2015-10-09 10:48:31 -04:00
|
|
|
# -*- coding: utf-8 -*-
|
|
|
|
# Copyright 2015 OpenMarket Ltd
|
|
|
|
#
|
|
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
# you may not use this file except in compliance with the License.
|
|
|
|
# You may obtain a copy of the License at
|
|
|
|
#
|
|
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
#
|
|
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
# See the License for the specific language governing permissions and
|
|
|
|
# limitations under the License.
|
|
|
|
|
|
|
|
from twisted.internet import defer
|
|
|
|
|
2015-11-09 14:29:32 -05:00
|
|
|
from .background_updates import BackgroundUpdateStore
|
2015-11-05 11:10:54 -05:00
|
|
|
from synapse.api.errors import SynapseError
|
2015-10-16 09:37:14 -04:00
|
|
|
from synapse.storage.engines import PostgresEngine, Sqlite3Engine
|
2015-10-09 10:48:31 -04:00
|
|
|
|
2015-11-04 12:57:44 -05:00
|
|
|
import logging
|
|
|
|
|
|
|
|
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
|
|
|
2015-11-09 14:29:32 -05:00
|
|
|
class SearchStore(BackgroundUpdateStore):
|
|
|
|
|
|
|
|
EVENT_SEARCH_UPDATE_NAME = "event_search"
|
|
|
|
|
2015-11-10 10:50:58 -05:00
|
|
|
def __init__(self, hs):
|
|
|
|
super(SearchStore, self).__init__(hs)
|
|
|
|
self.register_background_update_handler(
|
|
|
|
self.EVENT_SEARCH_UPDATE_NAME, self._background_reindex_search
|
|
|
|
)
|
2015-11-09 14:29:32 -05:00
|
|
|
|
|
|
|
@defer.inlineCallbacks
|
|
|
|
def _background_reindex_search(self, progress, batch_size):
|
2015-11-10 11:20:13 -05:00
|
|
|
target_min_stream_id = progress["target_min_stream_id_inclusive"]
|
|
|
|
max_stream_id = progress["max_stream_id_exclusive"]
|
2015-11-09 14:29:32 -05:00
|
|
|
rows_inserted = progress.get("rows_inserted", 0)
|
|
|
|
|
|
|
|
INSERT_CLUMP_SIZE = 1000
|
|
|
|
TYPES = ["m.room.name", "m.room.message", "m.room.topic"]
|
|
|
|
|
|
|
|
def reindex_search_txn(txn):
|
|
|
|
sql = (
|
|
|
|
"SELECT stream_id, event_id FROM events"
|
|
|
|
" WHERE ? <= stream_ordering AND stream_ordering < ?"
|
|
|
|
" AND (%s)"
|
|
|
|
" ORDER BY stream_ordering DESC"
|
|
|
|
" LIMIT ?"
|
|
|
|
) % (" OR ".join("type = '%s'" % TYPES),)
|
|
|
|
|
|
|
|
txn.execute(sql, target_min_stream_id, max_stream_id, batch_size)
|
|
|
|
|
|
|
|
rows = txn.fetch_all()
|
|
|
|
if not rows:
|
|
|
|
return None
|
|
|
|
|
|
|
|
min_stream_id = rows[-1][0]
|
|
|
|
event_ids = [row[1] for row in rows]
|
|
|
|
|
|
|
|
events = self._get_events_txn(txn, event_ids)
|
|
|
|
|
|
|
|
event_search_rows = []
|
|
|
|
for event in events:
|
|
|
|
try:
|
|
|
|
event_id = event.event_id
|
|
|
|
room_id = event.room_id
|
|
|
|
content = event.content
|
|
|
|
if event.type == "m.room.message":
|
|
|
|
key = "content.body"
|
|
|
|
value = content["body"]
|
|
|
|
elif event.type == "m.room.topic":
|
|
|
|
key = "content.topic"
|
|
|
|
value = content["topic"]
|
|
|
|
elif event.type == "m.room.name":
|
|
|
|
key = "content.name"
|
|
|
|
value = content["name"]
|
2015-11-10 10:50:58 -05:00
|
|
|
except (KeyError, AttributeError):
