add tests for new data invalidation functions

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

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index 97e2e0374b644c4d7721ed30d38f6e573fab54b5..aacc622b8b634ff284e3f8c0b9c8a105392fd68e 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -1,11 +1,17 @@
+# 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.
 """
 Functions for bringing auxiliary data in the database up-to-date.
 """
-import json
 import logging
 from textwrap import dedent
+from pathlib import Path
 
-from psycopg2.extras import execute_values
+from psycopg2 import sql as pysql
 
 from nominatim.db.utils import execute_file
 from nominatim.db.sql_preprocessor import SQLPreprocessor
@@ -13,12 +19,7 @@ from nominatim.version import NOMINATIM_VERSION
 
 LOG = logging.getLogger()
 
-
-def recompute_word_counts(dsn, sql_dir):
-    """ Compute the frequency of full-word search terms.
-    """
-    execute_file(dsn, sql_dir / 'words_from_search_name.sql')
-
+OSM_TYPE = {'N': 'node', 'W': 'way', 'R': 'relation'}
 
 def _add_address_level_rows_from_entry(rows, entry):
     """ Converts a single entry from the JSON format for address rank
@@ -49,7 +50,7 @@ def load_address_levels(conn, table, levels):
         _add_address_level_rows_from_entry(rows, entry)
 
     with conn.cursor() as cur:
-        cur.execute('DROP TABLE IF EXISTS {}'.format(table))
+        cur.drop_table(table)
 
         cur.execute("""CREATE TABLE {} (country_code varchar(2),
                                         class TEXT,
@@ -57,18 +58,22 @@ def load_address_levels(conn, table, levels):
                                         rank_search SMALLINT,
                                         rank_address SMALLINT)""".format(table))
 
-        execute_values(cur, "INSERT INTO {} VALUES %s".format(table), rows)
+        cur.execute_values(pysql.SQL("INSERT INTO {} VALUES %s")
+                           .format(pysql.Identifier(table)), rows)
 
         cur.execute('CREATE UNIQUE INDEX ON {} (country_code, class, type)'.format(table))
 
     conn.commit()
 
-def load_address_levels_from_file(conn, config_file):
-    """ Replace the `address_levels` table with the contents of the config
-        file.
+
+def load_address_levels_from_config(conn, config):
+    """ Replace the `address_levels` table with the content as
+        defined in the given configuration. Uses the parameter
+        NOMINATIM_ADDRESS_LEVEL_CONFIG to determine the location of the
+        configuration file.
     """
-    with config_file.open('r') as fdesc:
-        load_address_levels(conn, 'address_levels', json.load(fdesc))
+    cfg = config.load_sub_configuration('', config='ADDRESS_LEVEL_CONFIG')
+    load_address_levels(conn, 'address_levels', cfg)
 
 
 def create_functions(conn, config, enable_diff_updates=True, enable_debug=False):
@@ -97,7 +102,7 @@ PHP_CONST_DEFS = (
     ('Database_DSN', 'DATABASE_DSN', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),
-    ('Log_File', 'LOG_FILE', str),
+    ('Log_File', 'LOG_FILE', Path),
     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
@@ -165,7 +170,12 @@ def _quote_php_variable(var_type, config, conf_name):
     if not getattr(config, conf_name):
         return 'false'
 
-    quoted = getattr(config, conf_name).replace("'", "\\'")
+    if var_type == Path:
+        value = str(config.get_path(conf_name))
+    else:
+        value = getattr(config, conf_name)
+
+    quoted = value.replace("'", "\\'")
     return f"'{quoted}'"
 
 
@@ -201,3 +211,28 @@ def setup_website(basedir, config, conn):
             (basedir / script).write_text(template.format('reverse-only-search.php'), 'utf-8')
         else:
             (basedir / script).write_text(template.format(script), 'utf-8')
+
+
+def invalidate_osm_object(osm_type, osm_id, conn, recursive=True):
+    """ Mark the given OSM object for reindexing. When 'recursive' is set
+        to True (the default), then all dependent objects are marked for
+        reindexing as well.
+
+        'osm_type' must be on of 'N' (node), 'W' (way) or 'R' (relation).
+        If the given object does not exist, then nothing happens.
+    """
+    assert osm_type in ('N', 'R', 'W')
+
+    LOG.warning("Invalidating OSM %s %s%s.",
+                OSM_TYPE[osm_type], osm_id,
+                ' and its dependent places' if recursive else '')
+
+    with conn.cursor() as cur:
+        if recursive:
+            sql = """SELECT place_force_update(place_id)
+                     FROM placex WHERE osm_type = %s and osm_id = %s"""
+        else:
+            sql = """UPDATE placex SET indexed_status = 2
+                     WHERE osm_type = %s and osm_id = %s"""
+
+        cur.execute(sql, (osm_type, osm_id))