use word_token length when penalizing against postcodes

[nominatim.git] / src / nominatim_db / cli.py

diff --git a/src/nominatim_db/cli.py b/src/nominatim_db/cli.py
index 9a54f338e0f61daecf8fe9e0de0a85a9d53f48f0..8d4585c67b36028f875a6b71bf01566750ec42a2 100644 (file)
--- a/src/nominatim_db/cli.py
+++ b/src/nominatim_db/cli.py
@@ -2,16 +2,15 @@
 #
 # This file is part of Nominatim. (https://nominatim.org)
 #
-# Copyright (C) 2024 by the Nominatim developer community.
+# Copyright (C) 2025 by the Nominatim developer community.
 # For a full list of authors see the git log.
 """
 Command-line interface to the Nominatim functions for import, update,
 database administration and querying.
 """
-from typing import Optional, Any
+from typing import Optional, List, Mapping
 import importlib
 import logging
-import os
 import sys
 import argparse
 import asyncio
@@ -25,6 +24,7 @@ from .clicmd.args import NominatimArgs, Subcommand
 
 LOG = logging.getLogger()
 
+
 class CommandlineParser:
     """ Wraps some of the common functions for parsing the command line
         and setting up subcommands.
@@ -57,7 +57,6 @@ class CommandlineParser:
         group.add_argument('-j', '--threads', metavar='NUM', type=int,
                            help='Number of parallel threads to use')
 
-
     def nominatim_version_text(self) -> str:
         """ Program name and version number as string
         """
@@ -66,7 +65,6 @@ class CommandlineParser:
             text += f' ({version.GIT_COMMIT_HASH})'
         return text
 
-
     def add_subcommand(self, name: str, cmd: Subcommand) -> None:
         """ Add a subcommand to the parser. The subcommand must be a class
             with a function add_args() that adds the parameters for the
@@ -82,14 +80,14 @@ class CommandlineParser:
         parser.set_defaults(command=cmd)
         cmd.add_args(parser)
 
-
-    def run(self, **kwargs: Any) -> int:
+    def run(self, cli_args: Optional[List[str]],
+            environ: Optional[Mapping[str, str]]) -> int:
         """ Parse the command line arguments of the program and execute the
             appropriate subcommand.
         """
         args = NominatimArgs()
         try:
-            self.parser.parse_args(args=kwargs.get('cli_args'), namespace=args)
+            self.parser.parse_args(args=cli_args, namespace=args)
         except SystemExit:
             return 1
 
@@ -103,31 +101,22 @@ class CommandlineParser:
 
         args.project_dir = Path(args.project_dir).resolve()
 
-        if 'cli_args' not in kwargs:
+        if cli_args is None:
             logging.basicConfig(stream=sys.stderr,
                                 format='%(asctime)s: %(message)s',
                                 datefmt='%Y-%m-%d %H:%M:%S',
                                 level=max(4 - args.verbose, 1) * 10)
 
-        args.config = Configuration(args.project_dir,
-                                    environ=kwargs.get('environ', os.environ))
-        args.config.set_libdirs(module=kwargs['module_dir'],
-                                osm2pgsql=kwargs['osm2pgsql_path'])
+        args.config = Configuration(args.project_dir, environ=environ)
 
         log = logging.getLogger()
         log.warning('Using project directory: %s', str(args.project_dir))
 
         try:
-            ret = args.command.run(args)
-
-            if args.config.TOKENIZER == 'legacy':
-                log.warning('WARNING: the "legacy" tokenizer is deprecated '
-                            'and will be removed in Nominatim 5.0.')
-
-            return ret
+            return args.command.run(args)
         except UsageError as exception:
             if log.isEnabledFor(logging.DEBUG):
-                raise # use Python's exception printing
+                raise  # use Python's exception printing
             log.fatal('FATAL: %s', exception)
 
         # If we get here, then execution has failed in some way.
@@ -144,7 +133,6 @@ class CommandlineParser:
 # a subcommand.
 #
 # No need to document the functions each time.
-# pylint: disable=C0111
 class AdminServe:
     """\
     Start a simple web server for serving the API.
@@ -169,15 +157,13 @@ class AdminServe:
                            choices=('falcon', 'starlette'),
                            help='Webserver framework to run. (default: falcon)')
 
-
     def run(self, args: NominatimArgs) -> int:
         asyncio.run(self.run_uvicorn(args))
 
         return 0
 
-
     async def run_uvicorn(self, args: NominatimArgs) -> None:
-        import uvicorn # pylint: disable=import-outside-toplevel
+        import uvicorn
 
         server_info = args.server.split(':', 1)
         host = server_info[0]
@@ -231,7 +217,7 @@ def get_set_parser() -> CommandlineParser:
         parser.add_subcommand('details', apicmd.APIDetails())
         parser.add_subcommand('status', apicmd.APIStatus())
     except ModuleNotFoundError as ex:
-        if not ex.name or 'nominatim_api' not in ex.name: # pylint: disable=E1135
+        if not ex.name or 'nominatim_api' not in ex.name:
             raise ex
 
         parser.parser.epilog = \
@@ -240,13 +226,19 @@ def get_set_parser() -> CommandlineParser:
             '\n    export, convert, serve, search, reverse, lookup, details, status'\
             "\n\nRun 'pip install nominatim-api' to install the package."
 
-
     return parser
 
 
-def nominatim(**kwargs: Any) -> int:
+def nominatim(cli_args: Optional[List[str]] = None,
+              environ: Optional[Mapping[str, str]] = None) -> int:
     """\
     Command-line tools for importing, updating, administrating and
     querying the Nominatim database.
+
+    'cli_args' is a list of parameters for the command to run. If not given,
+    sys.args will be used.
+
+    'environ' is the dictionary of environment variables containing the
+    Nominatim configuration. When None, the os.environ is inherited.
     """
-    return get_set_parser().run(**kwargs)
+    return get_set_parser().run(cli_args=cli_args, environ=environ)