remove warning about now removed legacy tokenizer

[nominatim.git] / test / bdd / steps / nominatim_environment.py

diff --git a/test/bdd/steps/nominatim_environment.py b/test/bdd/steps/nominatim_environment.py
index 168334b1659d381b1ea98875d0ce90a14f8e412e..1b0c7c5f4dba9c281c589029dca2de07b738e1b1 100644 (file)
--- a/test/bdd/steps/nominatim_environment.py
+++ b/test/bdd/steps/nominatim_environment.py
@@ -1,15 +1,21 @@
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2024 by the Nominatim developer community.
+# For a full list of authors see the git log.

 from pathlib import Path
 from pathlib import Path

-import sys
+import importlib

 import tempfile
 
 import tempfile
 

-import psycopg2
-import psycopg2.extras
+import psycopg
+from psycopg import sql as pysql

 
 

-sys.path.insert(1, str((Path(__file__) / '..' / '..' / '..' / '..').resolve()))
-
-from nominatim import cli
-from nominatim.config import Configuration
-from nominatim.tools import refresh
+from nominatim_db import cli
+from nominatim_db.config import Configuration
+from nominatim_db.db.connection import Connection, register_hstore, execute_scalar
+from nominatim_db.tools import refresh
+from nominatim_db.tokenizer import factory as tokenizer_factory

 from steps.utils import run_script
 
 class NominatimEnvironment:
 from steps.utils import run_script
 
 class NominatimEnvironment:


@@ -17,7 +23,6 @@ class NominatimEnvironment:
     """
 
     def __init__(self, config):
     """
 
     def __init__(self, config):

-        self.build_dir = Path(config['BUILDDIR']).resolve()

         self.src_dir = (Path(__file__) / '..' / '..' / '..' / '..').resolve()
         self.db_host = config['DB_HOST']
         self.db_port = config['DB_PORT']
         self.src_dir = (Path(__file__) / '..' / '..' / '..' / '..').resolve()
         self.db_host = config['DB_HOST']
         self.db_port = config['DB_PORT']


@@ -27,23 +32,26 @@ class NominatimEnvironment:
         self.test_db = config['TEST_DB']
         self.api_test_db = config['API_TEST_DB']
         self.api_test_file = config['API_TEST_FILE']
         self.test_db = config['TEST_DB']
         self.api_test_db = config['API_TEST_DB']
         self.api_test_file = config['API_TEST_FILE']

-        self.server_module_path = config['SERVER_MODULE_PATH']
+        self.tokenizer = config['TOKENIZER']
+        self.import_style = config['STYLE']

         self.reuse_template = not config['REMOVE_TEMPLATE']
         self.keep_scenario_db = config['KEEP_TEST_DB']
         self.reuse_template = not config['REMOVE_TEMPLATE']
         self.keep_scenario_db = config['KEEP_TEST_DB']

-        self.code_coverage_path = config['PHPCOV']
-        self.code_coverage_id = 1

 
 

-        self.default_config = Configuration(None, self.src_dir / 'settings').get_os_env()
+        self.default_config = Configuration(None).get_os_env()

         self.test_env = None
         self.template_db_done = False
         self.api_db_done = False
         self.website_dir = None
 
         self.test_env = None
         self.template_db_done = False
         self.api_db_done = False
         self.website_dir = None
 

+        if not hasattr(self, f"create_api_request_func_{config['API_ENGINE']}"):
+            raise RuntimeError(f"Unknown API engine '{config['API_ENGINE']}'")
+        self.api_engine = getattr(self, f"create_api_request_func_{config['API_ENGINE']}")()
+

     def connect_database(self, dbname):
         """ Return a connection to the database with the given name.
             Uses configured host, user and port.
         """
     def connect_database(self, dbname):
         """ Return a connection to the database with the given name.
             Uses configured host, user and port.
         """

-        dbargs = {'database': dbname}
+        dbargs = {'dbname': dbname, 'row_factory': psycopg.rows.dict_row}

         if self.db_host:
             dbargs['host'] = self.db_host
         if self.db_port:
         if self.db_host:
             dbargs['host'] = self.db_host
         if self.db_port:


@@ -52,16 +60,8 @@ class NominatimEnvironment:
             dbargs['user'] = self.db_user
         if self.db_pass:
             dbargs['password'] = self.db_pass
             dbargs['user'] = self.db_user
         if self.db_pass:
             dbargs['password'] = self.db_pass

-        conn = psycopg2.connect(**dbargs)
-        return conn
-
-    def next_code_coverage_file(self):
-        """ Generate the next name for a coverage file.
-        """
-        fn = Path(self.code_coverage_path) / "{:06d}.cov".format(self.code_coverage_id)
-        self.code_coverage_id += 1
+        return psycopg.connect(**dbargs)

