Special phrases documentation updated

[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 345e134975db02b77425634d4ecb4a08b8242661..937978b0a138899b754191593f354e31a96a68fe 100644 (file)
--- a/test/bdd/steps/nominatim_environment.py
+++ b/test/bdd/steps/nominatim_environment.py
@@ -10,6 +10,7 @@ sys.path.insert(1, str((Path(__file__) / '..' / '..' / '..' / '..').resolve()))
 from nominatim import cli
 from nominatim.config import Configuration
 from nominatim.tools import refresh
 from nominatim import cli
 from nominatim.config import Configuration
 from nominatim.tools import refresh

+from nominatim.tokenizer import factory as tokenizer_factory

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


@@ -27,6 +28,7 @@ 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.tokenizer = config['TOKENIZER']

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


@@ -95,6 +97,8 @@ class NominatimEnvironment:
         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())
         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.tokenizer is not None:
+            self.test_env['NOMINATIM_TOKENIZER'] = self.tokenizer

 
         if self.server_module_path:
             self.test_env['NOMINATIM_DATABASE_MODULE_PATH'] = self.server_module_path
 
         if self.server_module_path:
             self.test_env['NOMINATIM_DATABASE_MODULE_PATH'] = self.server_module_path


@@ -179,33 +183,46 @@ class NominatimEnvironment:
         """
         self.write_nominatim_config(self.api_test_db)
 
         """
         self.write_nominatim_config(self.api_test_db)
 

-        if self.api_db_done:
-            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'
+                self.test_env['NOMINATIM_WIKIPEDIA_DATA_PATH'] = str(testdata.resolve())

 
 

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

 
 

-        try:
-            self.run_nominatim('import', '--osm-file', str(self.api_test_file))
-            self.run_nominatim('add-data', '--tiger-data', str((testdata / 'tiger').resolve()))
-            self.run_nominatim('freeze')
+                    if self.tokenizer != 'legacy_icu':
+                        phrase_file = str((testdata / 'specialphrases_testdb.sql').resolve())
+                        run_script(['psql', '-d', self.api_test_db, '-f', phrase_file])
+                    else:
+                        csv_path = str((testdata / 'full_en_phrases_test.csv').resolve())
+                        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.create_tokenizer(self.get_test_config(), init_db=False)

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


@@ -222,13 +239,13 @@ class NominatimEnvironment:
         context.db.autocommit = True
         psycopg2.extras.register_hstore(context.db, globally=False)
 
         context.db.autocommit = True
         psycopg2.extras.register_hstore(context.db, globally=False)
 

-    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):