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