Merge remote-tracking branch 'upstream/master'

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

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index c00a2e10e0fa56b365991c738e2f455f3ebc51df..561bcf83ce4775f4fd19afa9761b368109f58476 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -1,30 +1,25 @@
+# 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.
 """
 """
 Functions for bringing auxiliary data in the database up-to-date.
 """

-import json

 import logging
 from textwrap import dedent
 import logging
 from textwrap import dedent

+from pathlib import Path

 
 

-from psycopg2.extras import execute_values
+from psycopg2 import sql as pysql

 
 

-from ..db.utils import execute_file
-from ..db.sql_preprocessor import SQLPreprocessor
-from ..version import NOMINATIM_VERSION
+from nominatim.db.utils import execute_file
+from nominatim.db.sql_preprocessor import SQLPreprocessor
+from nominatim.version import version_str

 
 LOG = logging.getLogger()
 
 
 LOG = logging.getLogger()
 

-def update_postcodes(dsn, sql_dir):
-    """ Recalculate postcode centroids and add, remove and update entries in the
-        location_postcode table. `conn` is an opne connection to the database.
-    """
-    execute_file(dsn, sql_dir / 'update-postcodes.sql')
-
-
-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
 
 def _add_address_level_rows_from_entry(rows, entry):
     """ Converts a single entry from the JSON format for address rank


@@ -55,36 +50,42 @@ def load_address_levels(conn, table, levels):
         _add_address_level_rows_from_entry(rows, entry)
 
     with conn.cursor() as cur:
         _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),
+        cur.execute(pysql.SQL("""CREATE TABLE {} (
+                                        country_code varchar(2),

                                         class TEXT,
                                         type TEXT,
                                         rank_search SMALLINT,
                                         class TEXT,
                                         type TEXT,
                                         rank_search SMALLINT,

-                                        rank_address SMALLINT)""".format(table))
+                                        rank_address SMALLINT)
+                              """).format(pysql.Identifier(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))
+        cur.execute(pysql.SQL('CREATE UNIQUE INDEX ON {} (country_code, class, type)')
+                    .format(pysql.Identifier(table)))

 
     conn.commit()
 
 
     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, sqllib_dir,
-                     enable_diff_updates=True, enable_debug=False):
+def create_functions(conn, config, enable_diff_updates=True, enable_debug=False):

     """ (Re)create the PL/pgSQL functions.
     """
     """ (Re)create the PL/pgSQL functions.
     """

-    sql = SQLPreprocessor(conn, config, sqllib_dir)
+    sql = SQLPreprocessor(conn, config)

 
     sql.run_sql_file(conn, 'functions.sql',
 
     sql.run_sql_file(conn, 'functions.sql',

-                     disable_diff_update=not enable_diff_updates,
+                     disable_diff_updates=not enable_diff_updates,

                      debug=enable_debug)
 
 
                      debug=enable_debug)
 
 


@@ -104,15 +105,12 @@ PHP_CONST_DEFS = (
     ('Database_DSN', 'DATABASE_DSN', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),
     ('Database_DSN', 'DATABASE_DSN', str),
     ('Default_Language', 'DEFAULT_LANGUAGE', str),
     ('Log_DB', 'LOG_DB', bool),

-    ('Log_File', 'LOG_FILE', str),
-    ('Max_Word_Frequency', 'MAX_WORD_FREQUENCY', int),
+    ('Log_File', 'LOG_FILE', Path),

     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),
     ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
     ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
     ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
     ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
     ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),

-    ('Term_Normalization_Rules', 'TERM_NORMALIZATION', str),
-    ('Use_Aux_Location_data', 'USE_AUX_LOCATION_DATA', bool),

     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
     ('MapIcon_URL', 'MAPICON_URL', str),
 )
     ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
     ('MapIcon_URL', 'MAPICON_URL', str),
 )


@@ -165,35 +163,78 @@ def recompute_importance(conn):
     conn.commit()
 
 
     conn.commit()
 
 

-def setup_website(basedir, phplib_dir, config):
+def _quote_php_variable(var_type, config, conf_name):
+    if var_type == bool:
+        return 'true' if config.get_bool(conf_name) else 'false'
+
+    if var_type == int:
+        return getattr(config, conf_name)
+
+    if not getattr(config, conf_name):
+        return 'false'
+
+    if var_type == Path:
+        value = str(config.get_path(conf_name))
+    else:
+        value = getattr(config, conf_name)
+
+    quoted = value.replace("'", "\\'")
+    return f"'{quoted}'"
+
+
+def setup_website(basedir, config, conn):

     """ Create the website script stubs.
     """
     if not basedir.exists():
         LOG.info('Creating website directory.')
         basedir.mkdir()
 
     """ Create the website script stubs.
     """
     if not basedir.exists():
         LOG.info('Creating website directory.')
         basedir.mkdir()
 

-    template = dedent("""\
+    template = dedent(f"""\

                       <?php
 
                       @define('CONST_Debug', $_GET['debug'] ?? false);
                       <?php
 
                       @define('CONST_Debug', $_GET['debug'] ?? false);

-                      @define('CONST_LibDir', '{0}');
-                      @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');
+                      @define('CONST_LibDir', '{config.lib_dir.php}');
+                      @define('CONST_TokenizerDir', '{config.project_dir / 'tokenizer'}');
+                      @define('CONST_NominatimVersion', '{version_str()}');

 
 

-                      """.format(phplib_dir, NOMINATIM_VERSION))
+                      """)

 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:
 
     for php_name, conf_name, var_type in PHP_CONST_DEFS:

-        if var_type == bool:
-            varout = 'true' if config.get_bool(conf_name) else 'false'
-        elif var_type == int:
-            varout = getattr(config, conf_name)
-        elif not getattr(config, conf_name):
-            varout = 'false'
-        else:
-            varout = "'{}'".format(getattr(config, conf_name).replace("'", "\\'"))
+        varout = _quote_php_variable(var_type, config, conf_name)
+
+        template += f"@define('CONST_{php_name}', {varout});\n"

 
 

-        template += "@define('CONST_{}', {});\n".format(php_name, varout)
+    template += f"\nrequire_once('{config.lib_dir.php}/website/{{}}');\n"

 
 

-    template += "\nrequire_once('{}/website/{{}}');\n".format(phplib_dir)
+    search_name_table_exists = bool(conn and conn.table_exists('search_name'))

 
     for script in WEBSITE_SCRIPTS:
 
     for script in WEBSITE_SCRIPTS:

-        (basedir / script).write_text(template.format(script), 'utf-8')
+        if not search_name_table_exists and script == 'search.php':
+            (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))