add documentation for sanitizer interface

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

diff --git a/test/python/conftest.py b/test/python/conftest.py
index d4649c24929b91835b6aa3dd7d34f985904ceed6..405262950b78319d2924274ad497f933b03490f2 100644 (file)
--- a/test/python/conftest.py
+++ b/test/python/conftest.py
@@ -1,56 +1,33 @@
-import importlib
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.

 import itertools
 import sys
 from pathlib import Path
 
 import psycopg2
 import itertools
 import sys
 from pathlib import Path
 
 import psycopg2

-import psycopg2.extras

 import pytest
 import pytest

-import tempfile
-
-SRC_DIR = Path(__file__) / '..' / '..' / '..'

 
 # always test against the source
 
 # always test against the source

-sys.path.insert(0, str(SRC_DIR.resolve()))
+SRC_DIR = (Path(__file__) / '..' / '..' / '..').resolve()
+sys.path.insert(0, str(SRC_DIR))

 
 from nominatim.config import Configuration
 from nominatim.db import connection
 from nominatim.db.sql_preprocessor import SQLPreprocessor
 
 from nominatim.config import Configuration
 from nominatim.db import connection
 from nominatim.db.sql_preprocessor import SQLPreprocessor

-from nominatim.db import properties
+import nominatim.tokenizer.factory

 
 import dummy_tokenizer
 
 import dummy_tokenizer

+import mocks
+from cursor import CursorForTesting

 
 

-class _TestingCursor(psycopg2.extras.DictCursor):
-    """ Extension to the DictCursor class that provides execution
-        short-cuts that simplify writing assertions.
-    """
-
-    def scalar(self, sql, params=None):
-        """ Execute a query with a single return value and return this value.
-            Raises an assertion when not exactly one row is returned.
-        """
-        self.execute(sql, params)
-        assert self.rowcount == 1
-        return self.fetchone()[0]
-
-    def row_set(self, sql, params=None):
-        """ Execute a query and return the result as a set of tuples.
-        """
-        self.execute(sql, params)

 
 

-        return set((tuple(row) for row in self))
-
-    def table_exists(self, table):
-        """ Check that a table with the given name exists in the database.
-        """
-        num = self.scalar("""SELECT count(*) FROM pg_tables
-                             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 src_dir():
+    return SRC_DIR

 
 
 @pytest.fixture
 
 
 @pytest.fixture


@@ -68,7 +45,7 @@ def temp_db(monkeypatch):
 
     conn.close()
 
 
     conn.close()
 

-    monkeypatch.setenv('NOMINATIM_DATABASE_DSN' , 'dbname=' + name)
+    monkeypatch.setenv('NOMINATIM_DATABASE_DSN', 'dbname=' + name)

 
     yield name
 
 
     yield name
 


@@ -111,86 +88,85 @@ def temp_db_cursor(temp_db):
     """
     conn = psycopg2.connect('dbname=' + temp_db)
     conn.set_isolation_level(0)
     """
     conn = psycopg2.connect('dbname=' + temp_db)
     conn.set_isolation_level(0)

-    with conn.cursor(cursor_factory=_TestingCursor) as cur:
+    with conn.cursor(cursor_factory=CursorForTesting) as cur:

         yield cur
     conn.close()
 
 
 @pytest.fixture
 def table_factory(temp_db_cursor):
         yield cur
     conn.close()
 
 
 @pytest.fixture
 def table_factory(temp_db_cursor):

+    """ A fixture that creates new SQL tables, potentially filled with
+        content.
+    """

     def mk_table(name, definition='id INT', content=None):
         temp_db_cursor.execute('CREATE TABLE {} ({})'.format(name, definition))
         if content is not None:
     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))
+            temp_db_cursor.execute_values("INSERT INTO {} VALUES %s".format(name), content)

 
     return mk_table
 
 
 @pytest.fixture
 
     return mk_table
 
 
 @pytest.fixture

-def def_config():
-    cfg = Configuration(None, SRC_DIR.resolve() / 'settings')
+def def_config(src_dir):
+    cfg = Configuration(None, src_dir / 'settings')

     cfg.set_libdirs(module='.', osm2pgsql='.',
     cfg.set_libdirs(module='.', osm2pgsql='.',

