X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/98dbc84836dd834dd41cb4ada675456746e872f1..a3767f9142d19970371c1dfd3bdb624fc94d3c58:/nominatim/cli.py diff --git a/nominatim/cli.py b/nominatim/cli.py index 9d1e36d1..70481b4a 100644 --- a/nominatim/cli.py +++ b/nominatim/cli.py @@ -3,12 +3,24 @@ Command-line interface to the Nominatim functions for import, update, database administration and querying. """ import sys +import os import argparse import logging from pathlib import Path from .config import Configuration -from .admin.exec_utils import run_legacy_script +from .tools.exec_utils import run_legacy_script, run_api_script + +from .indexer.indexer import Indexer + +def _num_system_cpus(): + try: + cpus = len(os.sched_getaffinity(0)) + except NotImplementedError: + cpus = None + + return cpus or os.cpu_count() + class CommandlineParser: """ Wraps some of the common functions for parsing the command line @@ -61,18 +73,30 @@ class CommandlineParser: if args.subcommand is None: return self.parser.print_help() - for arg in ('module_dir', 'osm2pgsql_path', 'phplib_dir', 'data_dir'): + 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) logging.basicConfig(stream=sys.stderr, - format='%(asctime)s %(levelname)s: %(message)s', + 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.command.run(args) + return args.command.run(args) + +##### Subcommand classes +# +# Each class needs to implement two functions: add_args() adds the CLI parameters +# for the subfunction, run() executes the subcommand. +# +# The class documentation doubles as the help text for the command. The +# first line is also used in the summary when calling the program without +# a subcommand. +# +# No need to document the functions each time. +# pylint: disable=C0111 class SetupAll: @@ -86,7 +110,7 @@ class SetupAll: group = group_name.add_mutually_exclusive_group(required=True) group.add_argument('--osm-file', help='OSM file to be imported.') - group.add_argument('--continue', nargs=1, dest='continue_at', + 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') @@ -180,7 +204,7 @@ class SetupSpecialPhrases: def run(args): if args.output.name != '': raise NotImplementedError('Only output to stdout is currently implemented.') - return run_legacy_script('specialphrases.php', '--wiki-import' , nominatim_env=args) + return run_legacy_script('specialphrases.php', '--wiki-import', nominatim_env=args) class UpdateReplication: @@ -213,7 +237,7 @@ class UpdateReplication: if args.init: params.append('--init-updates') if not args.update_functions: - params.apend('--no-update-functions') + params.append('--no-update-functions') elif args.check_for_updates: params.append('--check-for-updates') else: @@ -258,9 +282,10 @@ class UpdateAddData: @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'] + params = ['update.php'] if args.file: params.extend(('--import-file', args.file)) elif args.diff: @@ -270,7 +295,7 @@ class UpdateAddData: elif args.way: params.extend(('--import-way', args.way)) elif args.relation: - params.extend(('--import-relation' , 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) @@ -283,16 +308,35 @@ class UpdateIndex: @staticmethod def add_args(parser): - pass + group = parser.add_argument_group('Filter arguments') + group.add_argument('--boundaries-only', action='store_true', + help="""Index only administrative boundaries.""") + group.add_argument('--no-boundaries', action='store_true', + help="""Index everything except administrative boundaries.""") + group.add_argument('--minrank', '-r', type=int, metavar='RANK', default=0, + help='Minimum/starting rank') + group.add_argument('--maxrank', '-R', type=int, metavar='RANK', default=30, + help='Maximum/finishing rank') @staticmethod def run(args): - return run_legacy_script('update.php', '--index', nominatim_env=args) + indexer = Indexer(args.config.get_libpq_dsn(), + args.threads or _num_system_cpus() or 1) + + if not args.no_boundaries: + indexer.index_boundaries(args.minrank, args.maxrank) + if not args.boundaries_only: + indexer.index_by_rank(args.minrank, args.maxrank) + + if not args.no_boundaries and not args.boundaries_only: + indexer.update_status_table() + + return 0 class UpdateRefresh: """\ - Recompute auxillary data used by the indexing process. + Recompute auxiliary data used by the indexing process. These functions must not be run in parallel with other update commands. """ @@ -306,12 +350,12 @@ class UpdateRefresh: help='Compute frequency of full-word search terms') group.add_argument('--address-levels', action='store_true', help='Reimport address level configuration') - group.add_argument('--importance', action='store_true', - help='Recompute place importances (expensive!)') group.add_argument('--functions', action='store_true', help='Update the PL/pgSQL functions in the database') group.add_argument('--wiki-data', action='store_true', help='Update Wikipedia/data importance numbers.') + group.add_argument('--importance', action='store_true', + help='Recompute place importances (expensive!)') group.add_argument('--website', action='store_true', help='Refresh the directory that serves the scripts for the web API') group = parser.add_argument_group('Arguments for function refresh') @@ -331,9 +375,6 @@ class UpdateRefresh: if args.address_levels: run_legacy_script('update.php', '--update-address-levels', nominatim_env=args, throw_on_fail=True) - if args.importance: - run_legacy_script('update.php', '--recompute-importance', - nominatim_env=args, throw_on_fail=True) if args.functions: params = ['setup.php', '--create-functions', '--create-partition-functions'] if args.diffs: @@ -344,6 +385,10 @@ class UpdateRefresh: if args.wiki_data: run_legacy_script('setup.php', '--import-wikipedia-articles', nominatim_env=args, throw_on_fail=True) + # Attention: importance MUST come after wiki data import. + if args.importance: + run_legacy_script('update.php', '--recompute-importance', + nominatim_env=args, throw_on_fail=True) if args.website: run_legacy_script('setup.php', '--setup-website', nominatim_env=args, throw_on_fail=True) @@ -390,7 +435,7 @@ class AdminWarm: class QueryExport: """\ - Export addresses as CSV file from a Nominatim database. + Export addresses as CSV file from the database. """ @staticmethod @@ -435,24 +480,277 @@ class QueryExport: if args.restrict_to_country: params.extend(('--restrict-to-country', args.restrict_to_country)) if args.restrict_to_osm_node: - params.exted(('--restrict-to-osm-node', args.restrict_to_osm_node)) + params.extend(('--restrict-to-osm-node', args.restrict_to_osm_node)) if args.restrict_to_osm_way: - params.exted(('--restrict-to-osm-way', args.restrict_to_osm_way)) + params.extend(('--restrict-to-osm-way', args.restrict_to_osm_way)) if args.restrict_to_osm_relation: - params.exted(('--restrict-to-osm-relation', args.restrict_to_osm_relation)) + params.extend(('--restrict-to-osm-relation', args.restrict_to_osm_relation)) return run_legacy_script(*params, nominatim_env=args) -class QueryTodo: +STRUCTURED_QUERY = ( + ('street', 'housenumber and street'), + ('city', 'city, town or village'), + ('county', 'county'), + ('state', 'state'), + ('country', 'country'), + ('postalcode', 'postcode') +) + +EXTRADATA_PARAMS = ( + ('addressdetails', 'Include a breakdown of the address into elements.'), + ('extratags', """Include additional information if available + (e.g. wikipedia link, opening hours)."""), + ('namedetails', 'Include a list of alternative names.') +) + +DETAILS_SWITCHES = ( + ('addressdetails', 'Include a breakdown of the address into elements.'), + ('keywords', 'Include a list of name keywords and address keywords.'), + ('linkedplaces', 'Include a details of places that are linked with this one.'), + ('hierarchy', 'Include details of places lower in the address hierarchy.'), + ('group_hierarchy', 'Group the places by type.'), + ('polygon_geojson', 'Include geometry of result.') +) + +class APISearch: """\ - Todo + Execute API search query. """ + @staticmethod def add_args(parser): - pass + group = parser.add_argument_group('Query arguments') + group.add_argument('--query', + help='Free-form query string') + for name, desc in STRUCTURED_QUERY: + group.add_argument('--' + name, help='Structured query: ' + desc) + + group = parser.add_argument_group('Output arguments') + group.add_argument('--format', default='jsonv2', + choices=['xml', 'json', 'jsonv2', 'geojson', 