2 Implementation of the 'import' subcommand.
5 from pathlib import Path
9 from nominatim.db.connection import connect
10 from nominatim.db import status, properties
11 from nominatim.version import NOMINATIM_VERSION
12 from nominatim.errors import UsageError
14 # Do not repeat documentation of subcommand classes.
15 # pylint: disable=C0111
16 # Using non-top-level imports to avoid eventually unused imports.
17 # pylint: disable=E0012,C0415
19 LOG = logging.getLogger()
23 Create a new Nominatim database from an OSM file.
28 group_name = parser.add_argument_group('Required arguments')
29 group = group_name.add_mutually_exclusive_group(required=True)
30 group.add_argument('--osm-file', metavar='FILE', action='append',
31 help='OSM file to be imported'
32 ' (repeat for importing multiple files.')
33 group.add_argument('--continue', dest='continue_at',
34 choices=['load-data', 'indexing', 'db-postprocess'],
35 help='Continue an import that was interrupted')
36 group = parser.add_argument_group('Optional arguments')
37 group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
38 help='Size of cache to be used by osm2pgsql (in MB)')
39 group.add_argument('--reverse-only', action='store_true',
40 help='Do not create tables and indexes for searching')
41 group.add_argument('--no-partitions', action='store_true',
42 help=("Do not partition search indices "
43 "(speeds up import of single country extracts)"))
44 group.add_argument('--no-updates', action='store_true',
45 help="Do not keep tables that are only needed for "
46 "updating the database later")
47 group = parser.add_argument_group('Expert options')
48 group.add_argument('--ignore-errors', action='store_true',
49 help='Continue import even when errors in SQL are present')
50 group.add_argument('--index-noanalyse', action='store_true',
51 help='Do not perform analyse operations during index')
55 def run(args): # pylint: disable=too-many-statements
56 from ..tools import database_import, refresh, postcodes, freeze
57 from ..indexer.indexer import Indexer
58 from ..tokenizer import factory as tokenizer_factory
61 files = [Path(f) for f in args.osm_file]
63 if not fname.is_file():
64 LOG.fatal("OSM file '%s' does not exist.", fname)
65 raise UsageError('Cannot access file.')
67 if args.continue_at is None:
68 database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
71 rouser=args.config.DATABASE_WEBUSER)
73 LOG.warning('Importing OSM data file')
74 database_import.import_osm_data(files,
75 args.osm2pgsql_options(0, 1),
77 ignore_errors=args.ignore_errors)
79 with connect(args.config.get_libpq_dsn()) as conn:
80 LOG.warning('Create functions (1st pass)')
81 refresh.create_functions(conn, args.config, False, False)
82 LOG.warning('Create tables')
83 database_import.create_tables(conn, args.config,
84 reverse_only=args.reverse_only)
85 refresh.load_address_levels_from_file(conn, Path(args.config.ADDRESS_LEVEL_CONFIG))
86 LOG.warning('Create functions (2nd pass)')
87 refresh.create_functions(conn, args.config, False, False)
88 LOG.warning('Create table triggers')
89 database_import.create_table_triggers(conn, args.config)
90 LOG.warning('Create partition tables')
91 database_import.create_partition_tables(conn, args.config)
92 LOG.warning('Create functions (3rd pass)')
93 refresh.create_functions(conn, args.config, False, False)
95 LOG.warning('Importing wikipedia importance data')
96 data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
97 if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
99 LOG.error('Wikipedia importance dump file not found. '
100 'Will be using default importances.')
102 if args.continue_at is None or args.continue_at == 'load-data':
103 LOG.warning('Initialise tables')
104 with connect(args.config.get_libpq_dsn()) as conn:
105 database_import.truncate_data_tables(conn)
107 LOG.warning('Load data into placex table')
108 database_import.load_data(args.config.get_libpq_dsn(),
109 args.threads or psutil.cpu_count() or 1)
111 LOG.warning("Setting up tokenizer")
112 if args.continue_at is None or args.continue_at == 'load-data':
113 # (re)initialise the tokenizer data
114 tokenizer = tokenizer_factory.create_tokenizer(args.config)
116 # just load the tokenizer
117 tokenizer = tokenizer_factory.get_tokenizer_for_db(args.config)
119 if args.continue_at is None or args.continue_at == 'load-data':
120 LOG.warning('Calculate postcodes')
121 postcodes.update_postcodes(args.config.get_libpq_dsn(),
122 args.project_dir, tokenizer)
124 if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
125 if args.continue_at is not None and args.continue_at != 'load-data':
126 with connect(args.config.get_libpq_dsn()) as conn:
127 SetupAll._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
128 LOG.warning('Indexing places')
129 indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
130 args.threads or psutil.cpu_count() or 1)
131 indexer.index_full(analyse=not args.index_noanalyse)
133 LOG.warning('Post-process tables')
134 with connect(args.config.get_libpq_dsn()) as conn:
135 database_import.create_search_indices(conn, args.config,
136 drop=args.no_updates)
137 LOG.warning('Create search index for default country names.')
138 database_import.create_country_names(conn, tokenizer,
139 args.config.LANGUAGES)
142 freeze.drop_update_tables(conn)
143 tokenizer.finalize_import(args.config)
146 webdir = args.project_dir / 'website'
147 LOG.warning('Setup website at %s', webdir)
148 with connect(args.config.get_libpq_dsn()) as conn:
149 refresh.setup_website(webdir, args.config, conn)
151 with connect(args.config.get_libpq_dsn()) as conn:
153 dbdate = status.compute_database_date(conn)
154 status.set_status(conn, dbdate)
155 LOG.info('Database is at %s.', dbdate)
156 except Exception as exc: # pylint: disable=broad-except
157 LOG.error('Cannot determine date of database: %s', exc)
159 properties.set_property(conn, 'database_version',
160 '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
166 def _create_pending_index(conn, tablespace):
167 """ Add a supporting index for finding places still to be indexed.
169 This index is normally created at the end of the import process
170 for later updates. When indexing was partially done, then this
171 index can greatly improve speed going through already indexed data.
173 if conn.index_exists('idx_placex_pendingsector'):
176 with conn.cursor() as cur:
177 LOG.warning('Creating support index')
179 tablespace = 'TABLESPACE ' + tablespace
180 cur.execute("""CREATE INDEX idx_placex_pendingsector
181 ON placex USING BTREE (rank_address,geometry_sector)
182 {} WHERE indexed_status > 0
183 """.format(tablespace))