2016-01-06 21:26:29 -07:00
|
|
|
# Copyright 2015, 2016 OpenMarket Ltd
|
2015-10-09 08:48:31 -06:00
|
|
|
#
|
|
|
|
# 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.
|
|
|
|
|
2018-07-09 00:09:20 -06:00
|
|
|
import itertools
|
|
|
|
import logging
|
2022-02-15 06:47:05 -07:00
|
|
|
from typing import TYPE_CHECKING, Collection, Dict, Iterable, List, Optional, Set, Tuple
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
import attr
|
2018-07-09 00:09:20 -06:00
|
|
|
from unpaddedbase64 import decode_base64, encode_base64
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2018-07-09 00:09:20 -06:00
|
|
|
from synapse.api.constants import EventTypes, Membership
|
2019-12-11 06:07:25 -07:00
|
|
|
from synapse.api.errors import NotFoundError, SynapseError
|
2018-07-09 00:09:20 -06:00
|
|
|
from synapse.api.filtering import Filter
|
2021-01-26 08:50:21 -07:00
|
|
|
from synapse.events import EventBase
|
2018-10-30 11:33:41 -06:00
|
|
|
from synapse.storage.state import StateFilter
|
2022-05-16 09:35:31 -06:00
|
|
|
from synapse.types import JsonDict, StreamKeyType, UserID
|
2016-05-11 06:42:37 -06:00
|
|
|
from synapse.visibility import filter_events_for_client
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2021-01-26 08:50:21 -07:00
|
|
|
if TYPE_CHECKING:
|
2021-03-23 05:12:48 -06:00
|
|
|
from synapse.server import HomeServer
|
2021-01-26 08:50:21 -07:00
|
|
|
|
2015-10-09 08:48:31 -06:00
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
@attr.s(slots=True, frozen=True, auto_attribs=True)
|
|
|
|
class _SearchResult:
|
|
|
|
# The count of results.
|
|
|
|
count: int
|
|
|
|
# A mapping of event ID to the rank of that event.
|
|
|
|
rank_map: Dict[str, int]
|
|
|
|
# A list of the resulting events.
|
|
|
|
allowed_events: List[EventBase]
|
|
|
|
# A map of room ID to results.
|
|
|
|
room_groups: Dict[str, JsonDict]
|
|
|
|
# A set of event IDs to highlight.
|
|
|
|
highlights: Set[str]
|
|
|
|
|
|
|
|
|
2021-10-08 05:44:43 -06:00
|
|
|
class SearchHandler:
|
2021-01-26 08:50:21 -07:00
|
|
|
def __init__(self, hs: "HomeServer"):
|
2022-02-23 04:04:02 -07:00
|
|
|
self.store = hs.get_datastores().main
|
2021-10-08 05:44:43 -06:00
|
|
|
self.state_handler = hs.get_state_handler()
|
|
|
|
self.clock = hs.get_clock()
|
|
|
|
self.hs = hs
|
2019-05-09 06:21:57 -06:00
|
|
|
self._event_serializer = hs.get_event_client_serializer()
|
2022-03-18 11:49:32 -06:00
|
|
|
self._relations_handler = hs.get_relations_handler()
|
2019-10-23 10:25:54 -06:00
|
|
|
self.storage = hs.get_storage()
|
2022-05-25 05:59:04 -06:00
|
|
|
self.state_storage = self.storage.state
|
2019-12-11 06:07:25 -07:00
|
|
|
self.auth = hs.get_auth()
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2020-07-24 08:53:25 -06:00
|
|
|
async def get_old_rooms_from_upgraded_room(self, room_id: str) -> Iterable[str]:
|
2019-01-18 03:04:47 -07:00
|
|
|
"""Retrieves room IDs of old rooms in the history of an upgraded room.
|
|
|
|
|
|
|
|
We do so by checking the m.room.create event of the room for a
|
|
|
|
`predecessor` key. If it exists, we add the room ID to our return
|
|
|
|
list and then check that room for a m.room.create event and so on
|
|
|
|
until we can no longer find any more previous rooms.
|
|
|
|
|
|
|
|
The full list of all found rooms in then returned.
|
|
|
|
|
|
|
|
Args:
|
2020-07-24 08:53:25 -06:00
|
|
|
room_id: id of the room to search through.