 
 

-        return fn.resolve()

 
     def write_nominatim_config(self, dbname):
         """ Set up a custom test configuration that connects to the given
 
     def write_nominatim_config(self, dbname):
         """ Set up a custom test configuration that connects to the given


@@ -69,7 +69,10 @@ class NominatimEnvironment:
             be picked up by dotenv and creates a project directory with the
             appropriate website scripts.
         """
             be picked up by dotenv and creates a project directory with the
             appropriate website scripts.
         """

-        dsn = 'pgsql:dbname={}'.format(dbname)
+        if dbname.startswith('sqlite:'):
+            dsn = 'sqlite:dbname={}'.format(dbname[7:])
+        else:
+            dsn = 'pgsql:dbname={}'.format(dbname)

         if self.db_host:
             dsn += ';host=' + self.db_host
         if self.db_port:
         if self.db_host:
             dsn += ';host=' + self.db_host
         if self.db_port:


@@ -79,45 +82,54 @@ class NominatimEnvironment:
         if self.db_pass:
             dsn += ';password=' + self.db_pass
 
         if self.db_pass:
             dsn += ';password=' + self.db_pass
 

-        if self.website_dir is not None \
-           and self.test_env is not None \
-           and dsn == self.test_env['NOMINATIM_DATABASE_DSN']:
-            return # environment already set uo
-

         self.test_env = dict(self.default_config)
         self.test_env['NOMINATIM_DATABASE_DSN'] = dsn
         self.test_env = dict(self.default_config)
         self.test_env['NOMINATIM_DATABASE_DSN'] = dsn

+        self.test_env['NOMINATIM_LANGUAGES'] = 'en,de,fr,ja'

         self.test_env['NOMINATIM_FLATNODE_FILE'] = ''
         self.test_env['NOMINATIM_IMPORT_STYLE'] = 'full'
         self.test_env['NOMINATIM_USE_US_TIGER_DATA'] = 'yes'
         self.test_env['NOMINATIM_DATADIR'] = str((self.src_dir / 'data').resolve())
         self.test_env['NOMINATIM_SQLDIR'] = str((self.src_dir / 'lib-sql').resolve())
         self.test_env['NOMINATIM_CONFIGDIR'] = str((self.src_dir / 'settings').resolve())
         self.test_env['NOMINATIM_FLATNODE_FILE'] = ''
         self.test_env['NOMINATIM_IMPORT_STYLE'] = 'full'
         self.test_env['NOMINATIM_USE_US_TIGER_DATA'] = 'yes'
         self.test_env['NOMINATIM_DATADIR'] = str((self.src_dir / 'data').resolve())
         self.test_env['NOMINATIM_SQLDIR'] = str((self.src_dir / 'lib-sql').resolve())
         self.test_env['NOMINATIM_CONFIGDIR'] = str((self.src_dir / 'settings').resolve())

-        self.test_env['NOMINATIM_DATABASE_MODULE_SRC_PATH'] = str((self.build_dir / 'module').resolve())
-        self.test_env['NOMINATIM_OSM2PGSQL_BINARY'] = str((self.build_dir / 'osm2pgsql' / 'osm2pgsql').resolve())
-        self.test_env['NOMINATIM_NOMINATIM_TOOL'] = str((self.build_dir / 'nominatim').resolve())
-
-        if self.server_module_path:
-            self.test_env['NOMINATIM_DATABASE_MODULE_PATH'] = self.server_module_path
-        else:
-            # avoid module being copied into the temporary environment
-            self.test_env['NOMINATIM_DATABASE_MODULE_PATH'] = str((self.build_dir / 'module').resolve())
+        if self.tokenizer is not None:
+            self.test_env['NOMINATIM_TOKENIZER'] = self.tokenizer
+        if self.import_style is not None:
+            self.test_env['NOMINATIM_IMPORT_STYLE'] = self.import_style

 
         if self.website_dir is not None:
             self.website_dir.cleanup()
 
         self.website_dir = tempfile.TemporaryDirectory()
 
         if self.website_dir is not None:
             self.website_dir.cleanup()
 
         self.website_dir = tempfile.TemporaryDirectory()