'geocodejson'], + help='Format of result') + for name, desc in EXTRADATA_PARAMS: + group.add_argument('--' + name, action='store_true', help=desc) + + group.add_argument('--lang', '--accept-language', metavar='LANGS', + help='Preferred language order for presenting search results') + group.add_argument('--polygon-output', + choices=['geojson', 'kml', 'svg', 'text'], + help='Output geometry of results as a GeoJSON, KML, SVG or WKT.') + group.add_argument('--polygon-threshold', type=float, metavar='TOLERANCE', + help="""Simplify output geometry. + Parameter is difference tolerance in degrees.""") + + group = parser.add_argument_group('Result limitation') + group.add_argument('--countrycodes', metavar='CC,..', + help='Limit search results to one or more countries.') + group.add_argument('--exclude_place_ids', metavar='ID,..', + help='List of search object to be excluded') + group.add_argument('--limit', type=int, + help='Limit the number of returned results') + group.add_argument('--viewbox', metavar='X1,Y1,X2,Y2', + help='Preferred area to find search results') + group.add_argument('--bounded', action='store_true', + help='Strictly restrict results to viewbox area') + + group = parser.add_argument_group('Other arguments') + group.add_argument('--no-dedupe', action='store_false', dest='dedupe', + help='Do not remove duplicates from the result list') + + @staticmethod def run(args): - print("TODO: searching") + if args.query: + params = dict(q=args.query) + else: + params = {k : getattr(args, k) for k, _ in STRUCTURED_QUERY if getattr(args, k)} + + for param, _ in EXTRADATA_PARAMS: + if getattr(args, param): + params[param] = '1' + for param in ('format', 'countrycodes', 'exclude_place_ids', 'limit', 'viewbox'): + if getattr(args, param): + params[param] = getattr(args, param) + if args.lang: + params['accept-language'] = args.lang + if args.polygon_output: + params['polygon_' + args.polygon_output] = '1' + if args.polygon_threshold: + params['polygon_threshold'] = args.polygon_threshold + if args.bounded: + params['bounded'] = '1' + if not args.dedupe: + params['dedupe'] = '0' + + return run_api_script('search', args.project_dir, + phpcgi_bin=args.phpcgi_path, params=params) + +class APIReverse: + """\ + Execute API reverse query. + """ + + @staticmethod + def add_args(parser): + group = parser.add_argument_group('Query arguments') + group.add_argument('--lat', type=float, required=True, + help='Latitude of coordinate to look up (in WGS84)') + group.add_argument('--lon', type=float, required=True, + help='Longitude of coordinate to look up (in WGS84)') + group.add_argument('--zoom', type=int, + help='Level of detail required for the address') + + group = parser.add_argument_group('Output arguments') + group.add_argument('--format', default='jsonv2', + choices=['xml', 'json', 'jsonv2', 'geojson', 'geocodejson'], + help='Format of result') + for name, desc in EXTRADATA_PARAMS: + group.add_argument('--' + name, action='store_true', help=desc) + + group.add_argument('--lang', '--accept-language', metavar='LANGS', + help='Preferred language order for presenting search results') + group.add_argument('--polygon-output', + choices=['geojson', 'kml', 'svg', 'text'], + help='Output geometry of results as a GeoJSON, KML, SVG or WKT.') + group.add_argument('--polygon-threshold', type=float, metavar='TOLERANCE', + help="""Simplify output geometry. + Parameter is difference tolerance in degrees.""") + + + @staticmethod + def run(args): + params = dict(lat=args.lat, lon=args.lon) + if args.zoom is not None: + params['zoom'] = args.zoom + + for param, _ in EXTRADATA_PARAMS: + if getattr(args, param): + params[param] = '1' + if args.format: + params['format'] = args.format + if args.lang: + params['accept-language'] = args.lang + if args.polygon_output: + params['polygon_' + args.polygon_output] = '1' + if args.polygon_threshold: + params['polygon_threshold'] = args.polygon_threshold + + return run_api_script('reverse', args.project_dir, + phpcgi_bin=args.phpcgi_path, params=params) + + +class APILookup: + """\ + Execute API reverse query. + """ + + @staticmethod + def add_args(parser): + group = parser.add_argument_group('Query