]> git.openstreetmap.org Git - nominatim.git/blobdiff - nominatim/clicmd/setup.py
add type annotations for command line functions
[nominatim.git] / nominatim / clicmd / setup.py
index 92d0694359e62ab48a58a255143b4733dfd316f6..6ffa7afb8babb91c58d7f40c0af16cfbb32f110d 100644 (file)
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.
 """
 Implementation of the 'import' subcommand.
 """
 """
 Implementation of the 'import' subcommand.
 """
+from typing import Optional
+import argparse
 import logging
 from pathlib import Path
 
 import psutil
 
 import logging
 from pathlib import Path
 
 import psutil
 
-from ..tools.exec_utils import run_legacy_script
-from ..db.connection import connect
-from ..db import status, properties
-from ..version import NOMINATIM_VERSION
-from ..errors import UsageError
+from nominatim.config import Configuration
+from nominatim.db.connection import connect, Connection
+from nominatim.db import status, properties
+from nominatim.tokenizer.base import AbstractTokenizer
+from nominatim.version import version_str
+from nominatim.clicmd.args import NominatimArgs
+from nominatim.errors import UsageError
 
 # Do not repeat documentation of subcommand classes.
 # pylint: disable=C0111
 # Using non-top-level imports to avoid eventually unused imports.
 
 # Do not repeat documentation of subcommand classes.
 # pylint: disable=C0111
 # Using non-top-level imports to avoid eventually unused imports.
-# pylint: disable=E0012,C0415
+# pylint: disable=C0415
 
 LOG = logging.getLogger()
 
 class SetupAll:
     """\
     Create a new Nominatim database from an OSM file.
 
 LOG = logging.getLogger()
 
 class SetupAll:
     """\
     Create a new Nominatim database from an OSM file.
+
+    This sub-command sets up a new Nominatim database from scratch starting
+    with creating a new database in Postgresql. The user running this command
+    needs superuser rights on the database.
     """
 
     """
 
-    @staticmethod
-    def add_args(parser):
+    def add_args(self, parser: argparse.ArgumentParser) -> None:
         group_name = parser.add_argument_group('Required arguments')
         group_name = parser.add_argument_group('Required arguments')
-        group = group_name.add_mutually_exclusive_group(required=True)
-        group.add_argument('--osm-file', metavar='FILE',
-                           help='OSM file to be imported.')
-        group.add_argument('--continue', dest='continue_at',
+        group1 = group_name.add_mutually_exclusive_group(required=True)
+        group1.add_argument('--osm-file', metavar='FILE', action='append',
+                           help='OSM file to be imported'
+                                ' (repeat for importing multiple files)')
+        group1.add_argument('--continue', dest='continue_at',
                            choices=['load-data', 'indexing', 'db-postprocess'],
                            help='Continue an import that was interrupted')
                            choices=['load-data', 'indexing', 'db-postprocess'],
                            help='Continue an import that was interrupted')
-        group = parser.add_argument_group('Optional arguments')
-        group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
+        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)')
                            help='Size of cache to be used by osm2pgsql (in MB)')
-        group.add_argument('--reverse-only', action='store_true',
+        group2.add_argument('--reverse-only', action='store_true',
                            help='Do not create tables and indexes for searching')
                            help='Do not create tables and indexes for searching')
-        group.add_argument('--no-partitions', action='store_true',
+        group2.add_argument('--no-partitions', action='store_true',
                            help=("Do not partition search indices "
                                  "(speeds up import of single country extracts)"))
                            help=("Do not partition search indices "
                                  "(speeds up import of single country extracts)"))
-        group.add_argument('--no-updates', action='store_true',
+        group2.add_argument('--no-updates', action='store_true',
                            help="Do not keep tables that are only needed for "
                                 "updating the database later")
                            help="Do not keep tables that are only needed for "
                                 "updating the database later")
-        group = parser.add_argument_group('Expert options')
-        group.add_argument('--ignore-errors', action='store_true',
+        group2.add_argument('--offline', action='store_true',
+                           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')
                            help='Continue import even when errors in SQL are present')
-        group.add_argument('--index-noanalyse', action='store_true',
-                           help='Do not perform analyse operations during index')
+        group3.add_argument('--index-noanalyse', action='store_true',
+                           help='Do not perform analyse operations during index (expert only)')
 
