Earlier, the notification-blocking for messages from muted senders
was a side-effect of we never sending notifications for messages
with the "read" flag.
This commit decouples these two things, as a prep for having new
settings which will allow users to **always** receive email
notifications, including when/if they read the message during the
time the notifications is in the queue.
We still mark muted-sender messages as "read" when they are sent,
because that's desirable anyways.
This is a prep change for importing (and using) `dataclasses.field`
elsewhere in the same file, because pyflakes would throw "Import
module shodowed" errors otherwise.
This locks the message row while a reaction is being added/removed,
which will handle race conditions caused by deleting the message
at the same time.
We make sure that events work happens outside the transaction,
so that in case there's some problem with the queue processor, the
locks aren't held for too long.
As a nice side-effect, we also handle race conditions from double
adding reactions, because once the message is locked, a duplicate
request will wait till the earlier transaction commits, and hence
will not throw `IntegrityErrors`s (rather, will be handled in our
safety check in the /views code itself), which earlier had to be
handled explicitly.
This locks the message while creating a submessage, which
will handle race conditions caused by deleting the message
simultaneously.
We make sure that events work happens outside the transaction,
so that in case there's a problem with the queue processor,
the locks aren't held for too long.
Further commits will start locking the message rows while
adding related fields like reactions or submessages,
to handle races caused by deleting the message itself at the
same time.
The message locking implemented then will create a possibility
of deadlocks, where the related field transaction holds a lock
on the message row, and the message-delete transaction holds a
lock on the database row of the related field (which will also
need to be deleted when the message is deleted), and both
transactions wait for each other.
To prevent such a deadlock, we lock the message itself while
it is being deleted, so that the message-delete transaction
will have to wait till the other transaction (which is about
to delete the related field, and also holds a lock on the
message row) commits.
https://chat.zulip.org/#narrow/near/1185943 has more details.
This commit fixes a bug where moving messages between streams was
not allowed for non-admins when allow_community_topic_editing was
set to false and move_messages_between_streams_policy was set to
Realm.POLICY_MEMBERS_ONLY.
The bug is fixed by calling can_edit_content_or_topic only when
topic or content edit is there and not in the case where only
message is moved from one stream to another.
This commit extracts the logic of checking the message edit permissions,
like whether the sender is same as user, whether it is a (no topic)
message or whether community topic editing is allowed, into a separate
function.
This is a prep commit for fixing a bug where permission to move messages
between streams is affected by permission of editing topics.
Previously when enforcing the check to do not allow editing topics
after a certain time, we were checking whether 'content is None' and
considering it as that if content is None then there must be topic
edit.
But after adding support for moving messages between streams it can be
the case that we are neither changing topic nor content and just moving
streams, and the original code raises error if this is done after the
time limit of editing topics, which is wrong.
This commit fixes this by actually checking 'topic_name is not None'.
This will offer users who are self-hosting to adjust
this value. Moreover, this will help to reduce the
overall time taken to test `test_markdown.py` (since
this can be now overridden with `override_settings`
Django decorator).
This is done as a prep commit for #18641.
d66cbd2832 added these mentioning
"always_notify" for some reason, but always_notify clearly isn't a real
thing in this context so the comments need to be fixed to eliminate this
potential source of confusion.
These checks are more related to the API than the editability
or permissions logic, so it makes sense to handle them first
before further processing the request.
Also split the main test class to separate out the tests for
this logic.
This also simplifies some tests by reducing the data setup
required to reach failure.
Tweaked by tabbott to avoid losing the topic_name.strip().
modified_user=sub_info.user and modified_stream=sub_info.stream, added
by commit 6d1f9de7d3 (#16553), were
always coming from the last entry in the loop above, not from the
enclosing list comprehension.
Found by the Pylint rule undefined-loop-variable.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The old name `push_notify_user_ids` was misleading, because
it does not contain user ids which should be notified for
the current message, but rather user ids who have the online
push notifications setting enabled.
When the Tornado server is restarted during an upgrade, if
server has old events with the `push_notify_user_ids` fields,
the server will throw error after this rename. Hence, we need
to explicitly handle such cases while processing the event.
We should only show the referrer name in subject of invitation emails,
and show only 'Zulip' in the 'From' header. This helps in preventing
the email from being marked as suspicious by the detection systems
when they see an employee's name as sender of an email sent from an
unrelated domain.
