postcodes: add support for optional spaces

[nominatim.git] / nominatim / clicmd / refresh.py

diff --git a/nominatim/clicmd/refresh.py b/nominatim/clicmd/refresh.py
index fbc23350c7212478f8c8b0c9fa484ec154f9f070..ecc7498eff7ef0b448a274e8158171f593ccedd7 100644 (file)
--- a/nominatim/clicmd/refresh.py
+++ b/nominatim/clicmd/refresh.py
@@ -1,6 +1,13 @@
+# 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 'refresh' subcommand.
 """
 """
 Implementation of 'refresh' subcommand.
 """

+from argparse import ArgumentTypeError

 import logging
 from pathlib import Path
 
 import logging
 from pathlib import Path
 


@@ -13,11 +20,27 @@ from nominatim.db.connection import connect
 
 LOG = logging.getLogger()
 
 
 LOG = logging.getLogger()
 

+def _parse_osm_object(obj):
+    """ Parse the given argument into a tuple of OSM type and ID.
+        Raises an ArgumentError if the format is not recognized.
+    """
+    if len(obj) < 2 or obj[0].lower() not in 'nrw' or not obj[1:].isdigit():
+        raise ArgumentTypeError("Cannot parse OSM ID. Expect format: [N|W|R]<id>.")
+
+    return (obj[0].upper(), int(obj[1:]))
+
+

 class UpdateRefresh:
     """\
     Recompute auxiliary data used by the indexing process.
 
 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
     """
     def __init__(self):
         self.tokenizer = None


@@ -27,6 +50,8 @@ class UpdateRefresh:
         group = parser.add_argument_group('Data arguments')
         group.add_argument('--postcodes', action='store_true',
                            help='Update postcode centroid table')
         group = parser.add_argument_group('Data arguments')
         group.add_argument('--postcodes', action='store_true',
                            help='Update postcode centroid table')

+        group.add_argument('--word-tokens', action='store_true',
+                           help='Clean up search terms')

         group.add_argument('--word-counts', action='store_true',
                            help='Compute frequency of full-word search terms')
         group.add_argument('--address-levels', action='store_true',
         group.add_argument('--word-counts', action='store_true',
                            help='Compute frequency of full-word search terms')
         group.add_argument('--address-levels', action='store_true',


@@ -34,11 +59,20 @@ 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',
         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',
                            help='Refresh the directory that serves the scripts for the web API')
         group.add_argument('--importance', action='store_true',
                            help='Recompute place importances (expensive!)')
         group.add_argument('--website', action='store_true',
                            help='Refresh the directory that serves the scripts for the web API')

+        group.add_argument('--data-object', action='append',
+                           type=_parse_osm_object, metavar='OBJECT',
+                           help='Mark the given OSM object as requiring an update'
+                                ' (format: [NWR]<id>)')
+        group.add_argument('--data-area', action='append',
+                           type=_parse_osm_object, metavar='OBJECT',
+                           help='Mark the area around the given OSM object as requiring an update'
+                                ' (format: [NWR]<id>)')
+

         group = parser.add_argument_group('Arguments for function refresh')
         group.add_argument('--no-diff-updates', action='store_false', dest='diffs',
                            help='Do not enable code for propagating updates')
         group = parser.add_argument_group('Arguments for function refresh')
         group.add_argument('--no-diff-updates', action='store_false', dest='diffs',
                            help='Do not enable code for propagating updates')


@@ -46,7 +80,7 @@ class UpdateRefresh:
                            help='Enable debug warning statements in functions')
 
 
                            help='Enable debug warning statements in functions')
 
 

-    def run(self, args):
+    def run(self, args): #pylint: disable=too-many-branches

         from ..tools import refresh, postcodes
         from ..indexer.indexer import Indexer
 
         from ..tools import refresh, postcodes
         from ..indexer.indexer import Indexer
 


@@ -61,18 +95,22 @@ class UpdateRefresh:
                                   args.threads or 1)
                 indexer.index_postcodes()
             else:
                                   args.threads or 1)
                 indexer.index_postcodes()
             else:

-                LOG.error("The place table doesn\'t exist. " \
+                LOG.error("The place table doesn't exist. "

                           "Postcode updates on a frozen database is not possible.")
 
                           "Postcode updates on a frozen database is not possible.")
 

+        if args.word_tokens:
+            LOG.warning('Updating word tokens')
+            tokenizer = self._get_tokenizer(args.config)
+            tokenizer.update_word_tokens()
+

         if args.word_counts:
         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:
 
         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:
             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')
 
         if args.functions:
             LOG.warning('Create functions')


@@ -99,9 +137,21 @@ class UpdateRefresh:
         if args.website:
             webdir = args.project_dir / 'website'
             LOG.warning('Setting up website directory at %s', webdir)
         if args.website:
             webdir = args.project_dir / 'website'
             LOG.warning('Setting up website directory at %s', webdir)

+            # This is a little bit hacky: call the tokenizer setup, so that
+            # the tokenizer directory gets repopulated as well, in case it
+            # wasn't there yet.
+            self._get_tokenizer(args.config)

             with connect(args.config.get_libpq_dsn()) as conn:
                 refresh.setup_website(webdir, args.config, conn)
 
             with connect(args.config.get_libpq_dsn()) as conn:
                 refresh.setup_website(webdir, args.config, conn)
 

+        if args.data_object or args.data_area:
+            with connect(args.config.get_libpq_dsn()) as conn:
+                for obj in args.data_object or []:
+                    refresh.invalidate_osm_object(*obj, conn, recursive=False)
+                for obj in args.data_area or []:
+                    refresh.invalidate_osm_object(*obj, conn, recursive=True)
+                conn.commit()
+

         return 0
 
 
         return 0