|
2019-01-18 03:04:47 -07:00
|
|
|
|
|
|
|
Returns:
|
2020-07-24 08:53:25 -06:00
|
|
|
Predecessor room ids
|
2019-01-18 03:04:47 -07:00
|
|
|
"""
|
|
|
|
|
|
|
|
historical_room_ids = []
|
|
|
|
|
2019-12-11 06:07:25 -07:00
|
|
|
# The initial room must have been known for us to get this far
|
2020-05-11 13:12:39 -06:00
|
|
|
predecessor = await self.store.get_room_predecessor(room_id)
|
2019-01-18 03:04:47 -07:00
|
|
|
|
2019-12-11 06:07:25 -07:00
|
|
|
while True:
|
2019-01-18 03:04:47 -07:00
|
|
|
if not predecessor:
|
2019-12-11 06:07:25 -07:00
|
|
|
# We have reached the end of the chain of predecessors
|
|
|
|
break
|
|
|
|
|
|
|
|
if not isinstance(predecessor.get("room_id"), str):
|
|
|
|
# This predecessor object is malformed. Exit here
|
|
|
|
break
|
|
|
|
|
|
|
|
predecessor_room_id = predecessor["room_id"]
|
|
|
|
|
|
|
|
# Don't add it to the list until we have checked that we are in the room
|
|
|
|
try:
|
2020-05-11 13:12:39 -06:00
|
|
|
next_predecessor_room = await self.store.get_room_predecessor(
|
2019-12-11 06:07:25 -07:00
|
|
|
predecessor_room_id
|
|
|
|
)
|
|
|
|
except NotFoundError:
|
|
|
|
# The predecessor is not a known room, so we are done here
|
2019-01-18 03:04:47 -07:00
|
|
|
break
|
|
|
|
|
2019-12-11 06:07:25 -07:00
|
|
|
historical_room_ids.append(predecessor_room_id)
|
2019-01-18 03:04:47 -07:00
|
|
|
|
2019-12-11 06:07:25 -07:00
|
|
|
# And repeat
|
|
|
|
predecessor = next_predecessor_room
|
2019-01-18 03:04:47 -07:00
|
|
|
|
2019-07-23 07:00:55 -06:00
|
|
|
return historical_room_ids
|
2019-01-18 03:04:47 -07:00
|
|
|
|
2021-01-26 08:50:21 -07:00
|
|
|
async def search(
|
|
|
|
self, user: UserID, content: JsonDict, batch: Optional[str] = None
|
|
|
|
) -> JsonDict:
|
2015-10-16 04:52:16 -06:00
|
|
|
"""Performs a full text search for a user.
|
|
|
|
|
|
|
|
Args:
|
2022-02-15 06:47:05 -07:00
|
|
|
user: The user performing the search.
|
2021-01-26 08:50:21 -07:00
|
|
|
content: Search parameters
|
|
|
|
batch: The next_batch parameter. Used for pagination.
|
2015-10-16 04:52:16 -06:00
|
|
|
|
|
|
|
Returns:
|
|
|
|
dict to be returned to the client with results of search
|
|
|
|
"""
|
|
|
|
|
2021-09-29 04:44:15 -06:00
|
|
|
if not self.hs.config.server.enable_search:
|
2018-12-04 04:01:02 -07:00
|
|
|
raise SynapseError(400, "Search is disabled on this homeserver")
|
|
|
|
|
2015-11-05 07:34:37 -07:00
|
|
|
batch_group = None
|
|
|
|
batch_group_key = None
|
|
|
|
batch_token = None
|
|
|
|
if batch:
|
|
|
|
try:
|
2018-09-06 08:22:23 -06:00
|
|
|
b = decode_base64(batch).decode("ascii")
|
2015-11-05 07:34:37 -07:00
|
|
|
batch_group, batch_group_key, batch_token = b.split("\n")
|
|
|
|
|
|
|
|
assert batch_group is not None
|
|
|
|
assert batch_group_key is not None
|
|
|
|
assert batch_token is not None
|
2017-10-23 08:52:32 -06:00
|
|
|
except Exception:
|
2015-11-05 07:34:37 -07:00
|
|
|
raise SynapseError(400, "Invalid batch")
|
|
|
|
|
2018-05-22 10:21:18 -06:00
|
|
|
logger.info(
|
|
|
|
"Search batch properties: %r, %r, %r",
|
|
|
|
batch_group,
|
|
|
|
batch_group_key,
|
|
|
|
batch_token,
|
|
|
|
)
|
|
|
|
|
|
|
|
logger.info("Search content: %s", content)
|
|
|
|
|
2015-10-13 08:22:14 -06:00
|
|
|
try:
|
2015-11-04 10:57:44 -07:00
|
|
|
room_cat = content["search_categories"]["room_events"]
|
2015-11-05 09:29:16 -07:00
|
|
|
|
|
|
|
# The actual thing to query in FTS
|
