X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/8f0885f6cb24f545a2f5021d53d8aec64a72bf9b..e0a7b94e49cb4c47d0a42f0d8108f826e9de59f3:/nominatim/tools/replication.py

diff --git a/nominatim/tools/replication.py b/nominatim/tools/replication.py
index f278556a..cbbf0585 100644
--- a/nominatim/tools/replication.py
+++ b/nominatim/tools/replication.py
@@ -1,12 +1,28 @@
+# 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.
 """
-import datetime
+import datetime as dt
+from enum import Enum
 import logging
+import time
 
-from osmium.replication.server import ReplicationServer
+from nominatim.db import status
+from nominatim.tools.exec_utils import run_osm2pgsql
+from nominatim.errors import UsageError
 
-from ..db import status
+try:
+    from osmium.replication.server import ReplicationServer
+    from osmium import WriteHandler
+except ImportError as exc:
+    logging.getLogger().fatal("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()
 
@@ -17,7 +33,7 @@ def init_replication(conn, base_url):
     date = status.compute_database_date(conn)
 
     # margin of error to make sure we get all data
-    date -= datetime.timedelta(hours=3)
+    date -= dt.timedelta(hours=3)
 
     repl = ReplicationServer(base_url)
 
@@ -27,7 +43,7 @@ def init_replication(conn, base_url):
         LOG.fatal("Cannot reach the configured replication service '%s'.\n"
                   "Does the URL point to a directory containing OSM update data?",
                   base_url)
-        raise RuntimeError("Failed to reach replication service")
+        raise UsageError("Failed to reach replication service")
 
     status.set_status(conn, date=date, seq=seq)
 
@@ -53,7 +69,64 @@ def check_for_updates(conn, base_url):
 
     if state.sequence <= seq:
         LOG.warning("Database is up to date.")
-        return 1
+        return 2
 
     LOG.warning("New data available (%i => %i).", seq, state.sequence)
     return 0
+
+class UpdateState(Enum):
+    """ Possible states after an update has run.
+    """
+
+    UP_TO_DATE = 0
+    MORE_PENDING = 2
+    NO_CHANGES = 3
+
+
+def update(conn, options):
+    """ Update database from the next batch of data. Returns the state of
+        updates according to `UpdateState`.
+    """
+    startdate, startseq, indexed = status.get_status(conn)
+
+    if startseq is None:
+        LOG.error("Replication not set up. "
+                  "Please run 'nominatim replication --init' first.")
+        raise UsageError("Replication not set up.")
+
+    if not indexed and options['indexed_only']:
+        LOG.info("Skipping update. There is data that needs indexing.")
+        return UpdateState.MORE_PENDING
+
+    last_since_update = dt.datetime.now(dt.timezone.utc) - startdate
+    update_interval = dt.timedelta(seconds=options['update_interval'])
+    if last_since_update < update_interval:
+        duration = (update_interval - last_since_update).seconds
+        LOG.warning("Sleeping for %s sec before next update.", duration)
+        time.sleep(duration)
+
+    if options['import_file'].exists():
+        options['import_file'].unlink()
+
+    # Read updates into file.
+    repl = ReplicationServer(options['base_url'])
+
+    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
+
+    # 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)
+
+    return UpdateState.UP_TO_DATE