X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/5071710db7832d13341230c9942bc7030a0ec6dc..2dafc4cf4fc46ea4be293a458d565fbff645ac28:/nominatim/tools/replication.py?ds=sidebyside diff --git a/nominatim/tools/replication.py b/nominatim/tools/replication.py index c167a49f..846b9c34 100644 --- a/nominatim/tools/replication.py +++ b/nominatim/tools/replication.py @@ -1,26 +1,40 @@ +# 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 updating a database from a replication source. """ +from typing import ContextManager, MutableMapping, Any, Generator, cast, Iterator +from contextlib import contextmanager import datetime as dt from enum import Enum import logging import time +import types +import urllib.request as urlrequest +import requests from nominatim.db import status +from nominatim.db.connection import Connection from nominatim.tools.exec_utils import run_osm2pgsql from nominatim.errors import UsageError try: from osmium.replication.server import ReplicationServer from osmium import WriteHandler + from osmium import version as pyo_version except ImportError as exc: - logging.getLogger().fatal("pyosmium not installed. Replication functions not available.\n" - "To install pyosmium via pip: pip3 install osmium") + logging.getLogger().critical("pyosmium not installed. Replication functions not available.\n" + "To install pyosmium via pip: pip3 install osmium") raise UsageError("replication tools not available") from exc LOG = logging.getLogger() -def init_replication(conn, base_url): +def init_replication(conn: Connection, base_url: str, + socket_timeout: int = 60) -> None: """ Set up replication for the server at the given base URL. """ LOG.info("Using replication source: %s", base_url) @@ -29,9 +43,8 @@ def init_replication(conn, base_url): # margin of error to make sure we get all data date -= dt.timedelta(hours=3) - repl = ReplicationServer(base_url) - - seq = repl.timestamp_to_sequence(date) + with _make_replication_server(base_url, socket_timeout) as repl: + seq = repl.timestamp_to_sequence(date) if seq is None: LOG.fatal("Cannot reach the configured replication service '%s'.\n" @@ -41,10 +54,11 @@ def init_replication(conn, base_url): status.set_status(conn, date=date, seq=seq) - LOG.warning("Updates intialised at sequence %s (%s)", seq, date) + LOG.warning("Updates initialised at sequence %s (%s)", seq, date) -def check_for_updates(conn, base_url): +def check_for_updates(conn: Connection, base_url: str, + socket_timeout: int = 60) -> int: """ Check if new data is available from the replication service at the given base URL. """ @@ -55,7 +69,8 @@ def check_for_updates(conn, base_url): "Please run 'nominatim replication --init' first.") return 254 - state = ReplicationServer(base_url).get_state_info() + with _make_replication_server(base_url, socket_timeout) as repl: + state = repl.get_state_info() if state is None: LOG.error("Cannot get state for URL %s.", base_url) @@ -77,7 +92,8 @@ class UpdateState(Enum): NO_CHANGES = 3 -def update(conn, options): +def update(conn: Connection, options: MutableMapping[str, Any], + socket_timeout: int = 60) -> UpdateState: """ Update database from the next batch of data. Returns the state of updates according to `UpdateState`. """ @@ -88,6 +104,8 @@ def update(conn, options): "Please run 'nominatim replication --init' first.") raise UsageError("Replication not set up.") + assert startdate is not None + if not indexed and options['indexed_only']: LOG.info("Skipping update. There is data that needs indexing.") return UpdateState.MORE_PENDING @@ -103,24 +121,81 @@ def update(conn, options): options['import_file'].unlink() # Read updates into file. - repl = ReplicationServer(options['base_url']) + with _make_replication_server(options['base_url'], socket_timeout) as repl: + outhandler = WriteHandler(str(options['import_file'])) + endseq = repl.apply_diffs(outhandler, startseq + 1, + max_size=options['max_diff_size'] * 1024) + outhandler.close() + + if endseq is None: + return UpdateState.NO_CHANGES + + run_osm2pgsql_updates(conn, options) + + # Write the current status to the file + endstate = repl.get_state_info(endseq) + status.set_status(conn, endstate.timestamp if endstate else None, + seq=endseq, indexed=False) + + return UpdateState.UP_TO_DATE - outhandler = WriteHandler(str(options['import_file'])) - endseq = repl.apply_diffs(outhandler, startseq + 1, - max_size=options['max_diff_size'] * 1024) - outhandler.close() - if endseq is None: - return UpdateState.NO_CHANGES +def run_osm2pgsql_updates(conn: Connection, options: MutableMapping[str, Any]) -> None: + """ Run osm2pgsql in append mode. + """ + # Remove any stale deletion marks. + with conn.cursor() as cur: + cur.execute('TRUNCATE place_to_be_deleted') + conn.commit() # Consume updates with osm2pgsql. options['append'] = True options['disable_jit'] = conn.server_version_tuple() >= (11, 0) run_osm2pgsql(options) - # Write the current status to the file - endstate = repl.get_state_info(endseq) - status.set_status(conn, endstate.timestamp if endstate else None, - seq=endseq, indexed=False) + # Handle deletions + with conn.cursor() as cur: + cur.execute('SELECT flush_deleted_places()') + conn.commit() - return UpdateState.UP_TO_DATE + +def _make_replication_server(url: str, timeout: int) -> ContextManager[ReplicationServer]: + """ Returns a ReplicationServer in form of a context manager. + + Creates a light wrapper around older versions of pyosmium that did + not support the context manager interface. + """ + if hasattr(ReplicationServer, '__enter__'): + # Patches the open_url function for pyosmium >= 3.2 + # where the socket timeout is no longer respected. + def patched_open_url(self: ReplicationServer, url: urlrequest.Request) -> Any: + """ Download a resource from the given URL and return a byte sequence + of the content. + """ + headers = {"User-Agent" : f"Nominatim (pyosmium/{pyo_version.pyosmium_release})"} + + if self.session is not None: + return self.session.get(url.get_full_url(), + headers=headers, timeout=timeout or None, + stream=True) + + @contextmanager + def _get_url_with_session() -> Iterator[requests.Response]: + with requests.Session() as session: + request = session.get(url.get_full_url(), + headers=headers, timeout=timeout or None, + stream=True) + yield request + + return _get_url_with_session() + + repl = ReplicationServer(url) + setattr(repl, 'open_url', types.MethodType(patched_open_url, repl)) + + return cast(ContextManager[ReplicationServer], repl) + + @contextmanager + def get_cm() -> Generator[ReplicationServer, None, None]: + yield ReplicationServer(url) + + return get_cm()