X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/17bbe2637a32e93b19ff887a896bf1b444cd9792..78c19bc006d5957f183968d086226be4b2b65fe3:/nominatim/tools/replication.py?ds=sidebyside diff --git a/nominatim/tools/replication.py b/nominatim/tools/replication.py index db706bf6..edd63e49 100644 --- a/nominatim/tools/replication.py +++ b/nominatim/tools/replication.py @@ -7,21 +7,25 @@ """ Functions for updating a database from a replication source. """ -from typing import ContextManager, MutableMapping, Any, Generator, cast +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.db.connection import Connection, connect 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().critical("pyosmium not installed. Replication functions not available.\n" "To install pyosmium via pip: pip3 install osmium") @@ -29,7 +33,8 @@ except ImportError as exc: LOG = logging.getLogger() -def init_replication(conn: Connection, base_url: str) -> None: +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) @@ -38,9 +43,8 @@ def init_replication(conn: Connection, base_url: str) -> None: # 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" @@ -53,7 +57,8 @@ def init_replication(conn: Connection, base_url: str) -> None: LOG.warning("Updates initialised at sequence %s (%s)", seq, date) -def check_for_updates(conn: Connection, base_url: str) -> int: +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. """ @@ -64,7 +69,8 @@ def check_for_updates(conn: Connection, base_url: str) -> int: "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) @@ -86,11 +92,14 @@ class UpdateState(Enum): NO_CHANGES = 3 -def update(conn: Connection, options: MutableMapping[str, Any]) -> UpdateState: +def update(dsn: str, 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`. """ - startdate, startseq, indexed = status.get_status(conn) + with connect(dsn) as conn: + startdate, startseq, indexed = status.get_status(conn) + conn.commit() if startseq is None: LOG.error("Replication not set up. " @@ -114,7 +123,7 @@ def update(conn: Connection, options: MutableMapping[str, Any]) -> UpdateState: options['import_file'].unlink() # Read updates into file. - with _make_replication_server(options['base_url']) as repl: + 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) @@ -123,27 +132,71 @@ def update(conn: Connection, options: MutableMapping[str, Any]) -> UpdateState: 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) + with connect(dsn) as conn: + 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) + # 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) + conn.commit() return UpdateState.UP_TO_DATE -def _make_replication_server(url: str) -> ContextManager[ReplicationServer]: +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) + + # Handle deletions + with conn.cursor() as cur: + cur.execute('SELECT flush_deleted_places()') + conn.commit() + + +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__'): - return cast(ContextManager[ReplicationServer], ReplicationServer(url)) + # 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]: