]> git.openstreetmap.org Git - nominatim.git/blob - nominatim/clicmd/setup.py
Merge pull request #2192 from lonvia/database-versioning
[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 ..tools.exec_utils import run_legacy_script
10 from ..db.connection import connect
11 from ..db import status, properties
12 from ..version import NOMINATIM_VERSION
13 from ..errors import UsageError
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 SetupAll:
23     """\
24     Create a new Nominatim database from an OSM file.
25     """
26
27     @staticmethod
28     def add_args(parser):
29         group_name = parser.add_argument_group('Required arguments')
30         group = group_name.add_mutually_exclusive_group(required=True)
31         group.add_argument('--osm-file', metavar='FILE',
32                            help='OSM file to be imported.')
33         group.add_argument('--continue', dest='continue_at',
34                            choices=['load-data', 'indexing', 'db-postprocess'],
35                            help='Continue an import that was interrupted')
36         group = parser.add_argument_group('Optional arguments')
37         group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
38                            help='Size of cache to be used by osm2pgsql (in MB)')
39         group.add_argument('--reverse-only', action='store_true',
40                            help='Do not create tables and indexes for searching')
41         group.add_argument('--no-partitions', action='store_true',
42                            help="""Do not partition search indices
43                                    (speeds up import of single country extracts)""")
44         group.add_argument('--no-updates', action='store_true',
45                            help="""Do not keep tables that are only needed for
46                                    updating the database later""")
47         group = parser.add_argument_group('Expert options')
48         group.add_argument('--ignore-errors', action='store_true',
49                            help='Continue import even when errors in SQL are present')
50         group.add_argument('--index-noanalyse', action='store_true',
51                            help='Do not perform analyse operations during index')
52
53
54     @staticmethod
55     def run(args): # pylint: disable=too-many-statements
56         from ..tools import database_import
57         from ..tools import refresh
58         from ..indexer.indexer import Indexer
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             LOG.warning('Create functions (1st pass)')
83             with connect(args.config.get_libpq_dsn()) as conn:
84                 refresh.create_functions(conn, args.config, args.sqllib_dir,
85                                          False, False)
86
87             LOG.warning('Create tables')
88             params = ['setup.php', '--create-tables', '--create-partition-tables']
89             if args.reverse_only:
90                 params.append('--reverse-only')
91             run_legacy_script(*params, nominatim_env=args,
92                               throw_on_fail=not args.ignore_errors)
93
94             LOG.warning('Create functions (2nd pass)')
95             with connect(args.config.get_libpq_dsn()) as conn:
96                 refresh.create_functions(conn, args.config, args.sqllib_dir,
97                                          False, False)
98
99             LOG.warning('Importing wikipedia importance data')
100             data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
101             if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
102                                                  data_path) > 0:
103                 LOG.error('Wikipedia importance dump file not found. '
104                           'Will be using default importances.')
105
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         if args.continue_at is None or args.continue_at == 'load-data':
111             LOG.warning('Load data into placex table')
112             database_import.load_data(args.config.get_libpq_dsn(),
113                                       args.data_dir,
114                                       args.threads or psutil.cpu_count() or 1)
115
116             LOG.warning('Calculate postcodes')
117             run_legacy_script('setup.php', '--calculate-postcodes',
118                               nominatim_env=args, throw_on_fail=not args.ignore_errors)
119
120         if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
121             LOG.warning('Indexing places')
122             indexer = Indexer(args.config.get_libpq_dsn(),
123                               args.threads or psutil.cpu_count() or 1)
124             indexer.index_full(analyse=not args.index_noanalyse)
125
126         LOG.warning('Post-process tables')
127         params = ['setup.php', '--create-search-indices', '--create-country-names']
128         if args.no_updates:
129             params.append('--drop')
130         run_legacy_script(*params, nominatim_env=args, throw_on_fail=not args.ignore_errors)
131
132         webdir = args.project_dir / 'website'
133         LOG.warning('Setup website at %s', webdir)
134         refresh.setup_website(webdir, args.phplib_dir, args.config)
135
136         with connect(args.config.get_libpq_dsn()) as conn:
137             try:
138                 dbdate = status.compute_database_date(conn)
139                 status.set_status(conn, dbdate)
140                 LOG.info('Database is at %s.', dbdate)
141             except Exception as exc: # pylint: disable=broad-except
142                 LOG.error('Cannot determine date of database: %s', exc)
143
144             properties.set_property(conn, 'database_version',
145                                     '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
146
147         return 0