2015-11-04 10:57:44 -07:00
|
|
|
search_term = room_cat["search_term"]
|
2015-11-05 09:29:16 -07:00
|
|
|
|
|
|
|
# Which "keys" to search over in FTS query
|
2015-11-04 10:57:44 -07:00
|
|
|
keys = room_cat.get(
|
2015-10-14 03:35:50 -06:00
|
|
|
"keys", ["content.body", "content.name", "content.topic"]
|
|
|
|
)
|
2015-11-05 09:29:16 -07:00
|
|
|
|
|
|
|
# Filter to apply to results
|
2015-11-04 10:57:44 -07:00
|
|
|
filter_dict = room_cat.get("filter", {})
|
2015-11-05 09:29:16 -07:00
|
|
|
|
2020-10-23 10:38:40 -06:00
|
|
|
# What to order results by (impacts whether pagination can be done)
|
2015-11-04 10:57:44 -07:00
|
|
|
order_by = room_cat.get("order_by", "rank")
|
2015-11-05 09:29:16 -07:00
|
|
|
|
2015-11-20 07:16:42 -07:00
|
|
|
# Return the current state of the rooms?
|
|
|
|
include_state = room_cat.get("include_state", False)
|
|
|
|
|
2015-11-05 09:29:16 -07:00
|
|
|
# Include context around each event?
|
2015-11-04 10:57:44 -07:00
|
|
|
event_context = room_cat.get("event_context", None)
|
2022-02-15 06:47:05 -07:00
|
|
|
before_limit = after_limit = None
|
|
|
|
include_profile = False
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2015-11-05 09:29:16 -07:00
|
|
|
# Group results together? May allow clients to paginate within a
|
|
|
|
# group
|
2015-11-04 10:57:44 -07:00
|
|
|
group_by = room_cat.get("groupings", {}).get("group_by", {})
|
|
|
|
group_keys = [g["key"] for g in group_by]
|
|
|
|
|
2015-10-28 12:25:11 -06:00
|
|
|
if event_context is not None:
|
|
|
|
before_limit = int(event_context.get("before_limit", 5))
|
|
|
|
after_limit = int(event_context.get("after_limit", 5))
|
2015-11-20 07:16:42 -07:00
|
|
|
|
|
|
|
# Return the historic display name and avatar for the senders
|
|
|
|
# of the events?
|
2015-11-20 04:39:44 -07:00
|
|
|
include_profile = bool(event_context.get("include_profile", False))
|
2015-10-13 08:22:14 -06:00
|
|
|
except KeyError:
|
|
|
|
raise SynapseError(400, "Invalid search query")
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2015-11-04 10:57:44 -07:00
|
|
|
if order_by not in ("rank", "recent"):
|
|
|
|
raise SynapseError(400, "Invalid order by: %r" % (order_by,))
|
|
|
|
|
|
|
|
if set(group_keys) - {"room_id", "sender"}:
|
|
|
|
raise SynapseError(
|
|
|
|
400,
|
|
|
|
"Invalid group by keys: %r"
|
|
|
|
% (set(group_keys) - {"room_id", "sender"},),
|
|
|
|
)
|
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
return await self._search(
|
|
|
|
user,
|
|
|
|
batch_group,
|
|
|
|
batch_group_key,
|
|
|
|
batch_token,
|
|
|
|
search_term,
|
|
|
|
keys,
|
|
|
|
filter_dict,
|
|
|
|
order_by,
|
|
|
|
include_state,
|
|
|
|
group_keys,
|
|
|
|
event_context,
|
|
|
|
before_limit,
|
|
|
|
after_limit,
|
|
|
|
include_profile,
|
|
|
|
)
|
|
|
|
|
|
|
|
async def _search(
|
|
|
|
self,
|
|
|
|
user: UserID,
|
|
|
|
batch_group: Optional[str],
|
|
|
|
batch_group_key: Optional[str],
|
|
|
|
batch_token: Optional[str],
|
|
|
|
search_term: str,
|
|
|
|
keys: List[str],
|
|
|
|
filter_dict: JsonDict,
|
|
|
|
order_by: str,
|
|
|
|
include_state: bool,
|
|
|
|
group_keys: List[str],
|
|
|
|
event_context: Optional[bool],
|
|
|
|
before_limit: Optional[int],
|
|
|
|
after_limit: Optional[int],
|
|
|
|
include_profile: bool,
|
|
|
|
) -> JsonDict:
|
|
|
|
"""Performs a full text search for a user.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
user: The user performing the search.
