Merge pull request #2732 from lonvia/fix-ordering-address-parts

[nominatim.git] / nominatim / tools / special_phrases / sp_importer.py

diff --git a/nominatim/tools/special_phrases/sp_importer.py b/nominatim/tools/special_phrases/sp_importer.py
index a26ea8a6201ae8b10891a85e457fbbe12778e902..31bbc3551cfed82a086feb25953ea073d0a4c011 100644 (file)
--- a/nominatim/tools/special_phrases/sp_importer.py
+++ b/nominatim/tools/special_phrases/sp_importer.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.

 """
     Module containing the class handling the import
     of the special phrases.
 """
     Module containing the class handling the import
     of the special phrases.


@@ -8,15 +14,9 @@
     valids anymore are removed.
 """
 import logging
     valids anymore are removed.
 """
 import logging

-import os
-from os.path import isfile
-from pathlib import Path

 import re
 import re

-import subprocess
-import json

 
 

-from psycopg2.sql import Identifier, Literal, SQL
-from nominatim.errors import UsageError
+from psycopg2.sql import Identifier, SQL

 from nominatim.tools.special_phrases.importer_statistics import SpecialPhrasesImporterStatistics
 
 LOG = logging.getLogger()
 from nominatim.tools.special_phrases.importer_statistics import SpecialPhrasesImporterStatistics
 
 LOG = logging.getLogger()


@@ -33,9 +33,8 @@ class SPImporter():
 
         Take a sp loader which load the phrases from an external source.
     """
 
         Take a sp loader which load the phrases from an external source.
     """

-    def __init__(self, config, phplib_dir, db_connection, sp_loader) -> None:
+    def __init__(self, config, db_connection, sp_loader) -> None:

         self.config = config
         self.config = config

-        self.phplib_dir = phplib_dir

         self.db_connection = db_connection
         self.sp_loader = sp_loader
         self.statistics_handler = SpecialPhrasesImporterStatistics()
         self.db_connection = db_connection
         self.sp_loader = sp_loader
         self.statistics_handler = SpecialPhrasesImporterStatistics()


@@ -63,11 +62,10 @@ class SPImporter():
         # Store pairs of class/type for further processing
         class_type_pairs = set()
 
         # Store pairs of class/type for further processing
         class_type_pairs = set()
 

-        for loaded_phrases in self.sp_loader:
-            for phrase in loaded_phrases:
-                result = self._process_phrase(phrase)
-                if result:
-                    class_type_pairs.add(result)
+        for phrase in self.sp_loader.generate_phrases():
+            result = self._process_phrase(phrase)
+            if result:
+                class_type_pairs.add(result)

 
         self._create_place_classtype_table_and_indexes(class_type_pairs)
         if should_replace:
 
         self._create_place_classtype_table_and_indexes(class_type_pairs)
         if should_replace:


@@ -101,13 +99,8 @@ class SPImporter():
         """
             Load white and black lists from phrases-settings.json.
         """
         """
             Load white and black lists from phrases-settings.json.
         """

-        settings_path = (self.config.config_dir / 'phrase-settings.json').resolve()
+        settings = self.config.load_sub_configuration('phrase-settings.json')

 
 

-        if self.config.PHRASE_CONFIG:
-            settings_path = self._convert_php_settings_if_needed(self.config.PHRASE_CONFIG)
-
-        with settings_path.open("r") as json_settings:
-            settings = json.load(json_settings)

         return settings['blackList'], settings['whiteList']
 
     def _check_sanity(self, phrase):
         return settings['blackList'], settings['whiteList']
 
     def _check_sanity(self, phrase):


@@ -197,39 +190,41 @@ class SPImporter():
 
     def _create_place_classtype_table(self, sql_tablespace, phrase_class, phrase_type):
         """
 
     def _create_place_classtype_table(self, sql_tablespace, phrase_class, phrase_type):
         """

-            Create table place_classtype of the given phrase_class/phrase_type if doesn't exit.
+            Create table place_classtype of the given phrase_class/phrase_type
+            if doesn't exit.

         """
         table_name = _classtype_table(phrase_class, phrase_type)
         """
         table_name = _classtype_table(phrase_class, phrase_type)

-        with self.db_connection.cursor() as db_cursor:
-            db_cursor.execute(SQL("""
-                    CREATE TABLE IF NOT EXISTS {{}} {}
-                    AS SELECT place_id AS place_id,st_centroid(geometry) AS centroid FROM placex
-                    WHERE class = {{}} AND type = {{}}""".format(sql_tablespace))
-                              .format(Identifier(table_name), Literal(phrase_class),
-                                      Literal(phrase_type)))
+        with self.db_connection.cursor() as cur:
+            cur.execute(SQL("""CREATE TABLE IF NOT EXISTS {} {} AS
+                                 SELECT place_id AS place_id,
+                                        st_centroid(geometry) AS centroid
+                                 FROM placex
+                                 WHERE class = %s AND type = %s
+                             """).format(Identifier(table_name), SQL(sql_tablespace)),
+                        (phrase_class, phrase_type))

