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 ..tools.exec_utils import run_legacy_script
  10 from ..db.connection import connect
  11 from ..db import status
  12 from ..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
  59         if args.osm_file and not Path(args.osm_file).is_file():
  60             LOG.fatal("OSM file '%s' does not exist.", args.osm_file)
  61             raise UsageError('Cannot access file.')
  62
  63         if args.continue_at is None:
  64             database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
  65                                                     args.data_dir,
  66                                                     args.no_partitions,
  67                                                     rouser=args.config.DATABASE_WEBUSER)
  68
  69             LOG.warning('Installing database module')
  70             with connect(args.config.get_libpq_dsn()) as conn:
  71                 database_import.install_module(args.module_dir, args.project_dir,
  72                                                args.config.DATABASE_MODULE_PATH,
  73                                                conn=conn)
  74
  75             LOG.warning('Importing OSM data file')
  76             database_import.import_osm_data(Path(args.osm_file),
  77                                             args.osm2pgsql_options(0, 1),
  78                                             drop=args.no_updates)
  79
  80             LOG.warning('Create functions (1st pass)')
  81             with connect(args.config.get_libpq_dsn()) as conn:
  82                 refresh.create_functions(conn, args.config, args.sqllib_dir,
  83                                          False, False)
  84
  85             LOG.warning('Create tables')
  86             params = ['setup.php', '--create-tables', '--create-partition-tables']
  87             if args.reverse_only:
  88                 params.append('--reverse-only')
  89             run_legacy_script(*params, nominatim_env=args)
  90
  91             LOG.warning('Create functions (2nd pass)')
  92             with connect(args.config.get_libpq_dsn()) as conn:
  93                 refresh.create_functions(conn, args.config, args.sqllib_dir,
  94                                          False, False)
  95
  96             LOG.warning('Importing wikipedia importance data')
  97             data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
  98             if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
  99                                                  data_path) > 0:
 100                 LOG.error('Wikipedia importance dump file not found. '
 101                           'Will be using default importances.')
 102
 103             LOG.warning('Initialise tables')
 104             with connect(args.config.get_libpq_dsn()) as conn:
 105                 database_import.truncate_data_tables(conn, args.config.MAX_WORD_FREQUENCY)
 106
 107         if args.continue_at is None or args.continue_at == 'load-data':
 108             LOG.warning('Load data into placex table')
 109             database_import.load_data(args.config.get_libpq_dsn(),
 110                                       args.data_dir,
 111                                       args.threads or psutil.cpu_count() or 1)
 112
 113             LOG.warning('Calculate postcodes')
 114             run_legacy_script('setup.php', '--calculate-postcodes', nominatim_env=args)
 115
 116         if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
 117             LOG.warning('Indexing places')
 118             indexer = Indexer(args.config.get_libpq_dsn(),
 119                               args.threads or psutil.cpu_count() or 1)
 120             indexer.index_full(analyse=not args.index_noanalyse)
 121
 122         LOG.warning('Post-process tables')
 123         params = ['setup.php', '--create-search-indices', '--create-country-names']
 124         if args.no_updates:
 125             params.append('--drop')
 126         run_legacy_script(*params, nominatim_env=args)
 127
 128         webdir = args.project_dir / 'website'
 129         LOG.warning('Setup website at %s', webdir)
 130         refresh.setup_website(webdir, args.phplib_dir, args.config)
 131
 132         with connect(args.config.get_libpq_dsn()) as conn:
 133             try:
 134                 dbdate = status.compute_database_date(conn)
 135                 status.set_status(conn, dbdate)
 136                 LOG.info('Database is at %s.', dbdate)
 137             except Exception as exc: # pylint: disable=broad-except
 138                 LOG.error('Cannot determine date of database: %s', exc)
 139
 140         return 0