|
2015-11-09 14:29:32 -05:00
|
|
|
# If the event is missing a necessary field then
|
|
|
|
# skip over it.
|
|
|
|
continue
|
|
|
|
|
|
|
|
event_search_rows.append((event_id, room_id, key, value))
|
|
|
|
|
|
|
|
if isinstance(self.database_engine, PostgresEngine):
|
|
|
|
sql = (
|
|
|
|
"INSERT INTO event_search (event_id, room_id, key, vector)"
|
|
|
|
" VALUES (?,?,?,to_tsvector('english', ?))"
|
|
|
|
)
|
|
|
|
elif isinstance(self.database_engine, Sqlite3Engine):
|
|
|
|
sql = (
|
|
|
|
"INSERT INTO event_search (event_id, room_id, key, value)"
|
|
|
|
" VALUES (?,?,?,?)"
|
|
|
|
)
|
|
|
|
else:
|
|
|
|
# This should be unreachable.
|
|
|
|
raise Exception("Unrecognized database engine")
|
|
|
|
|
|
|
|
for index in range(0, len(event_search_rows), INSERT_CLUMP_SIZE):
|
2015-11-10 10:50:58 -05:00
|
|
|
clump = event_search_rows[index:index + INSERT_CLUMP_SIZE]
|
2015-11-09 14:29:32 -05:00
|
|
|
txn.execute_many(sql, clump)
|
|
|
|
|
|
|
|
progress = {
|
2015-11-10 11:20:13 -05:00
|
|
|
"target_min_stream_id_inclusive": target_min_stream_id,
|
|
|
|
"max_stream_id_exclusive": min_stream_id,
|
2015-11-09 14:29:32 -05:00
|
|
|
"rows_inserted": rows_inserted + len(event_search_rows)
|
|
|
|
}
|
|
|
|
|
|
|
|
self._background_update_progress_txn(
|
|
|
|
txn, self.EVENT_SEARCH_UPDATE_NAME, progress
|
|
|
|
)
|
|
|
|
|
|
|
|
return len(event_search_rows)
|
|
|
|
|
|
|
|
result = yield self.runInteration(
|
|
|
|
self.EVENT_SEARCH_UPDATE_NAME, reindex_search_txn
|
|
|
|
)
|
|
|
|
|
|
|
|
if result is None:
|
2015-11-10 10:50:58 -05:00
|
|
|
yield self._end_background_update(self.EVENT_SEARCH_UPDATE_NAME)
|
2015-11-09 14:29:32 -05:00
|
|
|
|
|
|
|
defer.returnValue(result)
|
|
|
|
|
2015-10-09 10:48:31 -04:00
|
|
|
@defer.inlineCallbacks
|
2015-10-13 10:22:14 -04:00
|
|
|
def search_msgs(self, room_ids, search_term, keys):
|
2015-10-16 11:46:48 -04:00
|
|
|
"""Performs a full text search over events with given keys.
|
2015-10-16 06:28:12 -04:00
|
|
|
|
|
|
|
Args:
|
|
|
|
room_ids (list): List of room ids to search in
|
|
|
|
search_term (str): Search term to search for
|
|
|
|
keys (list): List of keys to search in, currently supports
|
2015-10-16 11:46:48 -04:00
|
|
|
"content.body", "content.name", "content.topic"
|
2015-10-16 06:28:12 -04:00
|
|
|
|
|
|
|
Returns:
|
2015-11-05 09:34:37 -05:00
|
|
|
list of dicts
|
2015-10-16 06:28:12 -04:00
|
|
|
"""
|
2015-10-09 10:48:31 -04:00
|
|
|
clauses = []
|
|
|
|
args = []
|
|
|
|
|
2015-10-22 08:25:22 -04:00
|
|
|
# Make sure we don't explode because the person is in too many rooms.
|
2015-10-22 11:54:56 -04:00
|
|
|
# We filter the results below regardless.
|
2015-10-22 11:18:35 -04:00
|
|
|
if len(room_ids) < 500:
|
2015-10-22 08:25:22 -04:00
|
|
|
clauses.append(
|
|
|
|
"room_id IN (%s)" % (",".join(["?"] * len(room_ids)),)
|
|
|
|
)
|
|
|
|
args.extend(room_ids)
|
2015-10-12 05:49:53 -04:00
|
|
|
|
2015-10-13 10:22:14 -04:00
|
|
|
local_clauses = []
|
|
|
|
for key in keys:
|
|
|
|
local_clauses.append("key = ?")