|
|
|
|
batch_group: Pagination information.
|
|
|
|
batch_group_key: Pagination information.
|
|
|
|
batch_token: Pagination information.
|
|
|
|
search_term: Search term to search for
|
|
|
|
keys: List of keys to search in, currently supports
|
|
|
|
"content.body", "content.name", "content.topic"
|
|
|
|
filter_dict: The JSON to build a filter out of.
|
|
|
|
order_by: How to order the results. Valid values ore "rank" and "recent".
|
|
|
|
include_state: True if the state of the room at each result should
|
|
|
|
be included.
|
|
|
|
group_keys: A list of ways to group the results. Valid values are
|
|
|
|
"room_id" and "sender".
|
|
|
|
event_context: True to include contextual events around results.
|
|
|
|
before_limit:
|
|
|
|
The number of events before a result to include as context.
|
|
|
|
|
|
|
|
Only used if event_context is True.
|
|
|
|
after_limit:
|
|
|
|
The number of events after a result to include as context.
|
|
|
|
|
|
|
|
Only used if event_context is True.
|
|
|
|
include_profile: True if historical profile information should be
|
|
|
|
included in the event context.
|
|
|
|
|
|
|
|
Only used if event_context is True.
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
dict to be returned to the client with results of search
|
|
|
|
"""
|
2021-11-09 06:10:58 -07:00
|
|
|
search_filter = Filter(self.hs, filter_dict)
|
2015-10-20 10:09:53 -06:00
|
|
|
|
2015-10-14 02:46:31 -06:00
|
|
|
# TODO: Search through left rooms too
|
2020-05-11 13:12:39 -06:00
|
|
|
rooms = await self.store.get_rooms_for_local_user_where_membership_is(
|
2015-10-13 08:22:14 -06:00
|
|
|
user.to_string(),
|
|
|
|
membership_list=[Membership.JOIN],
|
|
|
|
# membership_list=[Membership.JOIN, Membership.LEAVE, Membership.Ban],
|
2015-10-12 03:49:53 -06:00
|
|
|
)
|
2020-02-21 05:15:07 -07:00
|
|
|
room_ids = {r.room_id for r in rooms}
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2019-01-18 03:04:47 -07:00
|
|
|
# If doing a subset of all rooms seearch, check if any of the rooms
|
|
|
|
# are from an upgraded room, and search their contents as well
|
|
|
|
if search_filter.rooms:
|
2021-07-16 11:22:36 -06:00
|
|
|
historical_room_ids: List[str] = []
|
2019-01-22 04:12:48 -07:00
|
|
|
for room_id in search_filter.rooms:
|
2019-01-18 03:04:47 -07:00
|
|
|
# Add any previous rooms to the search if they exist
|
2020-05-11 13:12:39 -06:00
|
|
|
ids = await self.get_old_rooms_from_upgraded_room(room_id)
|
2019-01-18 03:04:47 -07:00
|
|
|
historical_room_ids += ids
|
|
|
|
|
|
|
|
# Prevent any historical events from being filtered
|
2019-01-22 04:12:48 -07:00
|
|
|
search_filter = search_filter.with_room_ids(historical_room_ids)
|
|
|
|
|
|
|
|
room_ids = search_filter.filter_rooms(room_ids)
|
2019-01-18 03:04:47 -07:00
|
|
|
|
2015-11-05 07:34:37 -07:00
|
|
|
if batch_group == "room_id":
|
2015-11-05 10:26:19 -07:00
|
|
|
room_ids.intersection_update({batch_group_key})
|
2015-11-05 07:34:37 -07:00
|
|
|
|
2015-11-30 10:45:31 -07:00
|
|
|
if not room_ids:
|
2019-07-23 07:00:55 -06:00
|
|
|
return {
|
|
|
|
"search_categories": {
|
|
|
|
"room_events": {"results": [], "count": 0, "highlights": []}
|
2015-11-30 10:45:31 -07:00
|
|
|
}
|
2019-07-23 07:00:55 -06:00
|
|
|
}
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
sender_group: Optional[Dict[str, JsonDict]]
|
2015-11-05 09:29:16 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if order_by == "rank":
|
|
|
|
search_result, sender_group = await self._search_by_rank(
|
|
|
|
user, room_ids, search_term, keys, search_filter
|
|
|
|
)