arguments') + group.add_argument('--id', metavar='OSMID', + action='append', required=True, dest='ids', + help='OSM id to lookup in format (may be repeated)') + + group = parser.add_argument_group('Output arguments') + group.add_argument('--format', default='jsonv2', + choices=['xml', 'json', 'jsonv2', 'geojson', 'geocodejson'], + help='Format of result') + for name, desc in EXTRADATA_PARAMS: + group.add_argument('--' + name, action='store_true', help=desc) + + group.add_argument('--lang', '--accept-language', metavar='LANGS', + help='Preferred language order for presenting search results') + group.add_argument('--polygon-output', + choices=['geojson', 'kml', 'svg', 'text'], + help='Output geometry of results as a GeoJSON, KML, SVG or WKT.') + group.add_argument('--polygon-threshold', type=float, metavar='TOLERANCE', + help="""Simplify output geometry. + Parameter is difference tolerance in degrees.""") + + + @staticmethod + def run(args): + params = dict(osm_ids=','.join(args.ids)) + + for param, _ in EXTRADATA_PARAMS: + if getattr(args, param): + params[param] = '1' + if args.format: + params['format'] = args.format + if args.lang: + params['accept-language'] = args.lang + if args.polygon_output: + params['polygon_' + args.polygon_output] = '1' + if args.polygon_threshold: + params['polygon_threshold'] = args.polygon_threshold + + return run_api_script('lookup', args.project_dir, + phpcgi_bin=args.phpcgi_path, params=params) + + +class APIDetails: + """\ + Execute API lookup query. + """ + + @staticmethod + def add_args(parser): + group = parser.add_argument_group('Query arguments') + objs = group.add_mutually_exclusive_group(required=True) + objs.add_argument('--node', '-n', type=int, + help="Look up the OSM node with the given ID.") + objs.add_argument('--way', '-w', type=int, + help="Look up the OSM way with the given ID.") + objs.add_argument('--relation', '-r', type=int, + help="Look up the OSM relation with the given ID.") + objs.add_argument('--place_id', '-p', type=int, + help='Database internal identifier of the OSM object to look up.') + group.add_argument('--class', dest='object_class', + help="""Class type to disambiguated multiple entries + of the same object.""") + + group = parser.add_argument_group('Output arguments') + for name, desc in DETAILS_SWITCHES: + group.add_argument('--' + name, action='store_true', help=desc) + group.add_argument('--lang', '--accept-language', metavar='LANGS', + help='Preferred language order for presenting search results') + + @staticmethod + def run(args): + if args.node: + params = dict(osmtype='N', osmid=args.node) + elif args.way: + params = dict(osmtype='W', osmid=args.node) + elif args.relation: + params = dict(osmtype='R', osmid=args.node) + else: + params = dict(place_id=args.place_id) + if args.object_class: + params['class'] = args.object_class + for name, _ in DETAILS_SWITCHES: + params[name] = '1' if getattr(args, name) else '0' + + return run_api_script('details', args.project_dir, + phpcgi_bin=args.phpcgi_path, params=params) + + +class APIStatus: + """\ + Execute API status query. + """ + + @staticmethod + def add_args(parser): + group = parser.add_argument_group('API parameters') + group.add_argument('--format', default='text', choices=['text', 'json'], + help='Format of result') + + @staticmethod + def run(args): + return run_api_script('status', args.project_dir, + phpcgi_bin=args.phpcgi_path, + params=dict(format=args.format)) def nominatim(**kwargs): @@ -476,10 +774,14 @@ def nominatim(**kwargs): parser.add_subcommand('refresh', UpdateRefresh) parser.add_subcommand('export', QueryExport) - parser.add_subcommand('search', QueryTodo) - parser.add_subcommand('reverse', QueryTodo) - parser.add_subcommand('lookup', QueryTodo) - parser.add_subcommand('details', QueryTodo) - parser.add_subcommand('status', QueryTodo) - parser.run(**kwargs) + if kwargs.get('phpcgi_path'): + parser.add_subcommand('search', APISearch) + parser.add_subcommand('reverse', APIReverse) + parser.add_subcommand('lookup', APILookup) + parser.add_subcommand('details', APIDetails) + parser.add_subcommand('status', APIStatus) + else: + parser.parser.epilog = 'php-cgi not found. Query commands not available.' + + return parser.run(**kwargs)