 
 
 
-    @staticmethod
-    def run(args): # pylint: disable=too-many-statements
-        from ..tools import database_import
-        from ..tools import refresh
+    def run(self, args: NominatimArgs) -> int: # pylint: disable=too-many-statements
+        from ..data import country_info
+        from ..tools import database_import, refresh, postcodes, freeze
         from ..indexer.indexer import Indexer
 
         from ..indexer.indexer import Indexer
 
-        if args.osm_file and not Path(args.osm_file).is_file():
-            LOG.fatal("OSM file '%s' does not exist.", args.osm_file)
-            raise UsageError('Cannot access file.')
+        country_info.setup_country_config(args.config)
 
         if args.continue_at is None:
 
         if args.continue_at is None:
+            files = args.get_osm_file_list()
+            if not files:
+                raise UsageError("No input files (use --osm-file).")
+
+            LOG.warning('Creating database')
             database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
             database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
-                                                    args.data_dir,
-                                                    args.no_partitions,
                                                     rouser=args.config.DATABASE_WEBUSER)
 
                                                     rouser=args.config.DATABASE_WEBUSER)
 
-            LOG.warning('Installing database module')
-            with connect(args.config.get_libpq_dsn()) as conn:
-                database_import.install_module(args.module_dir, args.project_dir,
-                                               args.config.DATABASE_MODULE_PATH,
-                                               conn=conn)
+            LOG.warning('Setting up country tables')
+            country_info.setup_country_tables(args.config.get_libpq_dsn(),
+                                              args.data_dir,
+                                              args.no_partitions)
 
             LOG.warning('Importing OSM data file')
 
             LOG.warning('Importing OSM data file')
-            database_import.import_osm_data(Path(args.osm_file),
+            database_import.import_osm_data(files,
                                             args.osm2pgsql_options(0, 1),
                                             drop=args.no_updates,
                                             ignore_errors=args.ignore_errors)
 
                                             args.osm2pgsql_options(0, 1),
                                             drop=args.no_updates,
                                             ignore_errors=args.ignore_errors)
 
-            with connect(args.config.get_libpq_dsn()) as conn:
-                LOG.warning('Create functions (1st pass)')
-                refresh.create_functions(conn, args.config, args.sqllib_dir,
-                                         False, False)
-                LOG.warning('Create tables')
-                database_import.create_tables(conn, args.config, args.sqllib_dir,
-                                              reverse_only=args.reverse_only)
-                refresh.load_address_levels_from_file(conn, Path(args.config.ADDRESS_LEVEL_CONFIG))
-                LOG.warning('Create functions (2nd pass)')
-                refresh.create_functions(conn, args.config, args.sqllib_dir,
-                                         False, False)
-                LOG.warning('Create table triggers')
-                database_import.create_table_triggers(conn, args.config, args.sqllib_dir)
-                LOG.warning('Create partition tables')
-                database_import.create_partition_tables(conn, args.config, args.sqllib_dir)
-                LOG.warning('Create functions (3rd pass)')
-                refresh.create_functions(conn, args.config, args.sqllib_dir,
-                                         False, False)
+            self._setup_tables(args.config, args.reverse_only)
 
             LOG.warning('Importing wikipedia importance data')
             data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
 
             LOG.warning('Importing wikipedia importance data')
             data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
@@ -105,47 +103,122 @@ class SetupAll:
                 LOG.error('Wikipedia importance dump file not found. '
                           'Will be using default importances.')
 
                 LOG.error('Wikipedia importance dump file not found. '
                           'Will be using default importances.')
 