|
|
|
|
# Unused return values for rank search.
|
|
|
|
global_next_batch = None
|
|
|
|
elif order_by == "recent":
|
|
|
|
search_result, global_next_batch = await self._search_by_recent(
|
|
|
|
user,
|
|
|
|
room_ids,
|
|
|
|
search_term,
|
|
|
|
keys,
|
|
|
|
search_filter,
|
|
|
|
batch_group,
|
|
|
|
batch_group_key,
|
|
|
|
batch_token,
|
|
|
|
)
|
|
|
|
# Unused return values for recent search.
|
|
|
|
sender_group = None
|
|
|
|
else:
|
|
|
|
# We should never get here due to the guard earlier.
|
|
|
|
raise NotImplementedError()
|
2015-11-27 09:40:42 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
logger.info("Found %d events to return", len(search_result.allowed_events))
|
2015-12-11 04:40:23 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
# If client has asked for "context" for each event (i.e. some surrounding
|
|
|
|
# events and state), fetch that
|
|
|
|
if event_context is not None:
|
|
|
|
# Note that before and after limit must be set in this case.
|
|
|
|
assert before_limit is not None
|
|
|
|
assert after_limit is not None
|
|
|
|
|
|
|
|
contexts = await self._calculate_event_contexts(
|
|
|
|
user,
|
|
|
|
search_result.allowed_events,
|
|
|
|
before_limit,
|
|
|
|
after_limit,
|
|
|
|
include_profile,
|
|
|
|
)
|
|
|
|
else:
|
|
|
|
contexts = {}
|
2015-10-20 10:09:53 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
# TODO: Add a limit
|
2015-12-11 04:40:23 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
state_results = {}
|
|
|
|
if include_state:
|
|
|
|
for room_id in {e.room_id for e in search_result.allowed_events}:
|
|
|
|
state = await self.state_handler.get_current_state(room_id)
|
|
|
|
state_results[room_id] = list(state.values())
|
2015-11-27 09:40:42 -07:00
|
|
|
|
2022-04-11 13:37:46 -06:00
|
|
|
aggregations = await self._relations_handler.get_bundled_aggregations(
|
|
|
|
# Generate an iterable of EventBase for all the events that will be
|
|
|
|
# returned, including contextual events.
|
|
|
|
itertools.chain(
|
|
|
|
# The events_before and events_after for each context.
|
|
|
|
itertools.chain.from_iterable(
|
2022-04-27 07:03:44 -06:00
|
|
|
itertools.chain(context["events_before"], context["events_after"])
|
2022-04-11 13:37:46 -06:00
|
|
|
for context in contexts.values()
|
2022-02-15 06:47:05 -07:00
|
|
|
),
|
2022-04-11 13:37:46 -06:00
|
|
|
# The returned events.
|
|
|
|
search_result.allowed_events,
|
|
|
|
),
|
|
|
|
user.to_string(),
|
|
|
|
)
|
2015-11-27 09:40:42 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
# We're now about to serialize the events. We should not make any
|
|
|
|
# blocking calls after this. Otherwise, the 'age' will be wrong.
|
2015-11-05 07:34:37 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
time_now = self.clock.time_msec()
|
2015-10-12 03:49:53 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
for context in contexts.values():
|
|
|
|
context["events_before"] = self._event_serializer.serialize_events(
|
2022-04-27 07:03:44 -06:00
|
|
|
context["events_before"], time_now, bundle_aggregations=aggregations
|
2022-02-15 06:47:05 -07:00
|
|
|
)
|
|
|
|
context["events_after"] = self._event_serializer.serialize_events(
|
2022-04-27 07:03:44 -06:00
|
|
|
context["events_after"], time_now, bundle_aggregations=aggregations
|
2015-11-04 10:57:44 -07:00
|
|
|
)
|
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
results = [
|
|
|
|
{
|
|
|
|
"rank": search_result.rank_map[e.event_id],
|
|
|
|
"result": self._event_serializer.serialize_event(
|
|
|
|
e, time_now, bundle_aggregations=aggregations
|
|
|
|
),
|
|
|
|
"context": contexts.get(e.event_id, {}),
|
|
|
|
}
|
|
|
|
for e in search_result.allowed_events
|
|
|
|
]
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
rooms_cat_res: JsonDict = {
|
|
|
|
"results": results,
|
|
|
|
"count": search_result.count,
|
|
|
|
"highlights": list(search_result.highlights),
|
|
|
|
}
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if state_results:
|
|
|
|
rooms_cat_res["state"] = {
|
|
|
|
room_id: self._event_serializer.serialize_events(state_events, time_now)
|
|
|
|
for room_id, state_events in state_results.items()
|
|
|
|
}
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if search_result.room_groups and "room_id" in group_keys:
|
|
|
|
rooms_cat_res.setdefault("groups", {})[
|
|
|
|
"room_id"
|
|
|
|
] = search_result.room_groups
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if sender_group and "sender" in group_keys:
|
|
|
|
rooms_cat_res.setdefault("groups", {})["sender"] = sender_group
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if global_next_batch:
|
|
|
|
rooms_cat_res["next_batch"] = global_next_batch
|
2015-12-11 04:40:23 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
return {"search_categories": {"room_events": rooms_cat_res}}
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
async def _search_by_rank(
|
|
|
|
self,
|
|
|
|
user: UserID,
|
|
|
|
room_ids: Collection[str],
|
|
|
|
search_term: str,
|
|
|
|
keys: Iterable[str],
|
|
|
|
search_filter: Filter,
|
|
|
|
) -> Tuple[_SearchResult, Dict[str, JsonDict]]:
|
|
|
|
"""
|
|
|
|
Performs a full text search for a user ordering by rank.