The behavior is already the same for reminder invitation emails where
we do not show name and only 'Zulip' in the 'From' header.
Fixes#18256.
Commit 1a7ddd9ea3 “Fix
UserActivityInterval overlap bug” introduced a mathematically
incorrect assertion about how intervals work. There’s a third way two
intervals could overlap: both the start and end of the old interval
could be inside the new interval. This probably can’t happen here
because the old interval should be at least as long as the new
interval. However, a correct overlap test can be formulated in a
simpler way anyway.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Convert this function that absolutely makes a stream web public.
We already have do_change_stream_invite_only to convert
streams to public and private streams.
We also update all the fields that should be set when a stream
is made web public.
After re-assignment, mypy will still think the type of
`widget_content` to be `str`, not `Dict`. So we need to
create a new variable.
This is a prep change for stronger type checking in this
code.
Previously only admins were allowed to move messages between streams
and admins are allowed to post in any stream irresepctive of stream
post policy, so there was no need to check for stream post policy.
But as we now allow other members to also move messages, we need
to check whether the user who is moving the message is allowed
to post to the target stream (i.e. stream to which the messages
are being moved) and thus we allow moving messages only if the
user is allowed to post in target stream.
This new function optimizes how we fetch subscriptions
for streams. Basically, it excludes most long-term-idle
users from the query.
With 8k users, of which all but 400 are long term idle,
this speeds up get_recipient_info from about 150ms
to 50ms.
Overall this change appears to save a factor of 2-3 in the backend
processing time for sending or editing a message in large, public
streams in chat.zulip.org (at 18K users today).
We combine the two loops into one, so that we
can check our flags before creating the
UserMessageList object.
And we lift a few calculations out of the loop.
For 8k users, with 95% long-term-idle, this was
about a 10x speedup for me. (~30ms -> 3ms)
As discussed in the comment, this is a critical scalability
optimization for organizations with thousands of users.
With substantial comment updates by tabbott.
This extends the /json/typing endpoint to also accept
stream_id and topic. With this change, the requests
sent to /json/typing should have these:
* `to`: a list set to
- recipients for a PM
- stream_id for a stream message
* `topic`, in case of stream message
along with `op`(start or stop).
On receiving a request with stream_id and topic, we send
typing events to clients with stream_typing_notifications set
to True for all users subscribed to that stream.
Remove content edit keys if present in edit_history_event
when passing to update_messages_for_topic_edit.
Since content edit is only applied to the edited_message,
this shouldn't be part of the rest of the messages for which
topic was edited. This was a bug identified by
editing topic and content of a message at the same time
when more than 1 message is affected.
model__id syntax implies needing a JOIN on the model table to fetch the
id. That's usually redundant, because the first table in the query
simply has a 'model_id' column, so the id can be fetched directly.
Django is actually smart enough to not do those redundant joins, but we
should still avoid this misguided syntax.
The exceptions are ManytoMany fields and queries doing a backward
relationship lookup. If "streams" is a many-to-many relationship, then
streams_id is invalid - streams__id syntax is needed. If "y" is a
foreign fields from X to Y:
class X:
y = models.ForeignKey(Y)
then object x of class X has the field x.y_id, but y of class Y doesn't
have y.x_id. Thus Y queries need to be done like
Y.objects.filter(x__id__in=some_list)
This was a mostly harmless bug, since those users cannot have active
clients, but fixing it will improve performance in any Zulip
organization where the vast majority of users are deactivated.
This commit adds an API to `zproject/urls.py` to edit/update
the realm linkifier. Its helper function to update the
database is added in `zerver/lib/actions.py`.
`zulip.yaml` is documented accordingly as well, clearly
stating that this API updates one linkifier at a time.
The tests are added for the API and helper function which
updates the realm linkifier.
Fixes#10830.
Fixes#17922.
These two places fetch subscriptions for the sake of getting user ids to
send events to. Clearly deactivated users should be excluded from that.
get_active_subscriptions_for_stream_id should allow specifying whether
subscriptions of deactivated users should be included in the result.
Active subs of deactivated users are a subtlety that's easy to miss
when writing relevant code, so we make include_deactivated_users a
mandatory kwarg - this will force callers to definitely give thought to
whether such subs should be included or not.
