nominatim/clicmd/refresh.py

   1 # SPDX-License-Identifier: GPL-2.0-only
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2022 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 Implementation of 'refresh' subcommand.
   9 """
  10 import logging
  11 from pathlib import Path
  12
  13 from nominatim.db.connection import connect
  14
  15 # Do not repeat documentation of subcommand classes.
  16 # pylint: disable=C0111
  17 # Using non-top-level imports to avoid eventually unused imports.
  18 # pylint: disable=E0012,C0415
  19
  20 LOG = logging.getLogger()
  21
  22 class UpdateRefresh:
  23     """\
  24     Recompute auxiliary data used by the indexing process.
  25
  26     This sub-commands updates various static data and functions in the database.
  27     It usually needs to be run after changing various aspects of the
  28     configuration. The configuration documentation will mention the exact
  29     command to use in such case.
  30
  31     Warning: the 'update' command must not be run in parallel with other update
  32              commands like 'replication' or 'add-data'.
  33     """
  34     def __init__(self):
  35         self.tokenizer = None
  36
  37     @staticmethod
  38     def add_args(parser):
  39         group = parser.add_argument_group('Data arguments')
  40         group.add_argument('--postcodes', action='store_true',
  41                            help='Update postcode centroid table')
  42         group.add_argument('--word-counts', action='store_true',
  43                            help='Compute frequency of full-word search terms')
  44         group.add_argument('--address-levels', action='store_true',
  45                            help='Reimport address level configuration')
  46         group.add_argument('--functions', action='store_true',
  47                            help='Update the PL/pgSQL functions in the database')
  48         group.add_argument('--wiki-data', action='store_true',
  49                            help='Update Wikipedia/data importance numbers')
  50         group.add_argument('--importance', action='store_true',
  51                            help='Recompute place importances (expensive!)')
  52         group.add_argument('--website', action='store_true',
  53                            help='Refresh the directory that serves the scripts for the web API')
  54         group = parser.add_argument_group('Arguments for function refresh')
  55         group.add_argument('--no-diff-updates', action='store_false', dest='diffs',
  56                            help='Do not enable code for propagating updates')
  57         group.add_argument('--enable-debug-statements', action='store_true',
  58                            help='Enable debug warning statements in functions')
  59
  60
  61     def run(self, args):
  62         from ..tools import refresh, postcodes
  63         from ..indexer.indexer import Indexer
  64
  65
  66         if args.postcodes:
  67             if postcodes.can_compute(args.config.get_libpq_dsn()):
  68                 LOG.warning("Update postcodes centroid")
  69                 tokenizer = self._get_tokenizer(args.config)
  70                 postcodes.update_postcodes(args.config.get_libpq_dsn(),
  71                                            args.project_dir, tokenizer)
  72                 indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
  73                                   args.threads or 1)
  74                 indexer.index_postcodes()
  75             else:
  76                 LOG.error("The place table doesn't exist. "
  77                           "Postcode updates on a frozen database is not possible.")
  78
  79         if args.word_counts:
  80             LOG.warning('Recompute word statistics')
  81             self._get_tokenizer(args.config).update_statistics()
  82
  83         if args.address_levels:
  84             LOG.warning('Updating address levels')
  85             with connect(args.config.get_libpq_dsn()) as conn:
  86                 refresh.load_address_levels_from_config(conn, args.config)
  87
  88         if args.functions:
  89             LOG.warning('Create functions')
  90             with connect(args.config.get_libpq_dsn()) as conn:
  91                 refresh.create_functions(conn, args.config,
  92                                          args.diffs, args.enable_debug_statements)
  93                 self._get_tokenizer(args.config).update_sql_functions(args.config)
  94
  95         if args.wiki_data:
  96             data_path = Path(args.config.WIKIPEDIA_DATA_PATH
  97                              or args.project_dir)
  98             LOG.warning('Import wikipdia article importance from %s', data_path)
  99             if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
 100                                                  data_path) > 0:
 101                 LOG.fatal('FATAL: Wikipedia importance dump file not found')
 102                 return 1
 103
 104         # Attention: importance MUST come after wiki data import.
 105         if args.importance:
 106             LOG.warning('Update importance values for database')
 107             with connect(args.config.get_libpq_dsn()) as conn:
 108                 refresh.recompute_importance(conn)
 109
 110         if args.website:
 111             webdir = args.project_dir / 'website'
 112             LOG.warning('Setting up website directory at %s', webdir)
 113             with connect(args.config.get_libpq_dsn()) as conn:
 114                 refresh.setup_website(webdir, args.config, conn)
 115
 116         return 0
 117
 118
 119     def _get_tokenizer(self, config):
 120         if self.tokenizer is None:
 121             from ..tokenizer import factory as tokenizer_factory
 122
 123             self.tokenizer = tokenizer_factory.get_tokenizer_for_db(config)
 124
 125         return self.tokenizer