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Args:
|
|
|
|
user: The user performing the search.
|
|
|
|
room_ids: List of room ids to search in
|
|
|
|
search_term: Search term to search for
|
|
|
|
keys: List of keys to search in, currently supports
|
|
|
|
"content.body", "content.name", "content.topic"
|
|
|
|
search_filter: The event filter to use.
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Returns:
|
|
|
|
A tuple of:
|
|
|
|
The search results.
|
|
|
|
A map of sender ID to results.
|
|
|
|
"""
|
|
|
|
rank_map = {} # event_id -> rank of event
|
|
|
|
# Holds result of grouping by room, if applicable
|
|
|
|
room_groups: Dict[str, JsonDict] = {}
|
|
|
|
# Holds result of grouping by sender, if applicable
|
|
|
|
sender_group: Dict[str, JsonDict] = {}
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
search_result = await self.store.search_msgs(room_ids, search_term, keys)
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if search_result["highlights"]:
|
|
|
|
highlights = search_result["highlights"]
|
|
|
|
else:
|
|
|
|
highlights = set()
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
results = search_result["results"]
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
# event_id -> rank of event
|
|
|
|
rank_map = {r["event"].event_id: r["rank"] for r in results}
|
2015-11-30 10:45:31 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
filtered_events = await search_filter.filter([r["event"] for r in results])
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
events = await filter_events_for_client(
|
|
|
|
self.storage, user.to_string(), filtered_events
|
|
|
|
)
|
2015-10-29 10:17:47 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
events.sort(key=lambda e: -rank_map[e.event_id])
|
|
|
|
allowed_events = events[: search_filter.limit]
|
2018-05-22 10:21:18 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
for e in allowed_events:
|
|
|
|
rm = room_groups.setdefault(
|
|
|
|
e.room_id, {"results": [], "order": rank_map[e.event_id]}
|
|
|
|
)
|
|
|
|
rm["results"].append(e.event_id)
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
s = sender_group.setdefault(
|
|
|
|
e.sender, {"results": [], "order": rank_map[e.event_id]}
|
|
|
|
)
|
|
|
|
s["results"].append(e.event_id)
|
|
|
|
|
|
|
|
return (
|
|
|
|
_SearchResult(
|
|
|
|
search_result["count"],
|
|
|
|
rank_map,
|
|
|
|
allowed_events,
|
|
|
|
room_groups,
|
|
|
|
highlights,
|
|
|
|
),
|
|
|
|
sender_group,
|
|
|
|
)
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
async def _search_by_recent(
|
|
|
|
self,
|
|
|
|
user: UserID,
|
|
|
|
room_ids: Collection[str],
|
|
|
|
search_term: str,
|
|
|
|
keys: Iterable[str],
|
|
|
|
search_filter: Filter,
|
|
|
|
batch_group: Optional[str],
|
|
|
|
batch_group_key: Optional[str],
|
|
|
|
batch_token: Optional[str],
|
|
|
|
) -> Tuple[_SearchResult, Optional[str]]:
|
|
|
|
"""
|
|
|
|
Performs a full text search for a user ordering by recent.
|
2018-05-22 10:26:46 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Args:
|
|
|
|
user: The user performing the search.
|
|
|
|
room_ids: List of room ids to search in
|
|
|
|
search_term: Search term to search for
|
|
|
|
keys: List of keys to search in, currently supports
|
|
|
|
"content.body", "content.name", "content.topic"
|
|
|
|
search_filter: The event filter to use.
|
|
|
|
batch_group: Pagination information.
|
|
|
|
batch_group_key: Pagination information.
|
|
|
|
batch_token: Pagination information.