This commit is just a refactoring, we keep original behavior everywhere
- there are places where subs of deactivates users should probably be
excluded but aren't - we don't fix that here, it'll be addressed in
follow-up commits.
The bulk deletion codepath was using dicts instead of user ids in the
event, as opposed to the other codepath which was adjusted to pass just
user ids before. We make the bulk codepath consistent with the other
one. Due to the dict-type events happening in 3.*, we move the goal for
deleting the compat code in process_notification to 5.0.
django.utils.translation.ugettext is a deprecated alias of
django.utils.translation.gettext as of Django 3.0, and will be removed
in Django 4.0.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
A bug in the implementation of replies to messages sent by outgoing
webhooks to private streams meant that an outgoing webhook bot could be
used to send messages to private streams that the user was not intended
to be able to send messages to.
Completely skipping stream access check in check_message whenever the
sender is an outgoing webhook bot is insecure, as it might allow someone
with access to the bot's API key to send arbitrary messages to all
streams in the organization. The check is only meant to be bypassed in
send_response_message, where the stream message is only being sent
because someone mentioned the bot in that stream (and thus the bot
posting there is the desired outcome). We get much better control over
what's going by passing an explicit argument to check_message when
skipping the access check is desirable.
* This introduces a new event type `realm_linkifiers` and
a new key for the initial data fetch of the same name.
Newer clients will be expected to use these.
* Backwards compatibility is ensured by changing neither
the current event nor the /register key. The data which
these hold is the same as before, but internally, it is
generated by processing the `realm_linkifiers` data.
We send both the old and the new event types to clients
whenever the linkifiers are changed.
Older clients will simply ignore the new event type, and
vice versa.
* The `realm/filters:GET` endpoint (which returns tuples)
is currently used by none of the official Zulip clients.
This commit replaces it with `realm/linkifiers:GET` which
returns data in the new dictionary format.
TODO: Update the `get_realm_filters` method in the API
bindings, to hit this new URL instead of the old one.
* This also updates the webapp frontend to use the newer
events and keys.
When a user is muted, in the same request,
we mark any existing unreads from that user
as read.
This is done for all types of messages
(PM/huddle/stream) and regardless of whether
the user was mentioned in them.
This will not break the unread count logic
of the web frontend, because that algorithm
decides which messages to mark as read based
only on the pointer location and the whitespace
at the bottom, not on what messages have already
been marked as read.
Messages sent by muted users are marked as read
as soon as they are sent (or, more accurately,
while creating the database entries itself), regardless
of type (stream/huddle/PM).
ede73ee4cd, makes it easy to
pass a list to `do_send_messages` containing user-ids for
whom the message should be marked as read.
We add the contents of this list to the set of muter IDs,
and then pass it on to `create_user_messages`.
This benefits from the caching behaviour of `get_muting_users`
and should not cause performance issues long term.
The consequence is that messages sent by muted users will
not contribute to unread counts and notifications.
This commit does not affect the unread messages
(if any) present just before muting, but only handles
subsequent messages. Old unreads will be handled in
further commits.
This makes it so that RealmAuditLog entries are
created when a user mutes/unmutes someone.
We don't really need to store the time, but we
do so anyways, because the `event_time` field
is currently a non-nullable one in the `RealmAuditLog`
model, and making it nullable would risk allowing
not specifying the time in other more important
code which also creates `RealmAuditLog` entries.
This also fixes an incorrect test of successfully
unmuting with the API. Earlier it did not mock
the time in the `views/muting.py` code to return
`mute_time`.
Previously, when unmuting a user, we used to make
two database fetches - one to verify that the user
is has been muted before, and one while actually
unmuting the user.
This reduces that to one, by passing around the
`MutedUser` object fetched in the first round.
Since the new function returns `Optional[MutedUser]`,
we need to use a hack for events tests, because
mypy does not yet use the type inferred from
`assert foo is not None` in nested functions like lambdas.
See python/mypy@8780d45507.
This commit implements a subtle optimization (described in more detail
in the comment) that can save a few hundred milliseconds in when the
sender sees that their message has sent when sending to very large
streams.
Fixes#17898.
We send the whole data set as a part of the event rather than
doing an add/remove operation for couple of reasons:
* This would make the client logic simpler.
