icu tokenizer: switch to matching against partial names

[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 811faf5cda6a9837a597feeb8a4afd4ebc548912..76f90cfa316af280262a066bbf5da77ccdb3d45d 100644 (file)
--- a/test/bdd/steps/nominatim_environment.py
+++ b/test/bdd/steps/nominatim_environment.py
@@ -7,8 +7,11 @@ import psycopg2.extras
 
 sys.path.insert(1, str((Path(__file__) / '..' / '..' / '..' / '..').resolve()))
 
 
 sys.path.insert(1, str((Path(__file__) / '..' / '..' / '..' / '..').resolve()))
 

+from nominatim import cli

 from nominatim.config import Configuration
 from nominatim.config import Configuration

+from nominatim.db.connection import _Connection

 from nominatim.tools import refresh
 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:


@@ -26,6 +29,7 @@ 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']


@@ -51,7 +55,7 @@ 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)
+        conn = psycopg2.connect(connection_factory=_Connection, **dbargs)

         return conn
 
     def next_code_coverage_file(self):
         return conn
 
     def next_code_coverage_file(self):


@@ -88,25 +92,59 @@ class NominatimEnvironment:
         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_FLATNODE_FILE'] = ''
         self.test_env['NOMINATIM_IMPORT_STYLE'] = 'full'
         self.test_env['NOMINATIM_USE_US_TIGER_DATA'] = 'yes'

-        self.test_env['NOMINATIM_DATADIR'] = self.src_dir / 'data'
-        self.test_env['NOMINATIM_SQLDIR'] = self.src_dir / 'lib-sql'
-        self.test_env['NOMINATIM_CONFIGDIR'] = self.src_dir / 'settings'
-        self.test_env['NOMINATIM_DATABASE_MODULE_SRC_PATH'] = self.build_dir / 'module'
-        self.test_env['NOMINATIM_OSM2PGSQL_BINARY'] = self.build_dir / 'osm2pgsql' / 'osm2pgsql'
-        self.test_env['NOMINATIM_NOMINATIM_TOOL'] = self.build_dir / 'nominatim'
+        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.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
         else:
             # avoid module being copied into the temporary environment
 
         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'] = self.build_dir / 'module'
+            self.test_env['NOMINATIM_DATABASE_MODULE_PATH'] = str((self.build_dir / 'module').resolve())

 
         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)
+
+        try:
+            conn = self.connect_database(dbname)
+        except:
+            conn = False
+        refresh.setup_website(Path(self.website_dir.name) / 'website',
+                              self.get_test_config(), conn)
+
+
+    def get_test_config(self):
+        cfg = Configuration(Path(self.website_dir.name), self.src_dir / 'settings',
+                            environ=self.test_env)
+        cfg.set_libdirs(module=self.build_dir / 'module',
+                        osm2pgsql=self.build_dir / 'osm2pgsql' / 'osm2pgsql',
+                        php=self.src_dir / 'lib-php',
+                        sql=self.src_dir / 'lib-sql',
+                        data=self.src_dir / 'data')
+        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):
 
 
     def db_drop_database(self, name):


@@ -131,34 +169,16 @@ class NominatimEnvironment:
         if self._reuse_or_drop_db(self.template_db):
             return
 
         if self._reuse_or_drop_db(self.template_db):
             return
 

-        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()
+        self.write_nominatim_config(self.template_db)

 
 

-            # execute osm2pgsql import on an empty file to get the right tables
+        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()
             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')
+                self.run_nominatim('import', '--osm-file', fd.name,
+                                             '--osm2pgsql-cache', '1',
+                                             '--ignore-errors')

         except:
             self.db_drop_database(self.template_db)
             raise
         except:
             self.db_drop_database(self.template_db)
             raise


@@ -169,57 +189,68 @@ class NominatimEnvironment:
         """
         self.write_nominatim_config(self.api_test_db)
 
         """
         self.write_nominatim_config(self.api_test_db)
 

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

 
 

-        self.api_db_done = True
+            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())

 
 

-        if self._reuse_or_drop_db(self.api_test_db):
-            return
+                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')

 
 

-        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())
+                    if self.tokenizer != '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

 
 

-        try:
-            self.run_setup_script('all', osm_file=self.api_test_file)
-            self.run_setup_script('import-tiger-data')
-            self.run_setup_script('drop')
-
-            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.
         """
         self.setup_template_db()
 
     def setup_db(self, context):
         """ Setup a test against a fresh, empty test database.
         """
         self.setup_template_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()
         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()

+        self.write_nominatim_config(self.test_db)

         context.db = self.connect_database(self.test_db)
         context.db.autocommit = True
         psycopg2.extras.register_hstore(context.db, globally=False)
 
         context.db = self.connect_database(self.test_db)
         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):


@@ -249,43 +280,34 @@ class NominatimEnvironment:
         """
         with db.cursor() as cur:
             while True:
         """
         with db.cursor() as cur:
             while True:

-                self.run_update_script('index')
+                self.run_nominatim('index')

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

-    def run_setup_script(self, *args, **kwargs):
-        """ Run the Nominatim setup script with the given arguments.
+    def run_nominatim(self, *cmdline):
+        """ Run the nominatim command-line tool via the library.

         """
         """

-        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:
         if self.website_dir is not None:

-            cwd = self.website_dir.name
-        else:
-            cwd = None
+            cmdline = list(cmdline) + ['--project-dir', self.website_dir.name]
+
+        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'),
+                      cli_args=cmdline,
+                      phpcgi_path='',
+                      environ=self.test_env)

 
 

-        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
             tables invoking the appropriate triggers.
         """
 
     def copy_from_place(self, db):
         """ Copy data from place to the placex and location_property_osmline
             tables invoking the appropriate triggers.
         """

-        self.run_setup_script('create-functions', 'create-partition-functions')
+        self.run_nominatim('refresh', '--functions', '--no-diff-updates')

 
         with db.cursor() as cur:
             cur.execute("""INSERT INTO placex (osm_type, osm_id, class, type,
 
         with db.cursor() as cur:
             cur.execute("""INSERT INTO placex (osm_type, osm_id, class, type,