]> git.openstreetmap.org Git - nominatim.git/blob - nominatim/clicmd/setup.py
indexer: reset query counter
[nominatim.git] / nominatim / clicmd / setup.py
1 """
2 Implementation of the 'import' subcommand.
3 """
4 import logging
5 from pathlib import Path
6
7 import psutil
8
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
13
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
18
19 LOG = logging.getLogger()
20
21 class SetupAll:
22     """\
23     Create a new Nominatim database from an OSM file.
24     """
25
26     @staticmethod
27     def add_args(parser):
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',
31                            help='OSM file to be imported.')
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')
51
52
53     @staticmethod
54     def run(args): # pylint: disable=too-many-statements
55         from ..tools import database_import
56         from ..tools import refresh
57         from ..indexer.indexer import Indexer
58         from ..tools import postcodes
59
60         if args.osm_file and not Path(args.osm_file).is_file():
61             LOG.fatal("OSM file '%s' does not exist.", args.osm_file)
62             raise UsageError('Cannot access file.')
63
64         if args.continue_at is None:
65             database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
66                                                     args.data_dir,
67                                                     args.no_partitions,
68                                                     rouser=args.config.DATABASE_WEBUSER)
69
70             LOG.warning('Installing database module')
71             with connect(args.config.get_libpq_dsn()) as conn:
72                 database_import.install_module(args.module_dir, args.project_dir,
73                                                args.config.DATABASE_MODULE_PATH,
74                                                conn=conn)
75
76             LOG.warning('Importing OSM data file')
77             database_import.import_osm_data(Path(args.osm_file),
78                                             args.osm2pgsql_options(0, 1),
79                                             drop=args.no_updates,
80                                             ignore_errors=args.ignore_errors)
81
82             with connect(args.config.get_libpq_dsn()) as conn:
83                 LOG.warning('Create functions (1st pass)')
84                 refresh.create_functions(conn, args.config, False, False)
85                 LOG.warning('Create tables')
86                 database_import.create_tables(conn, args.config,
87                                               reverse_only=args.reverse_only)
88                 refresh.load_address_levels_from_file(conn, Path(args.config.ADDRESS_LEVEL_CONFIG))
89                 LOG.warning('Create functions (2nd pass)')
90                 refresh.create_functions(conn, args.config, False, False)
91                 LOG.warning('Create table triggers')
92                 database_import.create_table_triggers(conn, args.config)
93                 LOG.warning('Create partition tables')
94                 database_import.create_partition_tables(conn, args.config)
95                 LOG.warning('Create functions (3rd pass)')
96                 refresh.create_functions(conn, args.config, False, False)
97
98             LOG.warning('Importing wikipedia importance data')
99             data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
100             if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
101                                                  data_path) > 0:
102                 LOG.error('Wikipedia importance dump file not found. '
103                           'Will be using default importances.')
104
105         if args.continue_at is None or args.continue_at == 'load-data':
106             LOG.warning('Initialise tables')
107             with connect(args.config.get_libpq_dsn()) as conn:
108                 database_import.truncate_data_tables(conn, args.config.MAX_WORD_FREQUENCY)
109
110             LOG.warning('Load data into placex table')
111             database_import.load_data(args.config.get_libpq_dsn(),
112                                       args.data_dir,
113                                       args.threads or psutil.cpu_count() or 1)
114
115             LOG.warning('Calculate postcodes')
116             postcodes.import_postcodes(args.config.get_libpq_dsn(), args.project_dir)
117
118         if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
119             if args.continue_at is not None and args.continue_at != 'load-data':
120                 with connect(args.config.get_libpq_dsn()) as conn:
121                     SetupAll._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
122             LOG.warning('Indexing places')
123             indexer = Indexer(args.config.get_libpq_dsn(),
124                               args.threads or psutil.cpu_count() or 1)
125             indexer.index_full(analyse=not args.index_noanalyse)
126
127         LOG.warning('Post-process tables')
128         with connect(args.config.get_libpq_dsn()) as conn:
129             database_import.create_search_indices(conn, args.config,
130                                                   drop=args.no_updates)
131             LOG.warning('Create search index for default country names.')
132             database_import.create_country_names(conn, args.config)
133
134         webdir = args.project_dir / 'website'
135         LOG.warning('Setup website at %s', webdir)
136         refresh.setup_website(webdir, args.config)
137
138         with connect(args.config.get_libpq_dsn()) as conn:
139             try:
140                 dbdate = status.compute_database_date(conn)
141                 status.set_status(conn, dbdate)
142                 LOG.info('Database is at %s.', dbdate)
143             except Exception as exc: # pylint: disable=broad-except
144                 LOG.error('Cannot determine date of database: %s', exc)
145
146             properties.set_property(conn, 'database_version',
147                                     '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
148
149         return 0
150
151
152     @staticmethod
153     def _create_pending_index(conn, tablespace):
154         """ Add a supporting index for finding places still to be indexed.
155
156             This index is normally created at the end of the import process
157             for later updates. When indexing was partially done, then this
158             index can greatly improve speed going through already indexed data.
159         """
160         if conn.index_exists('idx_placex_pendingsector'):
161             return
162
163         with conn.cursor() as cur:
164             LOG.warning('Creating support index')
165             if tablespace:
166                 tablespace = 'TABLESPACE ' + tablespace
167             cur.execute("""CREATE INDEX idx_placex_pendingsector
168                            ON placex USING BTREE (rank_address,geometry_sector)
169                            {} WHERE indexed_status > 0
170                         """.format(tablespace))
171         conn.commit()