatch):
     monkeypatch.setenv('NOMINATIM_USE_US_TIGER_DATA' , 'no')
     assert chkdb.check_tiger_table(temp_db_conn, def_config) == chkdb.CheckState.NOT_APPLICABLE

test/python/test_tools_database_import.py

@@ -24,6 +24,24 @@ def nonexistant_db():
     with conn.cursor() as cur:
         cur.execute('DROP DATABASE IF EXISTS {}'.format(dbname))
 
+@pytest.mark.parametrize("no_partitions", (True, False))
+def test_setup_skeleton(src_dir, nonexistant_db, no_partitions):
+    database_import.setup_database_skeleton('dbname=' + nonexistant_db,
+                                            src_dir / 'data', no_partitions)
+
+    conn = psycopg2.connect(database=nonexistant_db)
+
+    try:
+        with conn.cursor() as cur:
+            cur.execute("SELECT distinct partition FROM country_name")
+            partitions = set([r[0] for r in list(cur)])
+            if no_partitions:
+                assert partitions == set([0])
+            else:
+                assert len(partitions) > 10
+    finally:
+        conn.close()
+
 
 def test_create_db_success(nonexistant_db):
     database_import.create_db('dbname=' + nonexistant_db, rouser='www-data')
@@ -79,6 +97,22 @@ def test_install_module(tmp_path):
     assert outfile.stat().st_mode == 33261
 
 
+def test_install_module_custom(tmp_path):
+    (tmp_path / 'nominatim.so').write_text('TEST nomiantim.so')
+
+    database_import.install_module(tmp_path, tmp_path, str(tmp_path.resolve()))
+
+    assert not (tmp_path / 'module').exists()
+
+
+def test_install_module_fail_access(temp_db_conn, tmp_path):
+    (tmp_path / 'nominatim.so').write_text('TEST nomiantim.so')
+
+    with pytest.raises(UsageError, match='.*module cannot be accessed.*'):
+        database_import.install_module(tmp_path, tmp_path, '',
+                                       conn=temp_db_conn)
+
+
 def test_import_base_data(src_dir, temp_db, temp_db_cursor):
     temp_db_cursor.execute('CREATE EXTENSION hstore')
     temp_db_cursor.execute('CREATE EXTENSION postgis')
@@ -134,3 +168,35 @@ def test_import_osm_data_default_cache(temp_db_cursor,osm2pgsql_options):
     osm2pgsql_options['osm2pgsql_cache'] = 0
 
     database_import.import_osm_data(Path(__file__), osm2pgsql_options)
+
+
+def test_truncate_database_tables(temp_db_conn, temp_db_cursor, table_factory):
+    tables = ('word', 'placex', 'place_addressline', 'location_area',
+              'location_area_country', 'location_property',
+              'location_property_tiger', 'location_property_osmline',
+              'location_postcode', 'search_name', 'location_road_23')
+    for table in tables:
+        table_factory(table, content=(1, 2, 3))
+
+    database_import.truncate_data_tables(temp_db_conn, max_word_frequency=23)
+
+    for table in tables:
+        assert temp_db_cursor.table_rows(table) == 0
+
+
+@pytest.mark.parametrize("threads", (1, 5))
+def test_load_data(dsn, src_dir, place_row, placex_table, osmline_table, word_table,
+                   temp_db_cursor, threads):
+    for func in ('make_keywords', 'getorcreate_housenumber_id', 'make_standard_name'):
+        temp_db_cursor.execute("""CREATE FUNCTION {} (src TEXT)
+                                  RETURNS TEXT AS $$ SELECT 'a' $$ LANGUAGE SQL
+                               """.format(func))
+    for oid in range(100, 130):
+        place_row(osm_id=oid)
+    place_row(osm_type='W', osm_id=342, cls='place', typ='houses',
+              geom='SRID=4326;LINESTRING(0 0, 10 10)')
+
+    database_import.load_data(dsn, src_dir / 'data', threads)
+
+    assert temp_db_cursor.table_rows('placex') == 30
+    assert temp_db_cursor.table_rows('location_property_osmline') == 1