-                    php=SRC_DIR / 'lib-php',
-                    sql=SRC_DIR / 'lib-sql',
-                    data=SRC_DIR / 'data')
+                    php=src_dir / 'lib-php',
+                    sql=src_dir / 'lib-sql',
+                    data=src_dir / 'data')

     return cfg
 
     return cfg
 

-@pytest.fixture
-def src_dir():
-    return SRC_DIR.resolve()

 
 @pytest.fixture
 
 @pytest.fixture

-def tmp_phplib_dir():
-    with tempfile.TemporaryDirectory() as phpdir:
-        (Path(phpdir) / 'admin').mkdir()
-
-        yield Path(phpdir)
+def project_env(src_dir, tmp_path):
+    projdir = tmp_path / 'project'
+    projdir.mkdir()
+    cfg = Configuration(projdir, src_dir / 'settings')
+    cfg.set_libdirs(module='.', osm2pgsql='.',
+                    php=src_dir / 'lib-php',
+                    sql=src_dir / 'lib-sql',
+                    data=src_dir / 'data')
+    return cfg

 
 
 @pytest.fixture
 
 
 @pytest.fixture

-def property_table(table_factory):
+def property_table(table_factory, temp_db_conn):

     table_factory('nominatim_properties', 'property TEXT, value TEXT')
 
     table_factory('nominatim_properties', 'property TEXT, value TEXT')
 

+    return mocks.MockPropertyTable(temp_db_conn)
+
+

 @pytest.fixture
 @pytest.fixture

-def status_table(temp_db_conn):
+def status_table(table_factory):

     """ Create an empty version of the status table and
         the status logging table.
     """
     """ Create an empty version of the status table and
         the status logging table.
     """

-    with temp_db_conn.cursor() as cur:
-        cur.execute("""CREATE TABLE import_status (
-                           lastimportdate timestamp with time zone NOT NULL,
-                           sequence_id integer,
-                           indexed boolean
-                       )""")
-        cur.execute("""CREATE TABLE import_osmosis_log (
-                           batchend timestamp,
-                           batchseq integer,
-                           batchsize bigint,
-                           starttime timestamp,
-                           endtime timestamp,
-                           event text
-                           )""")
-    temp_db_conn.commit()
+    table_factory('import_status',
+                  """lastimportdate timestamp with time zone NOT NULL,
+                     sequence_id integer,
+                     indexed boolean""")
+    table_factory('import_osmosis_log',
+                  """batchend timestamp,
+                     batchseq integer,
+                     batchsize bigint,
+                     starttime timestamp,
+                     endtime timestamp,
+                     event text""")

 
 
 @pytest.fixture
 
 
 @pytest.fixture

-def place_table(temp_db_with_extensions, temp_db_conn):
+def place_table(temp_db_with_extensions, table_factory):

     """ Create an empty version of the place table.
     """
     """ Create an empty version of the place table.
     """

-    with temp_db_conn.cursor() as cur:
-        cur.execute("""CREATE TABLE place (
-                           osm_id int8 NOT NULL,
-                           osm_type char(1) NOT NULL,
-                           class text NOT NULL,
-                           type text NOT NULL,
-                           name hstore,
-                           admin_level smallint,
-                           address hstore,
-                           extratags hstore,
-                           geometry Geometry(Geometry,4326) NOT NULL)""")
-    temp_db_conn.commit()
+    table_factory('place',
+                  """osm_id int8 NOT NULL,
+                     osm_type char(1) NOT NULL,
+                     class text NOT NULL,
+                     type text NOT NULL,
+                     name hstore,
+                     admin_level smallint,
+                     address hstore,
+                     extratags hstore,
+                     geometry Geometry(Geometry,4326) NOT NULL""")

 
 
 @pytest.fixture
 
 
 @pytest.fixture


@@ -198,6 +174,7 @@ def place_row(place_table, temp_db_cursor):
     """ A factory for rows in the place table. The table is created as a
         prerequisite to the fixture.
     """
     """ A factory for rows in the place table. The table is created as a
         prerequisite to the fixture.
     """

+    psycopg2.extras.register_hstore(temp_db_cursor)

     idseq = itertools.count(1001)
     def _insert(osm_type='N', osm_id=None, cls='amenity', typ='cafe', names=None,
                 admin_level=None, address=None, extratags=None, geom=None):
     idseq = itertools.count(1001)
     def _insert(osm_type='N', osm_id=None, cls='amenity', typ='cafe', names=None,
                 admin_level=None, address=None, extratags=None, geom=None):