-        cfg = Configuration(None, self.src_dir / 'settings', environ=self.test_env)
-        refresh.setup_website(Path(self.website_dir.name) / 'website', self.src_dir / 'lib-php', cfg)
+
+
+    def get_test_config(self):
+        cfg = Configuration(Path(self.website_dir.name), environ=self.test_env)
+        return cfg
+
+    def get_libpq_dsn(self):
+        dsn = self.test_env['NOMINATIM_DATABASE_DSN']
+
+        def quote_param(param):
+            key, val = param.split('=')
+            val = val.replace('\\', '\\\\').replace("'", "\\'")
+            if ' ' in val:
+                val = "'" + val + "'"
+            return key + '=' + val
+
+        if dsn.startswith('pgsql:'):
+            # Old PHP DSN format. Convert before returning.
+            return ' '.join([quote_param(p) for p in dsn[6:].split(';')])
+
+        return dsn

 
 
     def db_drop_database(self, name):
         """ Drop the database with the given name.
         """
 
 
     def db_drop_database(self, name):
         """ Drop the database with the given name.
         """

-        conn = self.connect_database('postgres')
-        conn.set_isolation_level(0)
-        cur = conn.cursor()
-        cur.execute('DROP DATABASE IF EXISTS {}'.format(name))
-        conn.close()
+        with self.connect_database('postgres') as conn:
+            conn.autocommit = True
+            conn.execute(pysql.SQL('DROP DATABASE IF EXISTS')
+                         +  pysql.Identifier(name))

 
     def setup_template_db(self):
         """ Setup a template database that already contains common test data.
 
     def setup_template_db(self):
         """ Setup a template database that already contains common test data.


@@ -129,40 +141,23 @@ class NominatimEnvironment:
 
         self.template_db_done = True
 
 
         self.template_db_done = True
 

-        if self._reuse_or_drop_db(self.template_db):
-            return
+        self.write_nominatim_config(self.template_db)

 
 

-        try:
-            # call the first part of database setup
-            self.write_nominatim_config(self.template_db)
-            self.run_setup_script('create-db', 'setup-db')
-            # remove external data to speed up indexing for tests
-            conn = self.connect_database(self.template_db)
-            cur = conn.cursor()
-            cur.execute("""select tablename from pg_tables
-                           where tablename in ('gb_postcode', 'us_postcode')""")
-            for t in cur:
-                conn.cursor().execute('TRUNCATE TABLE {}'.format(t[0]))
-            conn.commit()
-            conn.close()
-
-            # execute osm2pgsql import on an empty file to get the right tables
-            with tempfile.NamedTemporaryFile(dir='/tmp', suffix='.xml') as fd:
-                fd.write(b'<osm version="0.6"></osm>')
-                fd.flush()
-                self.run_setup_script('import-data',
-                                      'ignore-errors',
-                                      'create-functions',
-                                      'create-tables',
-                                      'create-partition-tables',
-                                      'create-partition-functions',
-                                      'load-data',
-                                      'create-search-indices',
-                                      osm_file=fd.name,
-                                      osm2pgsql_cache='200')
-        except:
-            self.db_drop_database(self.template_db)
-            raise
+        if not self._reuse_or_drop_db(self.template_db):
+            try:
+                # execute nominatim import on an empty file to get the right tables
+                with tempfile.NamedTemporaryFile(dir='/tmp', suffix='.xml') as fd:
+                    fd.write(b'<osm version="0.6"></osm>')
+                    fd.flush()
+                    self.run_nominatim('import', '--osm-file', fd.name,
+                                                 '--osm2pgsql-cache', '1',
+                                                 '--ignore-errors',
+                                                 '--offline', '--index-noanalyse')
+            except:
+                self.db_drop_database(self.template_db)
+                raise
+
+        self.run_nominatim('refresh', '--functions')

 
 
     def setup_api_db(self):
 
 
     def setup_api_db(self):


@@ -170,130 +165,112 @@ class NominatimEnvironment:
         """
         self.write_nominatim_config(self.api_test_db)
 
         """
         self.write_nominatim_config(self.api_test_db)
 

-        if self.api_db_done:
+        if self.api_test_db.startswith('sqlite:'):

             return
 
             return
 

-        self.api_db_done = True
+        if not self.api_db_done:
+            self.api_db_done = True

 
 

-        if self._reuse_or_drop_db(self.api_test_db):
-            return
+            if not self._reuse_or_drop_db(self.api_test_db):
+                testdata = (Path(__file__) / '..' / '..' / '..' / 'testdb').resolve()
+                self.test_env['NOMINATIM_WIKIPEDIA_DATA_PATH'] = str(testdata)
+                simp_file = Path(self.website_dir.name) / 'secondary_importance.sql.gz'
+                simp_file.symlink_to(testdata / 'secondary_importance.sql.gz')

 
 