+        if args.continue_at is None or args.continue_at == 'load-data':
             LOG.warning('Initialise tables')
             with connect(args.config.get_libpq_dsn()) as conn:
             LOG.warning('Initialise tables')
             with connect(args.config.get_libpq_dsn()) as conn:
-                database_import.truncate_data_tables(conn, args.config.MAX_WORD_FREQUENCY)
+                database_import.truncate_data_tables(conn)
 
 
-        if args.continue_at is None or args.continue_at == 'load-data':
             LOG.warning('Load data into placex table')
             database_import.load_data(args.config.get_libpq_dsn(),
             LOG.warning('Load data into placex table')
             database_import.load_data(args.config.get_libpq_dsn(),
-                                      args.data_dir,
                                       args.threads or psutil.cpu_count() or 1)
 
                                       args.threads or psutil.cpu_count() or 1)
 
+        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':
             LOG.warning('Calculate postcodes')
             LOG.warning('Calculate postcodes')
-            run_legacy_script('setup.php', '--calculate-postcodes',
-                              nominatim_env=args, throw_on_fail=not args.ignore_errors)
+            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 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)
             LOG.warning('Indexing places')
             LOG.warning('Indexing places')
-            indexer = Indexer(args.config.get_libpq_dsn(),
+            indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
                               args.threads or psutil.cpu_count() or 1)
             indexer.index_full(analyse=not args.index_noanalyse)
 
         LOG.warning('Post-process tables')
         with connect(args.config.get_libpq_dsn()) as conn:
             database_import.create_search_indices(conn, args.config,
                               args.threads or psutil.cpu_count() or 1)
             indexer.index_full(analyse=not args.index_noanalyse)
 
         LOG.warning('Post-process tables')
         with connect(args.config.get_libpq_dsn()) as conn:
             database_import.create_search_indices(conn, args.config,
-                                                  args.sqllib_dir,
                                                   drop=args.no_updates)
             LOG.warning('Create search index for default country names.')
                                                   drop=args.no_updates)
             LOG.warning('Create search index for default country names.')
-            database_import.create_country_names(conn, args.config)
+            country_info.create_country_names(conn, tokenizer,
+                                              args.config.get_str_list('LANGUAGES'))
+            if args.no_updates:
+                freeze.drop_update_tables(conn)
+        tokenizer.finalize_import(args.config)
+
+        LOG.warning('Recompute word counts')
+        tokenizer.update_statistics()
 
         webdir = args.project_dir / 'website'
         LOG.warning('Setup website at %s', webdir)
 
         webdir = args.project_dir / 'website'
         LOG.warning('Setup website at %s', webdir)
-        refresh.setup_website(webdir, args.phplib_dir, args.config)
-
         with connect(args.config.get_libpq_dsn()) as conn:
         with connect(args.config.get_libpq_dsn()) as conn:
-            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)
+            refresh.setup_website(webdir, args.config, conn)
 
 
-            properties.set_property(conn, 'database_version',
-                                    '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
+        self._finalize_database(args.config.get_libpq_dsn(), args.offline)
 
         return 0
 
         return 0
+
+
+    def _setup_tables(self, config: Configuration, reverse_only: bool) -> None:
+        """ Set up the basic database layout: tables, indexes and functions.
+        """
+        from ..tools import database_import, refresh
+
+        with connect(config.get_libpq_dsn()) as conn:
+            LOG.warning('Create functions (1st pass)')
+            refresh.create_functions(conn, config, False, False)
+            LOG.warning('Create tables')
+            database_import.create_tables(conn, config, reverse_only=reverse_only)
+            refresh.load_address_levels_from_config(conn, config)
+            LOG.warning('Create functions (2nd pass)')
+            refresh.create_functions(conn, config, False, False)
+            LOG.warning('Create table triggers')
+            database_import.create_table_triggers(conn, config)
+            LOG.warning('Create partition tables')
+            database_import.create_partition_tables(conn, config)
+            LOG.warning('Create functions (3rd pass)')
+            refresh.create_functions(conn, config, False, False)
+
+
+    def _get_tokenizer(self, continue_at: Optional[str],
+                       config: Configuration) -> AbstractTokenizer:
+        """ Set up a new tokenizer or load an already initialised one.
+        """
+        from ..tokenizer import factory as tokenizer_factory
+
+        if continue_at is None or continue_at == 'load-data':
+            # (re)initialise the tokenizer data
+            return tokenizer_factory.create_tokenizer(config)
+
+        # just load the tokenizer
+        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())