add new command for cleaning word tokens

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

diff --git a/nominatim/clicmd/refresh.py b/nominatim/clicmd/refresh.py
index e696e7b6291d5df03d3480609cff8d8a7d8a5532..c741dcf63632fc0c01d8592a66f46d3be0c8bdbd 100644 (file)
--- a/nominatim/clicmd/refresh.py
+++ b/nominatim/clicmd/refresh.py
@@ -1,3 +1,9 @@
+# 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.
 """
@@ -17,14 +23,24 @@ 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):
         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',
@@ -32,7 +48,7 @@ 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',
@@ -43,38 +59,44 @@ class UpdateRefresh:
         group.add_argument('--enable-debug-statements', action='store_true',
                            help='Enable debug warning statements in functions')
 
-    @staticmethod
-    def run(args):
+
+    def run(self, args):
         from ..tools import refresh, postcodes
-        from ..tokenizer import factory as tokenizer_factory
         from ..indexer.indexer import Indexer
 
-        tokenizer = tokenizer_factory.get_tokenizer_for_db(args.config)
 
         if args.postcodes:
-            LOG.warning("Update postcodes centroid")
-            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()
+            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_tokens:
+            tokenizer = self._get_tokenizer(args.config)
+            tokenizer.update_word_tokens()
 
         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.diffs, args.enable_debug_statements)
-                tokenizer.update_sql_functions(args.config)
+                self._get_tokenizer(args.config).update_sql_functions(args.config)
 
         if args.wiki_data:
             data_path = Path(args.config.WIKIPEDIA_DATA_PATH
@@ -96,4 +118,14 @@ class UpdateRefresh:
             LOG.warning('Setting up website directory at %s', webdir)
             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