 
 
     def _create_place_classtype_indexes(self, sql_tablespace, phrase_class, phrase_type):
         """
             Create indexes on centroid and place_id for the place_classtype table.
         """
 
 
     def _create_place_classtype_indexes(self, sql_tablespace, phrase_class, phrase_type):
         """
             Create indexes on centroid and place_id for the place_classtype table.
         """

-        index_prefix = 'idx_place_classtype_{}_{}_'.format(phrase_class, phrase_type)
+        index_prefix = f'idx_place_classtype_{phrase_class}_{phrase_type}_'

         base_table = _classtype_table(phrase_class, phrase_type)
         # Index on centroid
         if not self.db_connection.index_exists(index_prefix + 'centroid'):
             with self.db_connection.cursor() as db_cursor:
         base_table = _classtype_table(phrase_class, phrase_type)
         # Index on centroid
         if not self.db_connection.index_exists(index_prefix + 'centroid'):
             with self.db_connection.cursor() as db_cursor:

-                db_cursor.execute(SQL("""
-                    CREATE INDEX {{}} ON {{}} USING GIST (centroid) {}""".format(sql_tablespace))
+                db_cursor.execute(SQL("CREATE INDEX {} ON {} USING GIST (centroid) {}")

                                   .format(Identifier(index_prefix + 'centroid'),
                                   .format(Identifier(index_prefix + 'centroid'),

-                                          Identifier(base_table)), sql_tablespace)
+                                          Identifier(base_table),
+                                          SQL(sql_tablespace)))

 
         # Index on place_id
         if not self.db_connection.index_exists(index_prefix + 'place_id'):
             with self.db_connection.cursor() as db_cursor:
 
         # Index on place_id
         if not self.db_connection.index_exists(index_prefix + 'place_id'):
             with self.db_connection.cursor() as db_cursor:

-                db_cursor.execute(SQL(
-                    """CREATE INDEX {{}} ON {{}} USING btree(place_id) {}""".format(sql_tablespace))
+                db_cursor.execute(SQL("CREATE INDEX {} ON {} USING btree(place_id) {}")

                                   .format(Identifier(index_prefix + 'place_id'),
                                   .format(Identifier(index_prefix + 'place_id'),

-                                          Identifier(base_table)))
+                                          Identifier(base_table),
+                                          SQL(sql_tablespace)))

 
 
     def _grant_access_to_webuser(self, phrase_class, phrase_type):
 
 
     def _grant_access_to_webuser(self, phrase_class, phrase_type):


@@ -248,42 +243,10 @@ class SPImporter():
             Delete the place_classtype tables.
         """
         LOG.warning('Cleaning database...')
             Delete the place_classtype tables.
         """
         LOG.warning('Cleaning database...')

-        # Array containing all queries to execute.
-        # Contains tuples of format (query, parameters)
-        queries_parameters = []

 
         # Delete place_classtype tables corresponding to class/type which
         # are not on the wiki anymore.
 
         # Delete place_classtype tables corresponding to class/type which
         # are not on the wiki anymore.

-        for table in self.table_phrases_to_delete:
-            self.statistics_handler.notify_one_table_deleted()
-            query = SQL('DROP TABLE IF EXISTS {}').format(Identifier(table))
-            queries_parameters.append((query, ()))
-

         with self.db_connection.cursor() as db_cursor:
         with self.db_connection.cursor() as db_cursor:

-            for query, parameters in queries_parameters:
-                db_cursor.execute(query, parameters)
-
-    def _convert_php_settings_if_needed(self, file_path):
-        """
-            Convert php settings file of special phrases to json file if it is still in php format.
-        """
-        if not isfile(file_path):
-            raise UsageError(str(file_path) + ' is not a valid file.')
-
-        file, extension = os.path.splitext(file_path)
-        json_file_path = Path(file + '.json').resolve()
-
-        if extension not in ('.php', '.json'):
-            raise UsageError('The custom NOMINATIM_PHRASE_CONFIG file has not a valid extension.')
-
-        if extension == '.php' and not isfile(json_file_path):
-            try:
-                subprocess.run(['/usr/bin/env', 'php', '-Cq',
-                                (self.phplib_dir / 'migration/PhraseSettingsToJson.php').resolve(),
-                                file_path], check=True)
-                LOG.warning('special_phrase configuration file has been converted to json.')
-            except subprocess.CalledProcessError:
-                LOG.error('Error while converting %s to json.', file_path)
-                raise
-
-        return json_file_path
+            for table in self.table_phrases_to_delete:
+                self.statistics_handler.notify_one_table_deleted()
+                db_cursor.drop_table(table)