]> git.openstreetmap.org Git - nominatim.git/blobdiff - nominatim/clicmd/setup.py
reintroduce cutoffs when searching for very frequent words
[nominatim.git] / nominatim / clicmd / setup.py
index b4dde6fe862939c804fae590c62486a0b85cc362..ccd6bd788ed315a45bc8d2e7b2d872bdb64c8de5 100644 (file)
@@ -15,10 +15,10 @@ from pathlib import Path
 import psutil
 
 from nominatim.config import Configuration
-from nominatim.db.connection import connect, Connection
+from nominatim.db.connection import connect
 from nominatim.db import status, properties
 from nominatim.tokenizer.base import AbstractTokenizer
-from nominatim.version import version_str
+from nominatim.version import NOMINATIM_VERSION
 from nominatim.clicmd.args import NominatimArgs
 from nominatim.errors import UsageError
 
@@ -39,14 +39,15 @@ class SetupAll:
     """
 
     def add_args(self, parser: argparse.ArgumentParser) -> None:
-        group_name = parser.add_argument_group('Required arguments')
-        group1 = group_name.add_mutually_exclusive_group(required=True)
+        group1 = parser.add_argument_group('Required arguments')
         group1.add_argument('--osm-file', metavar='FILE', action='append',
                            help='OSM file to be imported'
-                                ' (repeat for importing multiple files)')
+                                ' (repeat for importing multiple files)',
+                                default=None)
         group1.add_argument('--continue', dest='continue_at',
-                           choices=['load-data', 'indexing', 'db-postprocess'],
-                           help='Continue an import that was interrupted')
+                           choices=['import-from-file', 'load-data', 'indexing', 'db-postprocess'],
+                           help='Continue an import that was interrupted',
+                           default=None)
         group2 = parser.add_argument_group('Optional arguments')
         group2.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
                            help='Size of cache to be used by osm2pgsql (in MB)')
@@ -59,15 +60,17 @@ class SetupAll:
                            help="Do not keep tables that are only needed for "
                                 "updating the database later")
         group2.add_argument('--offline', action='store_true',
-                           help="Do not attempt to load any additional data from the internet")
+                            help="Do not attempt to load any additional data from the internet")
         group3 = parser.add_argument_group('Expert options')
         group3.add_argument('--ignore-errors', action='store_true',
                            help='Continue import even when errors in SQL are present')
         group3.add_argument('--index-noanalyse', action='store_true',
                            help='Do not perform analyse operations during index (expert only)')
+        group3.add_argument('--prepare-database', action='store_true',
+                            help='Create the database but do not import any data')
 
 
-    def run(self, args: NominatimArgs) -> int: # pylint: disable=too-many-statements
+    def run(self, args: NominatimArgs) -> int: # pylint: disable=too-many-statements, too-many-branches
         from ..data import country_info
         from ..tools import database_import, refresh, postcodes, freeze
         from ..indexer.indexer import Indexer
@@ -76,36 +79,61 @@ class SetupAll:
 
         country_info.setup_country_config(args.config)
 
-        if args.continue_at is None:
-            files = args.get_osm_file_list()
-            if not files:
-                raise UsageError("No input files (use --osm-file).")
+        if args.osm_file is None and args.continue_at is None and not args.prepare_database:
+            raise UsageError("No input files (use --osm-file).")
 
-            LOG.warning('Creating database')
-            database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
-                                                    rouser=args.config.DATABASE_WEBUSER)
+        if args.osm_file is not None and args.continue_at not in ('import-from-file', None):
+            raise UsageError(f"Cannot use --continue {args.continue_at} and --osm-file together.")
 
-            LOG.warning('Setting up country tables')
-            country_info.setup_country_tables(args.config.get_libpq_dsn(),
-                                              args.data_dir,
-                                              args.no_partitions)
+        if args.continue_at is not None and args.prepare_database:
+            raise UsageError(
+                "Cannot use --continue and --prepare-database together."
+            )
 
-            LOG.warning('Importing OSM data file')
-            database_import.import_osm_data(files,
-                                            args.osm2pgsql_options(0, 1),
-                                            drop=args.no_updates,
-                                            ignore_errors=args.ignore_errors)
 
-            self._setup_tables(args.config, args.reverse_only)
+        if args.prepare_database or args.continue_at is None:
+            LOG.warning('Creating database')
+            database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
+                                                        rouser=args.config.DATABASE_WEBUSER)
+            if args.prepare_database:
+                return 0
 
-            LOG.warning('Importing wikipedia importance data')
-            data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
-            if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
-                                                 data_path) > 0:
-                LOG.error('Wikipedia importance dump file not found. '
-                          'Will be using default importances.')
+        if args.continue_at in (None, 'import-from-file'):
+            files = args.get_osm_file_list()
+            if not files:
+                raise UsageError("No input files (use --osm-file).")
 