-        testdata = Path('__file__') / '..' / '..' / 'testdb'
-        self.test_env['NOMINATIM_TIGER_DATA_PATH'] = str((testdata / 'tiger').resolve())
-        self.test_env['NOMINATIM_WIKIPEDIA_DATA_PATH'] = str(testdata.resolve())
+                try:
+                    self.run_nominatim('import', '--osm-file', str(self.api_test_file))
+                    self.run_nominatim('add-data', '--tiger-data', str(testdata / 'tiger'))
+                    self.run_nominatim('freeze')

 
 

-        try:
-            self.run_nominatim('import', '--osm-file', str(self.api_test_file))
-            self.run_setup_script('import-tiger-data')
-            self.run_nominatim('freeze')
+                    csv_path = str(testdata / 'full_en_phrases_test.csv')
+                    self.run_nominatim('special-phrases', '--import-from-csv', csv_path)
+                except:
+                    self.db_drop_database(self.api_test_db)
+                    raise

 
 

-            phrase_file = str((testdata / 'specialphrases_testdb.sql').resolve())
-            run_script(['psql', '-d', self.api_test_db, '-f', phrase_file])
-        except:
-            self.db_drop_database(self.api_test_db)
-            raise
+        tokenizer_factory.get_tokenizer_for_db(self.get_test_config())

 
 
     def setup_unknown_db(self):
         """ Setup a test against a non-existing database.
         """
 
 
     def setup_unknown_db(self):
         """ Setup a test against a non-existing database.
         """

-        self.write_nominatim_config('UNKNOWN_DATABASE_NAME')
+        # The tokenizer needs an existing database to function.
+        # So start with the usual database
+        class _Context:
+            db = None
+
+        context = _Context()
+        self.setup_db(context)
+        tokenizer_factory.create_tokenizer(self.get_test_config(), init_db=False)
+
+        # Then drop the DB again
+        self.teardown_db(context, force_drop=True)

 
     def setup_db(self, context):
         """ Setup a test against a fresh, empty test database.
         """
         self.setup_template_db()
 
     def setup_db(self, context):
         """ Setup a test against a fresh, empty test database.
         """
         self.setup_template_db()

+        with self.connect_database(self.template_db) as conn:
+            conn.autocommit = True
+            conn.execute(pysql.SQL('DROP DATABASE IF EXISTS')
+                                   + pysql.Identifier(self.test_db))
+            conn.execute(pysql.SQL('CREATE DATABASE {} TEMPLATE = {}').format(
+                           pysql.Identifier(self.test_db),
+                           pysql.Identifier(self.template_db)))
+

         self.write_nominatim_config(self.test_db)
         self.write_nominatim_config(self.test_db)

-        conn = self.connect_database(self.template_db)
-        conn.set_isolation_level(0)
-        cur = conn.cursor()
-        cur.execute('DROP DATABASE IF EXISTS {}'.format(self.test_db))
-        cur.execute('CREATE DATABASE {} TEMPLATE = {}'.format(self.test_db, self.template_db))
-        conn.close()

         context.db = self.connect_database(self.test_db)
         context.db.autocommit = True
         context.db = self.connect_database(self.test_db)
         context.db.autocommit = True

-        psycopg2.extras.register_hstore(context.db, globally=False)
+        register_hstore(context.db)

 
 

-    def teardown_db(self, context):
+    def teardown_db(self, context, force_drop=False):

         """ Remove the test database, if it exists.
         """
         """ Remove the test database, if it exists.
         """

-        if 'db' in context:
+        if hasattr(context, 'db'):

             context.db.close()
 
             context.db.close()
 

-        if not self.keep_scenario_db:
+        if force_drop or not self.keep_scenario_db:

             self.db_drop_database(self.test_db)
 
     def _reuse_or_drop_db(self, name):
             self.db_drop_database(self.test_db)
 
     def _reuse_or_drop_db(self, name):

-        """ Check for the existance of the given DB. If reuse is enabled,
-            then the function checks for existance and returns True if the
+        """ Check for the existence of the given DB. If reuse is enabled,
+            then the function checks for existnce and returns True if the

             database is already there. Otherwise an existing database is
             dropped and always false returned.
         """
         if self.reuse_template:
             database is already there. Otherwise an existing database is
             dropped and always false returned.
         """
         if self.reuse_template:

