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
13 # Do not repeat documentation of subcommand classes.
14 # pylint: disable=C0111
15 # Using non-top-level imports to avoid eventually unused imports.
16 # pylint: disable=E0012,C0415
18 LOG = logging.getLogger()
22 Create a new Nominatim database from an OSM file.
27 group_name = parser.add_argument_group('Required arguments')
28 group = group_name.add_mutually_exclusive_group(required=True)
29 group.add_argument('--osm-file', metavar='FILE', action='append',
30 help='OSM file to be imported'
31 ' (repeat for importing multiple files.')
32 group.add_argument('--continue', dest='continue_at',
33 choices=['load-data', 'indexing', 'db-postprocess'],
34 help='Continue an import that was interrupted')
35 group = parser.add_argument_group('Optional arguments')
36 group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
37 help='Size of cache to be used by osm2pgsql (in MB)')
38 group.add_argument('--reverse-only', action='store_true',
39 help='Do not create tables and indexes for searching')
40 group.add_argument('--no-partitions', action='store_true',
41 help=("Do not partition search indices "
42 "(speeds up import of single country extracts)"))
43 group.add_argument('--no-updates', action='store_true',
44 help="Do not keep tables that are only needed for "
45 "updating the database later")
46 group = parser.add_argument_group('Expert options')
47 group.add_argument('--ignore-errors', action='store_true',
48 help='Continue import even when errors in SQL are present')
49 group.add_argument('--index-noanalyse', action='store_true',
50 help='Do not perform analyse operations during index')
55 from ..tools import database_import, refresh, postcodes, freeze, country_info
56 from ..indexer.indexer import Indexer
58 if args.continue_at is None:
59 files = args.get_osm_file_list()
61 LOG.warning('Creating database')
62 database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
63 rouser=args.config.DATABASE_WEBUSER)
65 LOG.warning('Setting up country tables')
66 country_info.setup_country_tables(args.config.get_libpq_dsn(),
70 LOG.warning('Importing OSM data file')
71 database_import.import_osm_data(files,
72 args.osm2pgsql_options(0, 1),
74 ignore_errors=args.ignore_errors)
76 SetupAll._setup_tables(args.config, args.reverse_only)
78 LOG.warning('Importing wikipedia importance data')
79 data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
80 if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
82 LOG.error('Wikipedia importance dump file not found. '
83 'Will be using default importances.')
85 if args.continue_at is None or args.continue_at == 'load-data':
86 LOG.warning('Initialise tables')
87 with connect(args.config.get_libpq_dsn()) as conn:
88 database_import.truncate_data_tables(conn)
90 LOG.warning('Load data into placex table')
91 database_import.load_data(args.config.get_libpq_dsn(),
92 args.threads or psutil.cpu_count() or 1)
94 LOG.warning("Setting up tokenizer")
95 tokenizer = SetupAll._get_tokenizer(args.continue_at, args.config)
97 if args.continue_at is None or args.continue_at == 'load-data':
98 LOG.warning('Calculate postcodes')
99 postcodes.update_postcodes(args.config.get_libpq_dsn(),
100 args.project_dir, tokenizer)
102 if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
103 if args.continue_at is not None and args.continue_at != 'load-data':
104 with connect(args.config.get_libpq_dsn()) as conn:
105 SetupAll._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
106 LOG.warning('Indexing places')
107 indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
108 args.threads or psutil.cpu_count() or 1)
109 indexer.index_full(analyse=not args.index_noanalyse)
111 LOG.warning('Post-process tables')
112 with connect(args.config.get_libpq_dsn()) as conn:
113 database_import.create_search_indices(conn, args.config,
114 drop=args.no_updates)
115 LOG.warning('Create search index for default country names.')
116 country_info.create_country_names(conn, tokenizer,
117 args.config.LANGUAGES)
120 freeze.drop_update_tables(conn)
121 tokenizer.finalize_import(args.config)
124 webdir = args.project_dir / 'website'
125 LOG.warning('Setup website at %s', webdir)
126 with connect(args.config.get_libpq_dsn()) as conn:
127 refresh.setup_website(webdir, args.config, conn)
129 with connect(args.config.get_libpq_dsn()) as conn:
130 SetupAll._set_database_date(conn)
131 properties.set_property(conn, 'database_version',
132 '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
138 def _setup_tables(config, reverse_only):
139 """ Set up the basic database layout: tables, indexes and functions.
141 from ..tools import database_import, refresh
143 with connect(config.get_libpq_dsn()) as conn:
144 LOG.warning('Create functions (1st pass)')
145 refresh.create_functions(conn, config, False, False)
146 LOG.warning('Create tables')
147 database_import.create_tables(conn, config, reverse_only=reverse_only)
148 refresh.load_address_levels_from_file(conn, Path(config.ADDRESS_LEVEL_CONFIG))
149 LOG.warning('Create functions (2nd pass)')
150 refresh.create_functions(conn, config, False, False)
151 LOG.warning('Create table triggers')
152 database_import.create_table_triggers(conn, config)
153 LOG.warning('Create partition tables')
154 database_import.create_partition_tables(conn, config)
155 LOG.warning('Create functions (3rd pass)')
156 refresh.create_functions(conn, config, False, False)
160 def _get_tokenizer(continue_at, config):
161 """ Set up a new tokenizer or load an already initialised one.
163 from ..tokenizer import factory as tokenizer_factory
165 if continue_at is None or continue_at == 'load-data':
166 # (re)initialise the tokenizer data
167 return tokenizer_factory.create_tokenizer(config)
169 # just load the tokenizer
170 return tokenizer_factory.get_tokenizer_for_db(config)
173 def _create_pending_index(conn, tablespace):
174 """ Add a supporting index for finding places still to be indexed.
176 This index is normally created at the end of the import process
177 for later updates. When indexing was partially done, then this
178 index can greatly improve speed going through already indexed data.
180 if conn.index_exists('idx_placex_pendingsector'):
183 with conn.cursor() as cur:
184 LOG.warning('Creating support index')
186 tablespace = 'TABLESPACE ' + tablespace
187 cur.execute("""CREATE INDEX idx_placex_pendingsector
188 ON placex USING BTREE (rank_address,geometry_sector)
189 {} WHERE indexed_status > 0
190 """.format(tablespace))
195 def _set_database_date(conn):
196 """ Determine the database date and set the status accordingly.
199 dbdate = status.compute_database_date(conn)
200 status.set_status(conn, dbdate)
201 LOG.info('Database is at %s.', dbdate)
202 except Exception as exc: # pylint: disable=broad-except
203 LOG.error('Cannot determine date of database: %s', exc)