-        if args.continue_at is None or args.continue_at == 'load-data':
+            if args.continue_at in ('import-from-file', None):
+                # Check if the correct plugins are installed
+                database_import.check_existing_database_plugins(args.config.get_libpq_dsn())
+                LOG.warning('Setting up country tables')
+                country_info.setup_country_tables(args.config.get_libpq_dsn(),
+                                                args.config.lib_dir.data,
+                                                args.no_partitions)
+
+                LOG.warning('Importing OSM data file')
+                database_import.import_osm_data(files,
+                                                args.osm2pgsql_options(0, 1),
+                                                drop=args.no_updates,
+                                                ignore_errors=args.ignore_errors)
+
+                LOG.warning('Importing wikipedia importance data')
+                data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
+                if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
+                                                    data_path) > 0:
+                    LOG.error('Wikipedia importance dump file not found. '
+                            'Calculating importance values of locations will not '
+                            'use Wikipedia importance data.')
+
+                LOG.warning('Importing secondary importance raster data')
+                if refresh.import_secondary_importance(args.config.get_libpq_dsn(),
+                                                    args.project_dir) != 0:
+                    LOG.error('Secondary importance file not imported. '
+                            'Falling back to default ranking.')
+
+                self._setup_tables(args.config, args.reverse_only)
+
+        if args.continue_at in ('import-from-file', 'load-data', None):
             LOG.warning('Initialise tables')
             with connect(args.config.get_libpq_dsn()) as conn:
                 database_import.truncate_data_tables(conn)
@@ -116,15 +144,13 @@ class SetupAll:
         LOG.warning("Setting up tokenizer")
         tokenizer = self._get_tokenizer(args.continue_at, args.config)
 
-        if args.continue_at is None or args.continue_at == 'load-data':
+        if args.continue_at in ('import-from-file', 'load-data', None):
             LOG.warning('Calculate postcodes')
             postcodes.update_postcodes(args.config.get_libpq_dsn(),
                                        args.project_dir, tokenizer)
 
-        if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
-            if args.continue_at is not None and args.continue_at != 'load-data':
-                with connect(args.config.get_libpq_dsn()) as conn:
-                    self._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
+        if args.continue_at in \
+            ('import-from-file', 'load-data', 'indexing', None):
             LOG.warning('Indexing places')
             indexer = Indexer(args.config.get_libpq_dsn(), tokenizer, num_threads)
             indexer.index_full(analyse=not args.index_noanalyse)
@@ -142,7 +168,7 @@ class SetupAll:
         tokenizer.finalize_import(args.config)
 
         LOG.warning('Recompute word counts')
-        tokenizer.update_statistics()
+        tokenizer.update_statistics(args.config, threads=num_threads)
 
         webdir = args.project_dir / 'website'
         LOG.warning('Setup website at %s', webdir)
@@ -181,7 +207,7 @@ class SetupAll:
         """
         from ..tokenizer import factory as tokenizer_factory
 
-        if continue_at is None or continue_at == 'load-data':
+        if continue_at in ('import-from-file', 'load-data', None):
             # (re)initialise the tokenizer data
             return tokenizer_factory.create_tokenizer(config)
 
@@ -189,37 +215,15 @@ class SetupAll:
         return tokenizer_factory.get_tokenizer_for_db(config)
 
 
-    def _create_pending_index(self, conn: Connection, tablespace: str) -> None:
-        """ Add a supporting index for finding places still to be indexed.
-
-            This index is normally created at the end of the import process
-            for later updates. When indexing was partially done, then this
-            index can greatly improve speed going through already indexed data.
-        """
-        if conn.index_exists('idx_placex_pendingsector'):
-            return
-
-        with conn.cursor() as cur:
-            LOG.warning('Creating support index')
-            if tablespace:
-                tablespace = 'TABLESPACE ' + tablespace
-            cur.execute(f"""CREATE INDEX idx_placex_pendingsector
-                            ON placex USING BTREE (rank_address,geometry_sector)
-                            {tablespace} WHERE indexed_status > 0
-                         """)
-        conn.commit()
-
-
     def _finalize_database(self, dsn: str, offline: bool) -> None:
         """ Determine the database date and set the status accordingly.
         """
         with connect(dsn) as conn:
-            if not offline:
-                try:
-                    dbdate = status.compute_database_date(conn)
-                    status.set_status(conn, dbdate)
-                    LOG.info('Database is at %s.', dbdate)
-                except Exception as exc: # pylint: disable=broad-except
-                    LOG.error('Cannot determine date of database: %s', exc)
-
-            properties.set_property(conn, 'database_version', version_str())
+            properties.set_property(conn, 'database_version', str(NOMINATIM_VERSION))
+
+            try:
+                dbdate = status.compute_database_date(conn, offline)
+                status.set_status(conn, dbdate)
+                LOG.info('Database is at %s.', dbdate)
+            except Exception as exc: # pylint: disable=broad-except
+                LOG.error('Cannot determine date of database: %s', exc)