+# 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
-from ..db import status
-from .exec_utils import run_osm2pgsql
-from ..errors import UsageError
+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
-except ModuleNotFoundError as exc:
- logging.getLogger().fatal("pyosmium not installed. Replication functions not available.\n"
- "To install pyosmium via pip: pip3 install osmium")
+ from osmium import version as pyo_version
+except ImportError as exc:
+ 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)
# 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"
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.
"""
"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)
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`.
"""
"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
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()
- 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
- if endseq is None:
- return UpdateState.NO_CHANGES
+ # Consume updates with osm2pgsql.
+ options['append'] = True
+ options['disable_jit'] = conn.server_version_tuple() >= (11, 0)
+ run_osm2pgsql(options)
- # Consume updates with osm2pgsql.
- options['append'] = True
- run_osm2pgsql(options)
-
- # Write the current status to the file
- endstate = repl.get_state_info(endseq)
- status.set_status(conn, endstate.timestamp, seq=endseq, indexed=False)
+ # 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
+
+
+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.
+ """
+ get_params = {
+ 'headers': {"User-Agent" : f"Nominatim (pyosmium/{pyo_version.pyosmium_release})"},
+ 'timeout': timeout or None,
+ 'stream': True
+ }
+
+ if self.session is not None:
+ return self.session.get(url.get_full_url(), **get_params)
+
+ @contextmanager
+ def _get_url_with_session() -> Iterator[requests.Response]:
+ with requests.Session() as session:
+ request = session.get(url.get_full_url(), **get_params) # type: ignore
+ yield request
+
+ return _get_url_with_session()
+
+ repl = ReplicationServer(url)
+ 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()