@@ -212,101 +189,57 @@ def place_row(place_table, temp_db_cursor):
 def placex_table(temp_db_with_extensions, temp_db_conn):
     """ Create an empty version of the place table.
     """
 def placex_table(temp_db_with_extensions, temp_db_conn):
     """ Create an empty version of the place table.
     """

-    with temp_db_conn.cursor() as cur:
-        cur.execute("""CREATE TABLE placex (
-                           place_id BIGINT,
-                           parent_place_id BIGINT,
-                           linked_place_id BIGINT,
-                           importance FLOAT,
-                           indexed_date TIMESTAMP,
-                           geometry_sector INTEGER,
-                           rank_address SMALLINT,
-                           rank_search SMALLINT,
-                           partition SMALLINT,
-                           indexed_status SMALLINT,
-                           osm_id int8,
-                           osm_type char(1),
-                           class text,
-                           type text,
-                           name hstore,
-                           admin_level smallint,
-                           address hstore,
-                           extratags hstore,
-                           geometry Geometry(Geometry,4326),
-                           wikipedia TEXT,
-                           country_code varchar(2),
-                           housenumber TEXT,
-                           postcode TEXT,
-                           centroid GEOMETRY(Geometry, 4326))""")
-    temp_db_conn.commit()
+    return mocks.MockPlacexTable(temp_db_conn)

 
 
 @pytest.fixture
 
 
 @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()
+def osmline_table(temp_db_with_extensions, table_factory):
+    table_factory('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)""")

 
 
 @pytest.fixture
 
 
 @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()
-
-
-@pytest.fixture
-def osm2pgsql_options(temp_db):
-    return dict(osm2pgsql='echo',
-                osm2pgsql_cache=10,
-                osm2pgsql_style='style.file',
-                threads=1,
-                dsn='dbname=' + temp_db,
-                flatnode_file='',
-                tablespaces=dict(slim_data='', slim_index='',
-                                 main_data='', main_index=''))
-
-@pytest.fixture
-def sql_preprocessor(temp_db_conn, tmp_path, monkeypatch, table_factory):
-    table_factory('country_name', 'partition INT', (0, 1, 2))
+def sql_preprocessor_cfg(tmp_path, table_factory, temp_db_with_extensions):
+    table_factory('country_name', 'partition INT', ((0, ), (1, ), (2, )))

     cfg = Configuration(None, SRC_DIR.resolve() / 'settings')
     cfg.set_libdirs(module='.', osm2pgsql='.', php=SRC_DIR / 'lib-php',
                     sql=tmp_path, data=SRC_DIR / 'data')
     cfg = Configuration(None, SRC_DIR.resolve() / 'settings')
     cfg.set_libdirs(module='.', osm2pgsql='.', php=SRC_DIR / 'lib-php',
                     sql=tmp_path, data=SRC_DIR / 'data')

+    return cfg
+

 
 

-    return SQLPreprocessor(temp_db_conn, cfg)
+@pytest.fixture
+def sql_preprocessor(sql_preprocessor_cfg, temp_db_conn):
+    return SQLPreprocessor(temp_db_conn, sql_preprocessor_cfg)

 
 
 @pytest.fixture
 
 
 @pytest.fixture

-def tokenizer_mock(monkeypatch, property_table, temp_db_conn):
+def tokenizer_mock(monkeypatch, property_table):

     """ Sets up the configuration so that the test dummy tokenizer will be
     """ Sets up the configuration so that the test dummy tokenizer will be

-        loaded.
+        loaded when the tokenizer factory is used. Also returns a factory
+        with which a new dummy tokenizer may be created.

     """
     monkeypatch.setenv('NOMINATIM_TOKENIZER', 'dummy')
 
     """
     monkeypatch.setenv('NOMINATIM_TOKENIZER', 'dummy')
 

-    def _import_dummy(module, *args, **kwargs):
+    def _import_dummy(*args, **kwargs):

         return dummy_tokenizer
 
         return dummy_tokenizer
 

-    monkeypatch.setattr(importlib, "import_module", _import_dummy)
-    properties.set_property(temp_db_conn, 'tokenizer', 'dummy')
+    monkeypatch.setattr(nominatim.tokenizer.factory, "_import_tokenizer", _import_dummy)
+    property_table.set('tokenizer', 'dummy')
+
+    def _create_tokenizer():
+        return dummy_tokenizer.DummyTokenizer(None, None)
+
+    return _create_tokenizer