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Returns:
|
|
|
|
A tuple of:
|
|
|
|
The search results.
|
|
|
|
Optionally, a pagination token.
|
|
|
|
"""
|
|
|
|
rank_map = {} # event_id -> rank of event
|
|
|
|
# Holds result of grouping by room, if applicable
|
|
|
|
room_groups: Dict[str, JsonDict] = {}
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
# Holds the next_batch for the entire result set if one of those exists
|
|
|
|
global_next_batch = None
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
highlights = set()
|
2015-11-20 04:39:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
room_events: List[EventBase] = []
|
|
|
|
i = 0
|
|
|
|
|
|
|
|
pagination_token = batch_token
|
|
|
|
|
|
|
|
# We keep looping and we keep filtering until we reach the limit
|
|
|
|
# or we run out of things.
|
|
|
|
# But only go around 5 times since otherwise synapse will be sad.
|
|
|
|
while len(room_events) < search_filter.limit and i < 5:
|
|
|
|
i += 1
|
|
|
|
search_result = await self.store.search_rooms(
|
|
|
|
room_ids,
|
|
|
|
search_term,
|
|
|
|
keys,
|
|
|
|
search_filter.limit * 2,
|
|
|
|
pagination_token=pagination_token,
|
|
|
|
)
|
2015-11-20 04:39:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if search_result["highlights"]:
|
|
|
|
highlights.update(search_result["highlights"])
|
|
|
|
|
|
|
|
count = search_result["count"]
|
|
|
|
|
|
|
|
results = search_result["results"]
|
|
|
|
|
|
|
|
results_map = {r["event"].event_id: r for r in results}
|
|
|
|
|
|
|
|
rank_map.update({r["event"].event_id: r["rank"] for r in results})
|
|
|
|
|
|
|
|
filtered_events = await search_filter.filter([r["event"] for r in results])
|
2018-10-30 11:33:41 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
events = await filter_events_for_client(
|
|
|
|
self.storage, user.to_string(), filtered_events
|
|
|
|
)
|
|
|
|
|
|
|
|
room_events.extend(events)
|
|
|
|
room_events = room_events[: search_filter.limit]
|
|
|
|
|
|
|
|
if len(results) < search_filter.limit * 2:
|
|
|
|
break
|
|
|
|
else:
|
|
|
|
pagination_token = results[-1]["pagination_token"]
|
|
|
|
|
|
|
|
for event in room_events:
|
|
|
|
group = room_groups.setdefault(event.room_id, {"results": []})
|
|
|
|
group["results"].append(event.event_id)
|
|
|
|
|
|
|
|
if room_events and len(room_events) >= search_filter.limit:
|
|
|
|
last_event_id = room_events[-1].event_id
|
|
|
|
pagination_token = results_map[last_event_id]["pagination_token"]
|
|
|
|
|
|
|
|
# We want to respect the given batch group and group keys so
|
|
|
|
# that if people blindly use the top level `next_batch` token
|
|
|
|
# it returns more from the same group (if applicable) rather
|
|
|
|
# than reverting to searching all results again.
|
|
|
|
if batch_group and batch_group_key:
|
|
|
|
global_next_batch = encode_base64(
|
|
|
|
(
|
|
|
|
"%s\n%s\n%s" % (batch_group, batch_group_key, pagination_token)
|
|
|
|
).encode("ascii")
|
|
|
|
)
|
|
|
|
else:
|
|
|
|
global_next_batch = encode_base64(
|
|
|
|
("%s\n%s\n%s" % ("all", "", pagination_token)).encode("ascii")
|
|
|
|
)
|
|
|
|
|
|
|
|
for room_id, group in room_groups.items():
|
|
|
|
group["next_batch"] = encode_base64(
|
|
|
|
("%s\n%s\n%s" % ("room_id", room_id, pagination_token)).encode(
|
|
|
|
"ascii"
|
2015-11-20 04:39:44 -07:00
|
|
|
)
|
2022-02-15 06:47:05 -07:00
|
|
|
)
|
2015-11-20 04:39:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
return (
|
|
|
|
_SearchResult(count, rank_map, room_events, room_groups, highlights),
|
|
|
|
global_next_batch,
|
|
|
|
)
|
2015-11-20 04:39:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
async def _calculate_event_contexts(
|
|
|
|
self,
|
|
|
|
user: UserID,
|
|
|
|
allowed_events: List[EventBase],
|
|
|
|
before_limit: int,
|
|
|
|
after_limit: int,
|
|
|
|
include_profile: bool,
|
|
|
|
) -> Dict[str, JsonDict]:
|
|
|
|
"""
|
|
|
|
Calculates the contextual events for any search results.
