2 Implementation of 'refresh' subcommand.
 
   5 from pathlib import Path
 
   7 from nominatim.db.connection import connect
 
   9 # Do not repeat documentation of subcommand classes.
 
  10 # pylint: disable=C0111
 
  11 # Using non-top-level imports to avoid eventually unused imports.
 
  12 # pylint: disable=E0012,C0415
 
  14 LOG = logging.getLogger()
 
  18     Recompute auxiliary data used by the indexing process.
 
  20     These functions must not be run in parallel with other update commands.
 
  25         group = parser.add_argument_group('Data arguments')
 
  26         group.add_argument('--postcodes', action='store_true',
 
  27                            help='Update postcode centroid table')
 
  28         group.add_argument('--word-counts', action='store_true',
 
  29                            help='Compute frequency of full-word search terms')
 
  30         group.add_argument('--address-levels', action='store_true',
 
  31                            help='Reimport address level configuration')
 
  32         group.add_argument('--functions', action='store_true',
 
  33                            help='Update the PL/pgSQL functions in the database')
 
  34         group.add_argument('--wiki-data', action='store_true',
 
  35                            help='Update Wikipedia/data importance numbers.')
 
  36         group.add_argument('--importance', action='store_true',
 
  37                            help='Recompute place importances (expensive!)')
 
  38         group.add_argument('--website', action='store_true',
 
  39                            help='Refresh the directory that serves the scripts for the web API')
 
  40         group = parser.add_argument_group('Arguments for function refresh')
 
  41         group.add_argument('--no-diff-updates', action='store_false', dest='diffs',
 
  42                            help='Do not enable code for propagating updates')
 
  43         group.add_argument('--enable-debug-statements', action='store_true',
 
  44                            help='Enable debug warning statements in functions')
 
  48         from ..tools import refresh
 
  49         from ..tokenizer import factory as tokenizer_factory
 
  52             LOG.warning("Update postcodes centroid")
 
  53             refresh.update_postcodes(args.config.get_libpq_dsn(), args.sqllib_dir)
 
  56             LOG.warning('Recompute frequency of full-word search terms')
 
  57             refresh.recompute_word_counts(args.config.get_libpq_dsn(), args.sqllib_dir)
 
  59         if args.address_levels:
 
  60             cfg = Path(args.config.ADDRESS_LEVEL_CONFIG)
 
  61             LOG.warning('Updating address levels from %s', cfg)
 
  62             with connect(args.config.get_libpq_dsn()) as conn:
 
  63                 refresh.load_address_levels_from_file(conn, cfg)
 
  66             LOG.warning('Create functions')
 
  67             with connect(args.config.get_libpq_dsn()) as conn:
 
  68                 refresh.create_functions(conn, args.config,
 
  69                                          args.diffs, args.enable_debug_statements)
 
  70                 tokenizer = tokenizer_factory.get_tokenizer_for_db(args.config)
 
  71                 tokenizer.update_sql_functions(args.config)
 
  74             data_path = Path(args.config.WIKIPEDIA_DATA_PATH
 
  76             LOG.warning('Import wikipdia article importance from %s', data_path)
 
  77             if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
 
  79                 LOG.fatal('FATAL: Wikipedia importance dump file not found')
 
  82         # Attention: importance MUST come after wiki data import.
 
  84             LOG.warning('Update importance values for database')
 
  85             with connect(args.config.get_libpq_dsn()) as conn:
 
  86                 refresh.recompute_importance(conn)
 
  89             webdir = args.project_dir / 'website'
 
  90             LOG.warning('Setting up website directory at %s', webdir)
 
  91             refresh.setup_website(webdir, args.config)