Merge remote-tracking branch 'upstream/master'

[nominatim.git] / nominatim / tools / replication.py

diff --git a/nominatim/tools/replication.py b/nominatim/tools/replication.py
index 53571706472036f80b546846f23b43fc905ec6c6..d93335b8b7169bcf9746970f4919c0db76066917 100644 (file)
--- a/nominatim/tools/replication.py
+++ b/nominatim/tools/replication.py
@@ -7,27 +7,34 @@
 """
 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)
@@ -36,9 +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"
@@ -51,7 +57,8 @@ def init_replication(conn, base_url):
     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.
     """
@@ -62,7 +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)
@@ -84,7 +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`.
     """
@@ -95,6 +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
@@ -110,7 +121,7 @@ def update(conn, options):
         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)
@@ -132,17 +143,43 @@ def update(conn, options):
     return UpdateState.UP_TO_DATE
 
 
-def _make_replication_server(url):
+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 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.
+            """
+            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():
+    def get_cm() -> Generator[ReplicationServer, None, None]:
         yield ReplicationServer(url)
 
     return get_cm()