]> git.openstreetmap.org Git - nominatim.git/blobdiff - nominatim/cli.py
reintroduce cutoffs when searching for very frequent words
[nominatim.git] / nominatim / cli.py
index 0bca03a31959b566c3d6b08137eac3dcf252a92a..720a8ece33e8e126422490735cdd7c055653ea4c 100644 (file)
@@ -1,26 +1,35 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2023 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
+import importlib
 import logging
 import os
 import sys
 import argparse
 from pathlib import Path
 
-from .config import Configuration
-from .tools.exec_utils import run_legacy_script, run_php_server
-from .errors import UsageError
-from . import clicmd
+from nominatim.config import Configuration
+from nominatim.tools.exec_utils import run_php_server
+from nominatim.errors import UsageError
+from nominatim import clicmd
+from nominatim import version
+from nominatim.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.
     """
-    def __init__(self, prog, description):
+    def __init__(self, prog: str, description: Optional[str]):
         self.parser = argparse.ArgumentParser(
             prog=prog,
             description=description,
@@ -29,6 +38,10 @@ class CommandlineParser:
         self.subs = self.parser.add_subparsers(title='available commands',
                                                dest='subcommand')
 
+        # Global arguments that only work if no sub-command given
+        self.parser.add_argument('--version', action='store_true',
+                                 help='Print Nominatim version and exit')
+
         # Arguments added to every sub-command
         self.default_args = argparse.ArgumentParser(add_help=False)
         group = self.default_args.add_argument_group('Default arguments')
@@ -45,11 +58,22 @@ class CommandlineParser:
                            help='Number of parallel threads to use')
 
 
-    def add_subcommand(self, name, cmd):
+    def nominatim_version_text(self) -> str:
+        """ Program name and version number as string
+        """
+        text = f'Nominatim version {version.NOMINATIM_VERSION!s}'
+        if version.GIT_COMMIT_HASH is not None:
+            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
             subcommand and a run() function that executes the command.
         """
+        assert cmd.__doc__ is not None
+
         parser = self.subs.add_parser(name, parents=[self.default_args],
                                       help=cmd.__doc__.split('\n', 1)[0],
                                       description=cmd.__doc__,
@@ -58,26 +82,37 @@ class CommandlineParser:
         parser.set_defaults(command=cmd)
         cmd.add_args(parser)
 
-    def run(self, **kwargs):
+
+    def run(self, **kwargs: Any) -> int:
         """ Parse the command line arguments of the program and execute the
             appropriate subcommand.
         """
-        args = self.parser.parse_args(args=kwargs.get('cli_args'))
+        args = NominatimArgs()
+        try:
+            self.parser.parse_args(args=kwargs.get('cli_args'), namespace=args)
+        except SystemExit:
+            return 1
+
+        if args.version:
+            print(self.nominatim_version_text())
+            return 0
 
         if args.subcommand is None:
             self.parser.print_help()
             return 1
 
-        for arg in ('module_dir', 'osm2pgsql_path', 'phplib_dir', 'data_dir', 'phpcgi_path'):
-            setattr(args, arg, Path(kwargs[arg]))
         args.project_dir = Path(args.project_dir).resolve()
 
-        logging.basicConfig(stream=sys.stderr,
-                            format='%(asctime)s: %(message)s',
-                            datefmt='%Y-%m-%d %H:%M:%S',
-                            level=max(4 - args.verbose, 1) * 10)
+        if 'cli_args' not in kwargs:
+            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, args.data_dir / 'settings')
+        args.config = Configuration(args.project_dir,
+                                    environ=kwargs.get('environ', os.environ))
+        args.config.set_libdirs(module=kwargs['module_dir'],
+                                osm2pgsql=kwargs['osm2pgsql_path'])
 
         log = logging.getLogger()
         log.warning('Using project directory: %s', str(args.project_dir))
@@ -93,7 +128,7 @@ class CommandlineParser:
         return 1
 
 
-##### Subcommand classes
+# Subcommand classes
 #
 # Each class needs to implement two functions: add_args() adds the CLI parameters
 # for the subfunction, run() executes the subcommand.
@@ -104,314 +139,90 @@ class CommandlineParser:
 #
 # No need to document the functions each time.
 # pylint: disable=C0111
-# Using non-top-level imports to make pyosmium optional for replication only.
-# pylint: disable=E0012,C0415
-
-
-class SetupAll:
-    """\
-    Create a new Nominatim database from an OSM file.
-    """
-
-    @staticmethod
-    def add_args(parser):
-        group_name = parser.add_argument_group('Required arguments')
-        group = group_name.add_mutually_exclusive_group(required=True)
-        group.add_argument('--osm-file',
-                           help='OSM file to be imported.')
-        group.add_argument('--continue', dest='continue_at',
-                           choices=['load-data', 'indexing', 'db-postprocess'],
-                           help='Continue an import that was interrupted')
-        group = parser.add_argument_group('Optional arguments')
-        group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
-                           help='Size of cache to be used by osm2pgsql (in MB)')
-        group.add_argument('--reverse-only', action='store_true',
-                           help='Do not create tables and indexes for searching')
-        group.add_argument('--enable-debug-statements', action='store_true',
-                           help='Include debug warning statements in SQL code')
-        group.add_argument('--no-partitions', action='store_true',
-                           help="""Do not partition search indices
-                                   (speeds up import of single country extracts)""")
-        group.add_argument('--no-updates', action='store_true',
-                           help="""Do not keep tables that are only needed for
-                                   updating the database later""")
-        group = parser.add_argument_group('Expert options')
-        group.add_argument('--ignore-errors', action='store_true',
-                           help='Continue import even when errors in SQL are present')
-        group.add_argument('--index-noanalyse', action='store_true',
-                           help='Do not perform analyse operations during index')
-
-
-    @staticmethod
-    def run(args):
-        params = ['setup.php']
-        if args.osm_file:
-            params.extend(('--all', '--osm-file', args.osm_file))
-        else:
-            if args.continue_at == 'load-data':
-                params.append('--load-data')
-            if args.continue_at in ('load-data', 'indexing'):
-                params.append('--index')
-            params.extend(('--create-search-indices', '--create-country-names',
-                           '--setup-website'))
-        if args.osm2pgsql_cache:
-            params.extend(('--osm2pgsql-cache', args.osm2pgsql_cache))
-        if args.reverse_only:
-            params.append('--reverse-only')
-        if args.enable_debug_statements:
-            params.append('--enable-debug-statements')
-        if args.no_partitions:
-            params.append('--no-partitions')
-        if args.no_updates:
-            params.append('--drop')
-        if args.ignore_errors:
-            params.append('--ignore-errors')
-        if args.index_noanalyse:
-            params.append('--index-noanalyse')
-
-        return run_legacy_script(*params, nominatim_env=args)
-
-
-class SetupFreeze:
+class AdminServe:
     """\
-    Make database read-only.
-
-    About half of data in the Nominatim database is kept only to be able to
-    keep the data up-to-date with new changes made in OpenStreetMap. This
-    command drops all this data and only keeps the part needed for geocoding
-    itself.
-
-    This command has the same effect as the `--no-updates` option for imports.
-    """
-
-    @staticmethod
-    def add_args(parser):
-        pass # No options
+    Start a simple web server for serving the API.
 
-    @staticmethod
-    def run(args):
-        return run_legacy_script('setup.php', '--drop', nominatim_env=args)
+    This command starts a built-in webserver to serve the website
+    from the current project directory. This webserver is only suitable
+    for testing and development. Do not use it in production setups!
 
+    There are different webservers available. The default 'php' engine
+    runs the classic PHP frontend. The other engines are Python servers
+    which run the new Python frontend code. This is highly experimental
+    at the moment and may not include the full API.
 
-class SetupSpecialPhrases:
-    """\
-    Maintain special phrases.
+    By the default, the webserver can be accessed at: http://127.0.0.1:8088
     """
 
-    @staticmethod
-    def add_args(parser):
-        group = parser.add_argument_group('Input arguments')
-        group.add_argument('--from-wiki', action='store_true',
-                           help='Pull special phrases from the OSM wiki.')
-        group = parser.add_argument_group('Output arguments')
-        group.add_argument('-o', '--output', default='-',
-                           help="""File to write the preprocessed phrases to.
-                                   If omitted, it will be written to stdout.""")
-
-    @staticmethod
-    def run(args):
-        if args.output != '-':
-            raise NotImplementedError('Only output to stdout is currently implemented.')
-        return run_legacy_script('specialphrases.php', '--wiki-import', nominatim_env=args)
-
+    def add_args(self, parser: argparse.ArgumentParser) -> None:
+        group = parser.add_argument_group('Server arguments')
+        group.add_argument('--server', default='127.0.0.1:8088',
+                           help='The address the server will listen to.')
+        group.add_argument('--engine', default='falcon',
+                           choices=('php', 'falcon', 'starlette'),
+                           help='Webserver framework to run. (default: falcon)')
 
-class UpdateAddData:
-    """\
-    Add additional data from a file or an online source.
 
-    Data is only imported, not indexed. You need to call `nominatim-update index`
-    to complete the process.
-    """
+    def run(self, args: NominatimArgs) -> int:
+        if args.engine == 'php':
+            if args.config.lib_dir.php is None:
+                raise UsageError("PHP frontend not configured.")
+            run_php_server(args.server, args.project_dir / 'website')
+        else:
+            import uvicorn # pylint: disable=import-outside-toplevel
+            server_info = args.server.split(':', 1)
+            host = server_info[0]
+            if len(server_info) > 1:
+                if not server_info[1].isdigit():
+                    raise UsageError('Invalid format for --server parameter. Use <host>:<port>')
+                port = int(server_info[1])
+            else:
+                port = 8088
 
-    @staticmethod
-    def add_args(parser):
-        group_name = parser.add_argument_group('Source')
-        group = group_name.add_mutually_exclusive_group(required=True)
-        group.add_argument('--file', metavar='FILE',
-                           help='Import data from an OSM file')
-        group.add_argument('--diff', metavar='FILE',
-                           help='Import data from an OSM diff file')
-        group.add_argument('--node', metavar='ID', type=int,
-                           help='Import a single node from the API')
-        group.add_argument('--way', metavar='ID', type=int,
-                           help='Import a single way from the API')
-        group.add_argument('--relation', metavar='ID', type=int,
-                           help='Import a single relation from the API')
-        group.add_argument('--tiger-data', metavar='DIR',
-                           help='Add housenumbers from the US TIGER census database.')
-        group = parser.add_argument_group('Extra arguments')
-        group.add_argument('--use-main-api', action='store_true',
-                           help='Use OSM API instead of Overpass to download objects')
-
-    @staticmethod
-    def run(args):
-        if args.tiger_data:
-            os.environ['NOMINATIM_TIGER_DATA_PATH'] = args.tiger_data
-            return run_legacy_script('setup.php', '--import-tiger-data', nominatim_env=args)
-
-        params = ['update.php']
-        if args.file:
-            params.extend(('--import-file', args.file))
-        elif args.diff:
-            params.extend(('--import-diff', args.diff))
-        elif args.node:
-            params.extend(('--import-node', args.node))
-        elif args.way:
-            params.extend(('--import-way', args.way))
-        elif args.relation:
-            params.extend(('--import-relation', args.relation))
-        if args.use_main_api:
-            params.append('--use-main-api')
-        return run_legacy_script(*params, nominatim_env=args)
-
-
-class AdminCheckDatabase:
-    """\
-    Check that the database is complete and operational.
-    """
+            server_module = importlib.import_module(f'nominatim.server.{args.engine}.server')
 
-    @staticmethod
-    def add_args(parser):
-        pass # No options
+            app = server_module.get_application(args.project_dir)
+            uvicorn.run(app, host=host, port=port)
 
-    @staticmethod
-    def run(args):
-        return run_legacy_script('check_import_finished.php', nominatim_env=args)
+        return 0
 
 
-class AdminWarm:
+def get_set_parser() -> CommandlineParser:
     """\
-    Warm database caches for search and reverse queries.
+    Initializes the parser and adds various subcommands for
+    nominatim cli.
     """
+    parser = CommandlineParser('nominatim', nominatim.__doc__)
 
-    @staticmethod
-    def add_args(parser):
-        group = parser.add_argument_group('Target arguments')
-        group.add_argument('--search-only', action='store_const', dest='target',
-                           const='search',
-                           help="Only pre-warm tables for search queries")
-        group.add_argument('--reverse-only', action='store_const', dest='target',
-                           const='reverse',
-                           help="Only pre-warm tables for reverse queries")
-
-    @staticmethod
-    def run(args):
-        params = ['warm.php']
-        if args.target == 'reverse':
-            params.append('--reverse-only')
-        if args.target == 'search':
-            params.append('--search-only')
-        return run_legacy_script(*params, nominatim_env=args)
-
-
-class QueryExport:
-    """\
-    Export addresses as CSV file from the database.
-    """
+    parser.add_subcommand('import', clicmd.SetupAll())
+    parser.add_subcommand('freeze', clicmd.SetupFreeze())
+    parser.add_subcommand('replication', clicmd.UpdateReplication())
 