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Args:
|
|
|
|
user: The user performing the search.
|
|
|
|
allowed_events: The search results.
|
|
|
|
before_limit:
|
|
|
|
The number of events before a result to include as context.
|
|
|
|
after_limit:
|
|
|
|
The number of events after a result to include as context.
|
|
|
|
include_profile: True if historical profile information should be
|
|
|
|
included in the event context.
|
2015-10-14 02:49:00 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
Returns:
|
|
|
|
A map of event ID to contextual information.
|
|
|
|
"""
|
|
|
|
now_token = self.hs.get_event_sources().get_current_token()
|
2015-10-12 03:49:53 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
contexts = {}
|
|
|
|
for event in allowed_events:
|
|
|
|
res = await self.store.get_events_around(
|
|
|
|
event.room_id, event.event_id, before_limit, after_limit
|
2022-02-08 07:21:20 -07:00
|
|
|
)
|
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
logger.info(
|
|
|
|
"Context for search returned %d and %d events",
|
|
|
|
len(res.events_before),
|
|
|
|
len(res.events_after),
|
2019-05-09 06:21:57 -06:00
|
|
|
)
|
2022-02-15 06:47:05 -07:00
|
|
|
|
|
|
|
events_before = await filter_events_for_client(
|
|
|
|
self.storage, user.to_string(), res.events_before
|
2019-05-09 06:21:57 -06:00
|
|
|
)
|
2015-10-28 12:25:11 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
events_after = await filter_events_for_client(
|
|
|
|
self.storage, user.to_string(), res.events_after
|
|
|
|
)
|
2015-11-20 07:16:42 -07:00
|
|
|
|
2022-02-18 08:57:26 -07:00
|
|
|
context: JsonDict = {
|
2022-02-15 06:47:05 -07:00
|
|
|
"events_before": events_before,
|
|
|
|
"events_after": events_after,
|
|
|
|
"start": await now_token.copy_and_replace(
|
2022-05-16 09:35:31 -06:00
|
|
|
StreamKeyType.ROOM, res.start
|
|
|
|
).to_string(self.store),
|
|
|
|
"end": await now_token.copy_and_replace(
|
|
|
|
StreamKeyType.ROOM, res.end
|
2022-02-15 06:47:05 -07:00
|
|
|
).to_string(self.store),
|
|
|
|
}
|
2015-11-20 07:16:42 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if include_profile:
|
|
|
|
senders = {
|
|
|
|
ev.sender
|
|
|
|
for ev in itertools.chain(events_before, [event], events_after)
|
2019-05-09 06:21:57 -06:00
|
|
|
}
|
2015-10-09 08:48:31 -06:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
if events_after:
|
|
|
|
last_event_id = events_after[-1].event_id
|
|
|
|
else:
|
|
|
|
last_event_id = event.event_id
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
state_filter = StateFilter.from_types(
|
|
|
|
[(EventTypes.Member, sender) for sender in senders]
|
2019-05-09 06:21:57 -06:00
|
|
|
)
|
|
|
|
|
2022-05-25 05:59:04 -06:00
|
|
|
state = await self.state_storage.get_state_for_event(
|
2022-02-15 06:47:05 -07:00
|
|
|
last_event_id, state_filter
|
|
|
|
)
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
context["profile_info"] = {
|
|
|
|
s.state_key: {
|
|
|
|
"displayname": s.content.get("displayname", None),
|
|
|
|
"avatar_url": s.content.get("avatar_url", None),
|
|
|
|
}
|
|
|
|
for s in state.values()
|
|
|
|
if s.type == EventTypes.Member and s.state_key in senders
|
|
|
|
}
|
2015-11-04 10:57:44 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
contexts[event.event_id] = context
|
2015-11-05 07:34:37 -07:00
|
|
|
|
2022-02-15 06:47:05 -07:00
|
|
|
return contexts
|