* The playground data is small enough for us to not worry
about performance.
Tweaked both `fetch_initial_state_data` and `apply_events` to
handle the new playground event.
Tests added to validate the event matches the expected schema.
Documented realm_playgrounds sections inside /events and
/register to support our openapi validation system in test_events.
Tweaked other tests like test_event_system.py and test_home.py
to account for the new event being generated.
Lastly, documented the changes to the API endpoints in
api/changelog.md and bumped API_FEATURE_LEVEL.
Tweaked by tabbott to add an `id` field in RealmPlayground objects
sent to clients, which is essential to sending the API request to
remove one.
Similar to the previous commit, we have added a `do_*` function
which does the deletion from the DB. The next commit handles sending
the events when both adding and deleting a playground entry.
Added the openAPI format data to zulip.yaml for DELETE
/realm/playgrounds/{playground_id}. Also added python and curl
examples to remove-playground.md.
Tests added.
This endpoint will allow clients to create a playground entry
containing the name, pygments language and url_prefix for the
playground of their choice.
Introduced the `do_*` function in-charge of creating the entry in
the model. Handling the process of sending events which will be
done in a follow up commit.
Added the openAPI format data to zulip.yaml for POST
/realm/playgrounds. Also added python and curl examples for using
the endpoint in its markdown documented (add-playground.md).
Tests added.
Adds backend code for the mute users feature.
This is just infrastructure work (database
interactions, helpers, tests, events, API docs
etc) and does not involve any behavioral/semantic
aspects of muted users.
Adds POST and DELETE endpoints, to keep the
URL scheme mostly consistent in terms of `users/me`.
TODOs:
1. Add tests for exporting `zulip_muteduser` database table.
2. Add dedicated methods to python-zulip-api to be used
in place of the current `client.call_endpoint` implementation.
Previously, if a user subscribed to a stream with
history_public_to_subscribers, and then was looking at old messages in
the stream, they would not get live-updates for that stream, because
of the structure in how notify_reaction_update only looked at
UserMessage rows (we had a previous workaround involving the
`historical` field in `UserMessage` which had already made it work if
the user themselves added the reaction).
We fix this by including all subscribers with history access in the
set of recipients for update events.
Fixes a bug that was confused with #16942.
Adding an additional `!` to the stream name each time a stream is
deactivated, to a maximum of 21 times, effectively limits number of
times a stream with a given name can be deactivated. This is unlikely
to come up in common usage, but may be confusing when testing.
Change what we prepend to deactivated stream names to something with
more entropy than just `!`, by instead prepending a substring of hash
of the stream's ID. `!`s. Using 128 bits of the hash means that it
will require more than 10^18th renames to have a 1% chance of collision.
Because too-long stream names are also truncated at 60 characters,
having this entropy in the beginning of the name also helps address
potential issues from stream names that differed only in, e.g. the
60th character.
Fixes#17016.
Instead of validating `op` value later, this commit does that
in `REQ`.
Also helps avoiding duplication of this validation when
stream typing notifications feature is added.
With the previous two commits deployed, we're ready to use the
denormalization to optimize the query.
With dev environment db prepared using
./manage.py populate_db --extra-users=2000 --extra-streams=400
this takes the execution time of the query in
bulk_get_subscriber_user_ids from 1.5-1.6s to 0.4-0.5s on my machine.
The new comment explains the issue in some detail, but basically if we
deactivate the bots first, then an error partway through is corrected
by a retry; if we deactivate the user first, then we may leak
undeactivated bots if a failure occurs.
This adds the is_user_active with the appropriate code for setting the
value correctly in the future. In the following commit a migration to
backfill the value for existing Subscriptions will be added.
To ensure correct user_profile.is_active handling also in tests, we
replace all direct .is_active mutation with calls to appropriate
functions.
These procedures should be done atomically overall, with the exception
of the code that sends events to avoid block if there's a delay
communicating with Tornado.
We add the savepoint=False on underlying function that already
executes inside an atomic context - to avoid the overhead of creating
savepoints where they aren't needed.
* `op` (operation) field, added in f6fb88549f, was never intended for
`custom_profile_fields` event. This commit removes the `op` as it doesn't
have any use in the code.