-    @staticmethod
-    def add_args(parser):
-        group = parser.add_argument_group('Output arguments')
-        group.add_argument('--output-type', default='street',
-                           choices=('continent', 'country', 'state', 'county',
-                                    'city', 'suburb', 'street', 'path'),
-                           help='Type of places to output (default: street)')
-        group.add_argument('--output-format',
-                           default='street;suburb;city;county;state;country',
-                           help="""Semicolon-separated list of address types
-                                   (see --output-type). Multiple ranks can be
-                                   merged into one column by simply using a
-                                   comma-separated list.""")
-        group.add_argument('--output-all-postcodes', action='store_true',
-                           help="""List all postcodes for address instead of
-                                   just the most likely one""")
-        group.add_argument('--language',
-                           help="""Preferred language for output
-                                   (use local name, if omitted)""")
-        group = parser.add_argument_group('Filter arguments')
-        group.add_argument('--restrict-to-country', metavar='COUNTRY_CODE',
-                           help='Export only objects within country')
-        group.add_argument('--restrict-to-osm-node', metavar='ID', type=int,
-                           help='Export only children of this OSM node')
-        group.add_argument('--restrict-to-osm-way', metavar='ID', type=int,
-                           help='Export only children of this OSM way')
-        group.add_argument('--restrict-to-osm-relation', metavar='ID', type=int,
-                           help='Export only children of this OSM relation')
-
-
-    @staticmethod
-    def run(args):
-        params = ['export.php',
-                  '--output-type', args.output_type,
-                  '--output-format', args.output_format]
-        if args.output_all_postcodes:
-            params.append('--output-all-postcodes')
-        if args.language:
-            params.extend(('--language', args.language))
-        if args.restrict_to_country:
-            params.extend(('--restrict-to-country', args.restrict_to_country))
-        if args.restrict_to_osm_node:
-            params.extend(('--restrict-to-osm-node', args.restrict_to_osm_node))
-        if args.restrict_to_osm_way:
-            params.extend(('--restrict-to-osm-way', args.restrict_to_osm_way))
-        if args.restrict_to_osm_relation:
-            params.extend(('--restrict-to-osm-relation', args.restrict_to_osm_relation))
-
-        return run_legacy_script(*params, nominatim_env=args)
+    parser.add_subcommand('special-phrases', clicmd.ImportSpecialPhrases())
 
+    parser.add_subcommand('add-data', clicmd.UpdateAddData())
+    parser.add_subcommand('index', clicmd.UpdateIndex())
+    parser.add_subcommand('refresh', clicmd.UpdateRefresh())
 
-class AdminServe:
-    """\
-    Start a simple web server for serving the API.
+    parser.add_subcommand('admin', clicmd.AdminFuncs())
 
-    This command starts the built-in PHP webserver to serve the website
-    from the current project directory. This webserver is only suitable
-    for testing and develop. Do not use it in production setups!
+    parser.add_subcommand('export', clicmd.QueryExport())
+    parser.add_subcommand('convert', clicmd.ConvertDB())
+    parser.add_subcommand('serve', AdminServe())
 
-    By the default, the webserver can be accessed at: http://127.0.0.1:8088
-    """
+    parser.add_subcommand('search', clicmd.APISearch())
+    parser.add_subcommand('reverse', clicmd.APIReverse())
+    parser.add_subcommand('lookup', clicmd.APILookup())
+    parser.add_subcommand('details', clicmd.APIDetails())
+    parser.add_subcommand('status', clicmd.APIStatus())
 
-    @staticmethod
-    def add_args(parser):
-        group = parser.add_argument_group('Server arguments')
-        group.add_argument('--server', default='127.0.0.1:8088',
-                           help='The address the server will listen to.')
-
-    @staticmethod
-    def run(args):
-        run_php_server(args.server, args.project_dir / 'website')
+    return parser
 
 
-def nominatim(**kwargs):
+def nominatim(**kwargs: Any) -> int:
     """\
     Command-line tools for importing, updating, administrating and
     querying the Nominatim database.
     """
-    parser = CommandlineParser('nominatim', nominatim.__doc__)
-
-    parser.add_subcommand('import', SetupAll)
-    parser.add_subcommand('freeze', SetupFreeze)
-    parser.add_subcommand('replication', clicmd.UpdateReplication)
-
-    parser.add_subcommand('check-database', AdminCheckDatabase)
-    parser.add_subcommand('warm', AdminWarm)
-
-    parser.add_subcommand('special-phrases', SetupSpecialPhrases)
-
-    parser.add_subcommand('add-data', UpdateAddData)
-    parser.add_subcommand('index', clicmd.UpdateIndex)
-    parser.add_subcommand('refresh', clicmd.UpdateRefresh)
-
-    parser.add_subcommand('export', QueryExport)
-    parser.add_subcommand('serve', AdminServe)
-
-    if kwargs.get('phpcgi_path'):
-        parser.add_subcommand('search', clicmd.APISearch)
-        parser.add_subcommand('reverse', clicmd.APIReverse)
-        parser.add_subcommand('lookup', clicmd.APILookup)
-        parser.add_subcommand('details', clicmd.APIDetails)
-        parser.add_subcommand('status', clicmd.APIStatus)
-    else:
-        parser.parser.epilog = 'php-cgi not found. Query commands not available.'
-
-    return parser.run(**kwargs)
+    return get_set_parser().run(**kwargs)