1 # SPDX-License-Identifier: GPL-2.0-only
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2022 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 Implementation of the 'import' subcommand.
11 from pathlib import Path
15 from nominatim.db.connection import connect
16 from nominatim.db import status, properties
17 from nominatim.version import NOMINATIM_VERSION
19 # Do not repeat documentation of subcommand classes.
20 # pylint: disable=C0111
21 # Using non-top-level imports to avoid eventually unused imports.
22 # pylint: disable=E0012,C0415
24 LOG = logging.getLogger()
28 Create a new Nominatim database from an OSM file.
30 This sub-command sets up a new Nominatim database from scratch starting
31 with creating a new database in Postgresql. The user running this command
32 needs superuser rights on the database.
37 group_name = parser.add_argument_group('Required arguments')
38 group = group_name.add_mutually_exclusive_group(required=True)
39 group.add_argument('--osm-file', metavar='FILE', action='append',
40 help='OSM file to be imported'
41 ' (repeat for importing multiple files)')
42 group.add_argument('--continue', dest='continue_at',
43 choices=['load-data', 'indexing', 'db-postprocess'],
44 help='Continue an import that was interrupted')
45 group = parser.add_argument_group('Optional arguments')
46 group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
47 help='Size of cache to be used by osm2pgsql (in MB)')
48 group.add_argument('--reverse-only', action='store_true',
49 help='Do not create tables and indexes for searching')
50 group.add_argument('--no-partitions', action='store_true',
51 help=("Do not partition search indices "
52 "(speeds up import of single country extracts)"))
53 group.add_argument('--no-updates', action='store_true',
54 help="Do not keep tables that are only needed for "
55 "updating the database later")
56 group = parser.add_argument_group('Expert options')
57 group.add_argument('--ignore-errors', action='store_true',
58 help='Continue import even when errors in SQL are present')
59 group.add_argument('--index-noanalyse', action='store_true',
60 help='Do not perform analyse operations during index (expert only)')
65 from ..tools import database_import, refresh, postcodes, freeze, country_info
66 from ..indexer.indexer import Indexer
68 country_info.setup_country_config(args.config)
70 if args.continue_at is None:
71 files = args.get_osm_file_list()
73 LOG.warning('Creating database')
74 database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
75 rouser=args.config.DATABASE_WEBUSER)
77 LOG.warning('Setting up country tables')
78 country_info.setup_country_tables(args.config.get_libpq_dsn(),
82 LOG.warning('Importing OSM data file')
83 database_import.import_osm_data(files,
84 args.osm2pgsql_options(0, 1),
86 ignore_errors=args.ignore_errors)
88 SetupAll._setup_tables(args.config, args.reverse_only)
90 LOG.warning('Importing wikipedia importance data')
91 data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
92 if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
94 LOG.error('Wikipedia importance dump file not found. '
95 'Will be using default importances.')
97 if args.continue_at is None or args.continue_at == 'load-data':
98 LOG.warning('Initialise tables')
99 with connect(args.config.get_libpq_dsn()) as conn:
100 database_import.truncate_data_tables(conn)
102 LOG.warning('Load data into placex table')
103 database_import.load_data(args.config.get_libpq_dsn(),
104 args.threads or psutil.cpu_count() or 1)
106 LOG.warning("Setting up tokenizer")
107 tokenizer = SetupAll._get_tokenizer(args.continue_at, args.config)
109 if args.continue_at is None or args.continue_at == 'load-data':
110 LOG.warning('Calculate postcodes')
111 postcodes.update_postcodes(args.config.get_libpq_dsn(),
112 args.project_dir, tokenizer)
114 if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
115 if args.continue_at is not None and args.continue_at != 'load-data':
116 with connect(args.config.get_libpq_dsn()) as conn:
117 SetupAll._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
118 LOG.warning('Indexing places')
119 indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
120 args.threads or psutil.cpu_count() or 1)
121 indexer.index_full(analyse=not args.index_noanalyse)
123 LOG.warning('Post-process tables')
124 with connect(args.config.get_libpq_dsn()) as conn:
125 database_import.create_search_indices(conn, args.config,
126 drop=args.no_updates)
127 LOG.warning('Create search index for default country names.')
128 country_info.create_country_names(conn, tokenizer,
129 args.config.LANGUAGES)
131 freeze.drop_update_tables(conn)
132 tokenizer.finalize_import(args.config)
134 LOG.warning('Recompute word counts')
135 tokenizer.update_statistics()
137 webdir = args.project_dir / 'website'
138 LOG.warning('Setup website at %s', webdir)
139 with connect(args.config.get_libpq_dsn()) as conn:
140 refresh.setup_website(webdir, args.config, conn)
142 SetupAll._set_database_date(args.config.get_libpq_dsn())
148 def _setup_tables(config, reverse_only):
149 """ Set up the basic database layout: tables, indexes and functions.
151 from ..tools import database_import, refresh
153 with connect(config.get_libpq_dsn()) as conn:
154 LOG.warning('Create functions (1st pass)')
155 refresh.create_functions(conn, config, False, False)
156 LOG.warning('Create tables')
157 database_import.create_tables(conn, config, reverse_only=reverse_only)
158 refresh.load_address_levels_from_config(conn, config)
159 LOG.warning('Create functions (2nd pass)')
160 refresh.create_functions(conn, config, False, False)
161 LOG.warning('Create table triggers')
162 database_import.create_table_triggers(conn, config)
163 LOG.warning('Create partition tables')
164 database_import.create_partition_tables(conn, config)
165 LOG.warning('Create functions (3rd pass)')
166 refresh.create_functions(conn, config, False, False)
170 def _get_tokenizer(continue_at, config):
171 """ Set up a new tokenizer or load an already initialised one.
173 from ..tokenizer import factory as tokenizer_factory
175 if continue_at is None or continue_at == 'load-data':
176 # (re)initialise the tokenizer data
177 return tokenizer_factory.create_tokenizer(config)
179 # just load the tokenizer
180 return tokenizer_factory.get_tokenizer_for_db(config)
183 def _create_pending_index(conn, tablespace):
184 """ Add a supporting index for finding places still to be indexed.
186 This index is normally created at the end of the import process
187 for later updates. When indexing was partially done, then this
188 index can greatly improve speed going through already indexed data.
190 if conn.index_exists('idx_placex_pendingsector'):
193 with conn.cursor() as cur:
194 LOG.warning('Creating support index')
196 tablespace = 'TABLESPACE ' + tablespace
197 cur.execute("""CREATE INDEX idx_placex_pendingsector
198 ON placex USING BTREE (rank_address,geometry_sector)
199 {} WHERE indexed_status > 0
200 """.format(tablespace))
205 def _set_database_date(dsn):
206 """ Determine the database date and set the status accordingly.
208 with connect(dsn) as conn:
210 dbdate = status.compute_database_date(conn)
211 status.set_status(conn, dbdate)
212 LOG.info('Database is at %s.', dbdate)
213 except Exception as exc: # pylint: disable=broad-except
214 LOG.error('Cannot determine date of database: %s', exc)
216 properties.set_property(conn, 'database_version',
217 '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))