* As a part of cleanup, this also eliminates the schema check warnings
for `custom_profile_fields` event, mentioned in #17568.
Emails are not unique, so we can only sensibly cache using keys formed
with both email and realm.
This requires adding a new cache key function for caching by delivery
email - user_profile_delivery_email_cache_key.
This add the schema checker, openapi schema, and also a test for
realm/deactivated event.
With several block comments by tabbott explaining the logic behind our
behavior here.
Part of #17568.
This is a prep commit which modifies the
`send_message_moved_breadcrumbs` function to take
message strings as input.
This is done to reuse the function in other places
like the /digress command.
Fixes#17238.
In process_new_human user, the queries were wrong, revoking all invites
sent to the email address, even in other realms than the one where the
new account just got created.
This commmit includes ROLE_MODERATOR in realm_user_count_by_role.
We also update test_change_role in test_audit_log.py to include
changes for moderator role as well.
This is a minor refactor which renames the
notify_topic_moved_streams function to
send_message_moved_breadcrumbs.
This is done because this function will be also used
for other things in the future, when moving streams
or when using the /digress command, for example.
c2526844e9 removed the `signups` queue
worker, and the command-line tool that enqueues to it -- but not the
automated process that enqueues during signups itself.
Remove the signup, since it is no longer in use.
We eliminate some redundant checks.
We also consistently provide a `subscribers` field
in our stream data with `[]`, even if our users
can't access subscribers. We therefore bump
the API version and tweak the docs. (See further
down for a detailed justification of the change.)
Even though it is sometimes fine to have redundant code
that is defensive in nature, some upcoming changes are gonna
move subscriber-related logic out of build_stream_dict_for_sub
for certain codepaths as part of our effort to streamline
the payload for subscribers within page_params.
So we can't rely on the code that I removed here
inside of build_stream_dict_for_sub.
Anyway, it makes more sense to do these checks explicitly
in the validate function.
The code in build_stream_dict_for_sub was almost effectively
a noop, since the validation function was already preventing
us from getting subscriber info. The only difference it
made was sometimes converting `[]` to `None`, and then
subsequently omitting the subscribers field.
Neither ZT nor the webapp make any distinction between
`[]` or <missing key> for the `subscribers` data in
`page_params`.
The webapp has had this code for a long time (and now
equivalent code elsewhere in this PR):
if (!Object.prototype.hasOwnProperty.call(sub, "subscribers")) {
sub.subscribers = new LazySet([]);
}
The webapp calculates access based on booleans, anyway:
sub.can_access_subscribers =
page_params.is_admin || sub.subscribed ||
(!page_params.is_guest && !sub.invite_only);
And ZT would choke if `subscribers` were missing, except that
it never gets to the relevant code due to other checks:
def get_other_subscribers_in_stream(<snip>):
assert stream_id is not None or stream_name is not None
if stream_id:
assert self.is_user_subscribed_to_stream(stream_id)
return [sub
for sub in self.stream_dict[stream_id]['subscribers']
if sub != self.user_id]
else:
return [sub
for _, stream in self.stream_dict.items()
for sub in stream['subscribers']
if stream['name'] == stream_name
if sub != self.user_id]
You could make a semantic argument that we should prefer
<missing key> to `[]` when subscribers aren't even available, but
we have precedent from the way that `bulk_get_subscriber_user_ids`
has traditionally populated its result:
result: Dict[int, List[int]] =
{stream["id"]: [] for stream in stream_dicts}
If we changed `stream_dicts` to `target_stream_dicts` we
would faciliate a move toward `None`, but it would just cause
headaches for other server code as well as the frontends
(which, to reiterate, already prefer the empty array
for convenience).
As my comment indicates, I would prefer to handle
this explicitly by raising JsonableError in an
else statement here, but it's not a big deal.
This function can probably be simplified with a
bit of work, mostly on the testing side to make
sure we are covering all edge cases, but that
is out of the scope of my current PR.
Abhijeet Prasad Bodas
Steve Howell
akshatdalton
Vishnu KS
sahil839
Mateusz Mandera
Anders Kaseorg
Aman Agrawal
Tim Abbott
PIG208
Dinesh
shanukun
Signior-X
Sumanth V Rao
LoopThrough-i-j
Siddharth Asthana
WookieMonkeys
tushar912
Alex Vandiver