X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/3206bf59df0213d24bd3e11df7dd2abaebf89911..83054af46fb21ea40c0bebb2c8a528937e5a49a0:/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 ef4e85bd..8906e03e 100644 --- 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. @@ -7,19 +13,36 @@ The phrases already present in the database which are not valids anymore are removed. """ +from typing import Iterable, Tuple, Mapping, Sequence, Optional, Set import logging -import os -from os.path import isfile -from pathlib import Path 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.config import Configuration +from nominatim.db.connection import Connection from nominatim.tools.special_phrases.importer_statistics import SpecialPhrasesImporterStatistics +from nominatim.tools.special_phrases.special_phrase import SpecialPhrase +from nominatim.tokenizer.base import AbstractTokenizer +from nominatim.typing import Protocol LOG = logging.getLogger() + +def _classtype_table(phrase_class: str, phrase_type: str) -> str: + """ Return the name of the table for the given class and type. + """ + return f'place_classtype_{phrase_class}_{phrase_type}' + + +class SpecialPhraseLoader(Protocol): + """ Protocol for classes implementing a loader for special phrases. + """ + + def generate_phrases(self) -> Iterable[SpecialPhrase]: + """ Generates all special phrase terms this loader can produce. + """ + + class SPImporter(): # pylint: disable-msg=too-many-instance-attributes """ @@ -27,22 +50,22 @@ class SPImporter(): 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: Configuration, conn: Connection, + sp_loader: SpecialPhraseLoader) -> None: self.config = config - self.phplib_dir = phplib_dir - self.db_connection = db_connection + self.db_connection = conn self.sp_loader = sp_loader self.statistics_handler = SpecialPhrasesImporterStatistics() self.black_list, self.white_list = self._load_white_and_black_lists() self.sanity_check_pattern = re.compile(r'^\w+$') # This set will contain all existing phrases to be added. # It contains tuples with the following format: (lable, class, type, operator) - self.word_phrases = set() - #This set will contain all existing place_classtype tables which doesn't match any - #special phrases class/type on the wiki. - self.table_phrases_to_delete = set() + self.word_phrases: Set[Tuple[str, str, str, str]] = set() + # This set will contain all existing place_classtype tables which doesn't match any + # special phrases class/type on the wiki. + self.table_phrases_to_delete: Set[str] = set() - def import_phrases(self, tokenizer, should_replace): + def import_phrases(self, tokenizer: AbstractTokenizer, should_replace: bool) -> None: """ Iterate through all SpecialPhrases extracted from the loader and import them into the database. @@ -54,16 +77,15 @@ class SPImporter(): LOG.warning('Special phrases importation starting') self._fetch_existing_place_classtype_tables() - #Store pairs of class/type for further processing + # 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.update(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) + self._create_classtype_table_and_indexes(class_type_pairs) if should_replace: self._remove_non_existent_tables_from_db() self.db_connection.commit() @@ -75,7 +97,7 @@ class SPImporter(): self.statistics_handler.notify_import_done() - def _fetch_existing_place_classtype_tables(self): + def _fetch_existing_place_classtype_tables(self) -> None: """ Fetch existing place_classtype tables. Fill the table_phrases_to_delete set of the class. @@ -91,20 +113,16 @@ class SPImporter(): for row in db_cursor: self.table_phrases_to_delete.add(row[0]) - def _load_white_and_black_lists(self): + def _load_white_and_black_lists(self) \ + -> Tuple[Mapping[str, Sequence[str]], Mapping[str, Sequence[str]]]: """ Load white and black lists from phrases-settings.json. """ - settings_path = (self.config.config_dir / 'phrase-settings.json').resolve() - - if self.config.PHRASE_CONFIG: - settings_path = self._convert_php_settings_if_needed(self.config.PHRASE_CONFIG) + settings = self.config.load_sub_configuration('phrase-settings.json') - with settings_path.open("r") as json_settings: - settings = json.load(json_settings) return settings['blackList'], settings['whiteList'] - def _check_sanity(self, phrase): + def _check_sanity(self, phrase: SpecialPhrase) -> bool: """ Check sanity of given inputs in case somebody added garbage in the wiki. If a bad class/type is detected the system will exit with an error. @@ -118,24 +136,24 @@ class SPImporter(): return False return True - def _process_phrase(self, phrase): + def _process_phrase(self, phrase: SpecialPhrase) -> Optional[Tuple[str, str]]: """ Processes the given phrase by checking black and white list and sanity. Return the class/type pair corresponding to the phrase. """ - #blacklisting: disallow certain class/type combinations + # blacklisting: disallow certain class/type combinations if phrase.p_class in self.black_list.keys() \ and phrase.p_type in self.black_list[phrase.p_class]: return None - #whitelisting: if class is in whitelist, allow only tags in the list + # whitelisting: if class is in whitelist, allow only tags in the list if phrase.p_class in self.white_list.keys() \ and phrase.p_type not in self.white_list[phrase.p_class]: return None - #sanity check, in case somebody added garbage in the wiki + # sanity check, in case somebody added garbage in the wiki if not self._check_sanity(phrase): self.statistics_handler.notify_one_phrase_invalid() return None @@ -143,10 +161,11 @@ class SPImporter(): self.word_phrases.add((phrase.p_label, phrase.p_class, phrase.p_type, phrase.p_operator)) - return set({(phrase.p_class, phrase.p_type)}) + return (phrase.p_class, phrase.p_type) - def _create_place_classtype_table_and_indexes(self, class_type_pairs): + def _create_classtype_table_and_indexes(self, + class_type_pairs: Iterable[Tuple[str, str]]) -> None: """ Create table place_classtype for each given pair. Also create indexes on place_id and centroid. @@ -155,7 +174,7 @@ class SPImporter(): sql_tablespace = self.config.TABLESPACE_AUX_DATA if sql_tablespace: - sql_tablespace = ' TABLESPACE '+sql_tablespace + sql_tablespace = ' TABLESPACE ' + sql_tablespace with self.db_connection.cursor() as db_cursor: db_cursor.execute("CREATE INDEX idx_placex_classtype ON placex (class, type)") @@ -164,23 +183,23 @@ class SPImporter(): phrase_class = pair[0] phrase_type = pair[1] - table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type) + table_name = _classtype_table(phrase_class, phrase_type) if table_name in self.table_phrases_to_delete: self.statistics_handler.notify_one_table_ignored() - #Remove this table from the ones to delete as it match a class/type - #still existing on the special phrases of the wiki. + # Remove this table from the ones to delete as it match a + # class/type still existing on the special phrases of the wiki. self.table_phrases_to_delete.remove(table_name) - #So dont need to create the table and indexes. + # So don't need to create the table and indexes. continue - #Table creation + # Table creation self._create_place_classtype_table(sql_tablespace, phrase_class, phrase_type) - #Indexes creation + # Indexes creation self._create_place_classtype_indexes(sql_tablespace, phrase_class, phrase_type) - #Grant access on read to the web user. + # Grant access on read to the web user. self._grant_access_to_webuser(phrase_class, phrase_type) self.statistics_handler.notify_one_table_created() @@ -189,94 +208,67 @@ class SPImporter(): db_cursor.execute("DROP INDEX idx_placex_classtype") - def _create_place_classtype_table(self, sql_tablespace, phrase_class, phrase_type): + def _create_place_classtype_table(self, sql_tablespace: str, + phrase_class: str, phrase_type: str) -> None: """ - 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 = 'place_classtype_{}_{}'.format(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))) - - - def _create_place_classtype_indexes(self, sql_tablespace, phrase_class, phrase_type): + table_name = _classtype_table(phrase_class, 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: str, + phrase_class: str, phrase_type: str) -> None: """ Create indexes on centroid and place_id for the place_classtype table. """ - index_prefix = 'idx_place_classtype_{}_{}_'.format(phrase_class, phrase_type) - base_table = 'place_classtype_{}_{}'.format(phrase_class, phrase_type) - #Index on centroid + 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: - 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'), - Identifier(base_table)), sql_tablespace) + Identifier(base_table), + SQL(sql_tablespace))) - #Index on place_id + # 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'), - 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: str, phrase_type: str) -> None: """ Grant access on read to the table place_classtype for the webuser. """ - table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type) + table_name = _classtype_table(phrase_class, phrase_type) with self.db_connection.cursor() as db_cursor: db_cursor.execute(SQL("""GRANT SELECT ON {} TO {}""") .format(Identifier(table_name), Identifier(self.config.DATABASE_WEBUSER))) - def _remove_non_existent_tables_from_db(self): + def _remove_non_existent_tables_from_db(self) -> None: """ Remove special phrases which doesn't exist on the wiki anymore. Delete the place_classtype tables. """ LOG.warning('Cleaning database...') - #Array containing all queries to execute. Contain tuples of format (query, parameters) - queries_parameters = [] - - #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, ())) + # Delete place_classtype tables corresponding to class/type which + # are not on the wiki anymore. 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.') - return json_file_path - except subprocess.CalledProcessError: - LOG.error('Error while converting %s to json.', file_path) - raise - else: - return json_file_path + for table in self.table_phrases_to_delete: + self.statistics_handler.notify_one_table_deleted() + db_cursor.drop_table(table)