-            conn = self.connect_database('postgres')
-            with conn.cursor() as cur:
-                cur.execute('select count(*) from pg_database where datname = %s',
-                            (name,))
-                if cur.fetchone()[0] == 1:
+            with self.connect_database('postgres') as conn:
+                num = execute_scalar(conn,
+                                     'select count(*) from pg_database where datname = %s',
+                                     (name,))
+                if num == 1:

                     return True
                     return True

-            conn.close()

         else:
             self.db_drop_database(name)
 
         return False
 
         else:
             self.db_drop_database(name)
 
         return False
 

+

     def reindex_placex(self, db):
         """ Run the indexing step until all data in the placex has
             been processed. Indexing during updates can produce more data
             to index under some circumstances. That is why indexing may have
             to be run multiple times.
         """
     def reindex_placex(self, db):
         """ Run the indexing step until all data in the placex has
             been processed. Indexing during updates can produce more data
             to index under some circumstances. That is why indexing may have
             to be run multiple times.
         """

-        with db.cursor() as cur:
-            while True:
-                self.run_nominatim('index')
+        self.run_nominatim('index')

 
 

-                cur.execute("SELECT 'a' FROM placex WHERE indexed_status != 0 LIMIT 1")
-                if cur.rowcount == 0:
-                    return

 
     def run_nominatim(self, *cmdline):
         """ Run the nominatim command-line tool via the library.
         """
 
     def run_nominatim(self, *cmdline):
         """ Run the nominatim command-line tool via the library.
         """

-        cli.nominatim(module_dir='',
-                      osm2pgsql_path=str(self.build_dir / 'osm2pgsql' / 'osm2pgsql'),
-                      phplib_dir=str(self.src_dir / 'lib-php'),
-                      sqllib_dir=str(self.src_dir / 'lib-sql'),
-                      data_dir=str(self.src_dir / 'data'),
-                      config_dir=str(self.src_dir / 'settings'),
+        if self.website_dir is not None:
+            cmdline = list(cmdline) + ['--project-dir', self.website_dir.name]
+
+        cli.nominatim(module_dir=None,
+                      osm2pgsql_path=None,

                       cli_args=cmdline,
                       cli_args=cmdline,

-                      phpcgi_path='',

                       environ=self.test_env)
 
                       environ=self.test_env)
 

-    def run_setup_script(self, *args, **kwargs):
-        """ Run the Nominatim setup script with the given arguments.
-        """
-        self.run_nominatim_script('setup', *args, **kwargs)
-
-    def run_update_script(self, *args, **kwargs):
-        """ Run the Nominatim update script with the given arguments.
-        """
-        self.run_nominatim_script('update', *args, **kwargs)
-
-    def run_nominatim_script(self, script, *args, **kwargs):
-        """ Run one of the Nominatim utility scripts with the given arguments.
-        """
-        cmd = ['/usr/bin/env', 'php', '-Cq']
-        cmd.append((Path(self.src_dir) / 'lib-php' / 'admin' / '{}.php'.format(script)).resolve())
-        cmd.extend(['--' + x for x in args])
-        for k, v in kwargs.items():
-            cmd.extend(('--' + k.replace('_', '-'), str(v)))
-
-        if self.website_dir is not None:
-            cwd = self.website_dir.name
-        else:
-            cwd = None
-
-        run_script(cmd, cwd=cwd, env=self.test_env)

 
     def copy_from_place(self, db):
         """ Copy data from place to the placex and location_property_osmline
 
     def copy_from_place(self, db):
         """ Copy data from place to the placex and location_property_osmline


@@ -316,3 +293,40 @@ class NominatimEnvironment:
                               WHERE class='place' and type='houses'
                                     and osm_type='W'
                                     and ST_GeometryType(geometry) = 'ST_LineString'""")
                               WHERE class='place' and type='houses'
                                     and osm_type='W'
                                     and ST_GeometryType(geometry) = 'ST_LineString'""")

+
+
+    def create_api_request_func_starlette(self):
+        import nominatim_api.server.starlette.server
+        from asgi_lifespan import LifespanManager
+        import httpx
+
+        async def _request(endpoint, params, project_dir, environ, http_headers):
+            app = nominatim_api.server.starlette.server.get_application(project_dir, environ)
+
+            async with LifespanManager(app):
+                async with httpx.AsyncClient(app=app, base_url="http://nominatim.test") as client:
+                    response = await client.get(f"/{endpoint}", params=params,
+                                                headers=http_headers)
+
+            return response.text, response.status_code
+
+        return _request
+
+
+    def create_api_request_func_falcon(self):
+        import nominatim_api.server.falcon.server
+        import falcon.testing
+
+        async def _request(endpoint, params, project_dir, environ, http_headers):
+            app = nominatim_api.server.falcon.server.get_application(project_dir, environ)
+
+            async with falcon.testing.ASGIConductor(app) as conductor:
+                response = await conductor.get(f"/{endpoint}", params=params,
+                                               headers=http_headers)
+
+            return response.text, response.status_code
+
+        return _request
+
+
+