X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/e14e7c6235a40dbec451146bcb3aaec013d659c9..810056349f0dde9e0ddeb6655cbf1aa48399b3a5:/nominatim/clicmd/refresh.py diff --git a/nominatim/clicmd/refresh.py b/nominatim/clicmd/refresh.py index 9dca4e42..82a61f54 100644 --- a/nominatim/clicmd/refresh.py +++ b/nominatim/clicmd/refresh.py @@ -4,7 +4,7 @@ Implementation of 'refresh' subcommand. import logging from pathlib import Path -from ..db.connection import connect +from nominatim.db.connection import connect # Do not repeat documentation of subcommand classes. # pylint: disable=C0111 @@ -17,8 +17,16 @@ class UpdateRefresh: """\ Recompute auxiliary data used by the indexing process. - These functions must not be run in parallel with other update commands. + This sub-commands updates various static data and functions in the database. + It usually needs to be run after changing various aspects of the + configuration. The configuration documentation will mention the exact + command to use in such case. + + Warning: the 'update' command must not be run in parallel with other update + commands like 'replication' or 'add-data'. """ + def __init__(self): + self.tokenizer = None @staticmethod def add_args(parser): @@ -32,7 +40,7 @@ class UpdateRefresh: group.add_argument('--functions', action='store_true', help='Update the PL/pgSQL functions in the database') group.add_argument('--wiki-data', action='store_true', - help='Update Wikipedia/data importance numbers.') + help='Update Wikipedia/data importance numbers') group.add_argument('--importance', action='store_true', help='Recompute place importances (expensive!)') group.add_argument('--website', action='store_true', @@ -43,29 +51,40 @@ class UpdateRefresh: group.add_argument('--enable-debug-statements', action='store_true', help='Enable debug warning statements in functions') - @staticmethod - def run(args): - from ..tools import refresh + + def run(self, args): + from ..tools import refresh, postcodes + from ..indexer.indexer import Indexer + if args.postcodes: - LOG.warning("Update postcodes centroid") - refresh.update_postcodes(args.config.get_libpq_dsn(), args.sqllib_dir) + if postcodes.can_compute(args.config.get_libpq_dsn()): + LOG.warning("Update postcodes centroid") + tokenizer = self._get_tokenizer(args.config) + postcodes.update_postcodes(args.config.get_libpq_dsn(), + args.project_dir, tokenizer) + indexer = Indexer(args.config.get_libpq_dsn(), tokenizer, + args.threads or 1) + indexer.index_postcodes() + else: + LOG.error("The place table doesn't exist. " + "Postcode updates on a frozen database is not possible.") if args.word_counts: - LOG.warning('Recompute frequency of full-word search terms') - refresh.recompute_word_counts(args.config.get_libpq_dsn(), args.sqllib_dir) + LOG.warning('Recompute word statistics') + self._get_tokenizer(args.config).update_statistics() if args.address_levels: - cfg = Path(args.config.ADDRESS_LEVEL_CONFIG) - LOG.warning('Updating address levels from %s', cfg) + LOG.warning('Updating address levels') with connect(args.config.get_libpq_dsn()) as conn: - refresh.load_address_levels_from_file(conn, cfg) + refresh.load_address_levels_from_config(conn, args.config) if args.functions: LOG.warning('Create functions') with connect(args.config.get_libpq_dsn()) as conn: - refresh.create_functions(conn, args.config, args.sqllib_dir, + refresh.create_functions(conn, args.config, args.diffs, args.enable_debug_statements) + self._get_tokenizer(args.config).update_sql_functions(args.config) if args.wiki_data: data_path = Path(args.config.WIKIPEDIA_DATA_PATH @@ -85,6 +104,16 @@ class UpdateRefresh: if args.website: webdir = args.project_dir / 'website' LOG.warning('Setting up website directory at %s', webdir) - refresh.setup_website(webdir, args.phplib_dir, args.config) + with connect(args.config.get_libpq_dsn()) as conn: + refresh.setup_website(webdir, args.config, conn) return 0 + + + def _get_tokenizer(self, config): + if self.tokenizer is None: + from ..tokenizer import factory as tokenizer_factory + + self.tokenizer = tokenizer_factory.get_tokenizer_for_db(config) + + return self.tokenizer