|
|
|
|
args.append(key)
|
|
|
|
|
|
|
|
clauses.append(
|
|
|
|
"(%s)" % (" OR ".join(local_clauses),)
|
|
|
|
)
|
2015-10-09 10:48:31 -04:00
|
|
|
|
2015-10-13 08:47:50 -04:00
|
|
|
if isinstance(self.database_engine, PostgresEngine):
|
|
|
|
sql = (
|
2015-10-22 11:54:56 -04:00
|
|
|
"SELECT ts_rank_cd(vector, query) AS rank, room_id, event_id"
|
2015-10-13 08:47:50 -04:00
|
|
|
" FROM plainto_tsquery('english', ?) as query, event_search"
|
|
|
|
" WHERE vector @@ query"
|
|
|
|
)
|
2015-10-16 09:37:14 -04:00
|
|
|
elif isinstance(self.database_engine, Sqlite3Engine):
|
2015-10-13 08:47:50 -04:00
|
|
|
sql = (
|
2015-10-23 08:23:48 -04:00
|
|
|
"SELECT rank(matchinfo(event_search)) as rank, room_id, event_id"
|
|
|
|
" FROM event_search"
|
2015-10-13 08:47:50 -04:00
|
|
|
" WHERE value MATCH ?"
|
|
|
|
)
|
2015-10-16 09:37:14 -04:00
|
|
|
else:
|
|
|
|
# This should be unreachable.
|
|
|
|
raise Exception("Unrecognized database engine")
|
2015-10-09 10:48:31 -04:00
|
|
|
|
|
|
|
for clause in clauses:
|
|
|
|
sql += " AND " + clause
|
|
|
|
|
2015-10-16 06:24:02 -04:00
|
|
|
# We add an arbitrary limit here to ensure we don't try to pull the
|
|
|
|
# entire table from the database.
|
2015-10-13 10:50:56 -04:00
|
|
|
sql += " ORDER BY rank DESC LIMIT 500"
|
2015-10-09 10:48:31 -04:00
|
|
|
|
|
|
|
results = yield self._execute(
|
2015-10-13 10:22:14 -04:00
|
|
|
"search_msgs", self.cursor_to_dict, sql, *([search_term] + args)
|
2015-10-09 10:48:31 -04:00
|
|
|
)
|
|
|
|
|
2015-10-22 11:54:56 -04:00
|
|
|
results = filter(lambda row: row["room_id"] in room_ids, results)
|
|
|
|
|
2015-10-09 10:48:31 -04:00
|
|
|
events = yield self._get_events([r["event_id"] for r in results])
|
|
|
|
|
|
|
|
event_map = {
|
|
|
|
ev.event_id: ev
|
|
|
|
for ev in events
|
|
|
|
}
|
|
|
|
|
2015-11-05 09:34:37 -05:00
|
|
|
defer.returnValue([
|
2015-10-09 10:48:31 -04:00
|
|
|
{
|
2015-11-05 09:34:37 -05:00
|
|
|
"event": event_map[r["event_id"]],
|
|
|
|
"rank": r["rank"],
|
|
|
|
}
|
|
|
|
for r in results
|
|
|
|
if r["event_id"] in event_map
|
|
|
|
])
|
2015-11-04 12:57:44 -05:00
|
|
|
|
|
|
|
@defer.inlineCallbacks
|
|
|
|
def search_room(self, room_id, search_term, keys, limit, pagination_token=None):
|
|
|
|
"""Performs a full text search over events with given keys.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
room_id (str): The room_id to search in
|
|
|
|
search_term (str): Search term to search for
|
|
|
|
keys (list): List of keys to search in, currently supports
|
|
|
|
"content.body", "content.name", "content.topic"
|
|
|
|
pagination_token (str): A pagination token previously returned
|
|
|
|
|
|
|
|
Returns:
|
2015-11-05 09:34:37 -05:00
|
|
|
list of dicts
|
2015-11-04 12:57:44 -05:00
|
|
|
"""
|
|
|
|
clauses = []
|
|
|
|
args = [search_term, room_id]
|
|
|
|
|
|
|
|
local_clauses = []
|
|
|
|
for key in keys:
|
|
|
|
local_clauses.append("key = ?")
|
|
|
|
args.append(key)
|
|
|
|
|
|
|
|
clauses.append(
|
|
|
|
"(%s)" % (" OR ".join(local_clauses),)
|
|
|
|
)
|
|
|
|
|
|
|
|
if pagination_token:
|
2015-11-05 11:10:54 -05:00
|
|
|
try:
|
|
|
|
topo, stream = pagination_token.split(",")
|
|
|
|
topo = int(topo)
|
|
|
|
stream = int(stream)
|
|
|
|
except:
|
|
|
|
raise SynapseError(400, "Invalid pagination token")
|
|
|
|
|
2015-11-04 12:57:44 -05:00
|
|
|
clauses.append(
|
|
|
|
"(topological_ordering < ?"
|
|
|
|
" OR (topological_ordering = ? AND stream_ordering < ?))"
|
|
|
|
)
|
|
|
|
args.extend([topo, topo, stream])
|
|
|
|
|
|
|
|
if isinstance(self.database_engine, PostgresEngine):
|
|
|
|
sql = (
|
|
|
|
"SELECT ts_rank_cd(vector, query) as rank,"
|
|
|
|
" topological_ordering, stream_ordering, room_id, event_id"
|
|
|
|
" FROM plainto_tsquery('english', ?) as query, event_search"
|
|
|
|
" NATURAL JOIN events"
|
|
|
|
" WHERE vector @@ query AND room_id = ?"
|
|
|
|
)
|
|
|
|
elif isinstance(self.database_engine, Sqlite3Engine):
|
|
|
|
sql = (
|
|
|
|
"SELECT rank(matchinfo(event_search)) as rank, room_id, event_id"
|
|
|
|
" topological_ordering, stream_ordering"
|
|
|
|
" FROM event_search"
|
|
|
|
" NATURAL JOIN events"
|
|
|
|
" WHERE value MATCH ? AND room_id = ?"
|
|
|
|
)
|
|
|
|
else:
|
|
|
|
# This should be unreachable.
|
|
|
|
raise Exception("Unrecognized database engine")
|
|
|
|
|
|
|
|
for clause in clauses:
|
|
|
|
sql += " AND " + clause
|
|
|
|
|
|
|
|
# We add an arbitrary limit here to ensure we don't try to pull the
|
|
|
|
# entire table from the database.
|
|
|
|
sql += " ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ?"
|
|
|
|
|
|
|
|
args.append(limit)
|
|
|
|
|
|
|
|
results = yield self._execute(
|
|
|
|
"search_rooms", self.cursor_to_dict, sql, *args
|
|
|
|
)
|
|
|
|
|
|
|
|
events = yield self._get_events([r["event_id"] for r in results])
|
|
|
|
|
|
|
|
event_map = {
|
|
|
|
ev.event_id: ev
|
|
|
|
for ev in events
|
|
|
|
}
|
|
|
|
|
2015-11-05 09:34:37 -05:00
|
|
|
defer.returnValue([
|
2015-11-04 12:57:44 -05:00
|
|
|
{
|
2015-11-05 09:34:37 -05:00
|
|
|
"event": event_map[r["event_id"]],
|
|
|
|
"rank": r["rank"],
|
|
|
|
"pagination_token": "%s,%s" % (
|
|
|
|
r["topological_ordering"], r["stream_ordering"]
|
|
|
|
),
|
|
|
|
}
|
|
|
|
for r in results
|
|
|
|
if r["event_id"] in event_map
|
|
|
|
])
|