This was doing bulk_create in a loop for each realm, which is too slow
on very large servers. Just do a single bulk_create with a reasonable
batch_size at the end.
This implements the core of the rewrite described in:
For the backend data model for UserPresence to one that supports much
more efficient queries and is more correct around handling of multiple
clients. The main loss of functionality is that we no longer track
which Client sent presence data (so we will no longer be able to say
using UserPresence "the user was last online on their desktop 15
minutes ago, but was online with their phone 3 minutes ago"). If we
consider that information important for the occasional investigation
query, we have can construct that answer data via UserActivity
already. It's not worth making Presence much more expensive/complex
to support it.
For slim_presence clients, this sends the same data format we sent
before, albeit with less complexity involved in constructing it. Note
that we at present will always send both last_active_time and
last_connected_time; we may revisit that in the future.
This commit doesn't include the finalizing migration, which drops the
UserPresenceOld table.
The way to deploy is to start the backfill migration with the server
down and then start the server *without* the user_presence queue worker,
to let the migration finish without having new data interfering with it.
Once the migration is done, the queue worker can be started, leading to
the presence data catching up to the current state as the queue worker
goes over the queued up events and updating the UserPresence table.
Co-authored-by: Mateusz Mandera <mateusz.mandera@zulip.com>
This removes the validator argument for 0423_realmfilter_url_template,
which do not really alter the database schema. It otherwise fails
the migration because the filter_format_validator function is removed.
Migration 0094_realm_filter_url_validator is modified because we can no
longer refer to filter_format_validator.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This swaps out url_format_string from all of our APIs and replaces it
with url_template. Note that the documentation changes in the following
commits will be squashed with this commit.
We change the "url_format" key to "url_template" for the
realm_linkifiers events in event_schema, along with updating
LinkifierDict. "url_template" is the name chosen to normalize
mixed usages of "url_format_string" and "url_format" throughout
the backend.
The markdown processor is updated to stop handling the format string
interpolation and delegate the task template expansion to the uri_template
library instead.
This change affects many test cases. We mostly just replace "%(name)s"
with "{name}", "url_format_string" with "url_template" to make sure that
they still pass. There are some test cases dedicated for testing "%"
escaping, which aren't relevant anymore and are subject to removal.
But for now we keep most of them as-is, and make sure that "%" is always
escaped since we do not use it for variable substitution any more.
Since url_format_string is not populated anymore, a migration is created
to remove this field entirely, and make url_template non-nullable since
we will always populate it. Note that it is possible to have
url_template being null after migration 0422 and before 0424, but
in practice, url_template will not be None after backfilling and the
backend now is always setting url_template.
With the removal of url_format_string, RealmFilter model will now be cleaned
with URL template checks, and the old checks for escapes are removed.
We also modified RealmFilter.clean to skip the validation when the
url_template is invalid. This avoids raising mulitple ValidationError's
when calling full_clean on a linkifier. But we might eventually want to
have a more centric approach to data validation instead of having
the same validation in both the clean method and the validator.
Fixes#23124.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This is implemented by replacing all matches of "%(var_name)s" in a URL
format string with "{var_name}". Since we do want to ensure that the
templates aren't broken after this migration, a RuntimeError is raised
to let the maintainer know that certain linkifier cannot be converted
automatically if it does not pass the uri_template.validate check.
Also, we need to escape "%%", which is used to represent "%" in the old
format string syntax, as well as "{" and "}", which is a part of the
URL template syntax.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This will later be used to expand matching linkifier patterns.
Making it nullable for now, but we will make it required in
the APIs.
As a part of this transition, we temporarily make url_format_string
nullable as well, which will be later removed. This allows us to
switch to populating url_template without caring about passing
url_format_string.
Note that the validators are imported in the migration because Django
otherwise diffs it and considers the schema to be different, generating
a migration, failing the "tools/test-migrations" test.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
208c0c3034 fixed this for new deleted users, but left existing users
with potentially invalid email addresses. This is problematic if the
realm is ever exported and re-imported, as the addresses will not
validate.
Add a migration which attempts to fix these invalid email addresses.
So far, we've used the BitField .authentication_methods on Realm
for tracking which backends are enabled for an organization. This
however made it a pain to add new backends (requiring altering the
column and a migration - particularly troublesome if someone wanted to
create their own custom auth backend for their server).
Instead this will be tracked through the existence of the appropriate
rows in the RealmAuthenticationMethods table.
This commit adds code to create a "Nobody" system user group
to realms which will be used in settings to represent "Nobody"
option.
We also add a migration to add this group to existing realms.
This commit adds PreregistrationRealm class which will be
similar to PreregistrationUser and will store initial
information of the realm before its creation as we are
changing the organization creation flow as per #24307.
Fixes part of #24307.
Currently, there is a checkbox setting for whether to
"Include realm name in subject of message notification emails".
This commit replaces the checkbox setting with a dropdown
having values: Automatic [default], Always, Never.
The Automatic option includes the realm name if, and only if,
there are multiple Zulip realms associated with the user's email.
Tests are added and(or) modified.
Fixes: #19905.
c7d0192755 added the unique constraint on
`user_profile_id,message_id,reaction_type,emoji_code`, but left the
existing constraint on `user_profile_id,message_id,emoji_name`. As
explained in the comment added in 3cd543ee98, `emoji_name` cannot be
trusted to be unique, as it is possible to have an Unicode emoji
reaction and a custom emoji with the same name on a message.
Remove the overly-constraining unique index, now that c7d0192755 has
provided the correct one.
This commit adds migration to fix extra_data field
of RealmAuditLog objects created on changing
can_remove_subscribers_group setting to add "property"
field since the same event type will now be used for
other group based stream settings that will be added
in future.
This will allow us to rely on the database to detect duplicate
`UserTopic`s (with the same `topic_name` with different cases)
and thus correctly throw IntegrityErrors when expected.
This is also important from a correctness point of view, since as
of now, when checking if topic is muted or requesting the backend for
muting a topic, the frontend does not check for case insensitivity.
There might exist duplicate UserTopics (in a case insensitive sense)
which need are removed before creating the new index.
The migration was tested manually using `./manage.py shell`.
Black 23 enforces some slightly more specific rules about empty line
counts and redundant parenthesis removal, but the result is still
compatible with Black 22.
(This does not actually upgrade our Python environment to Black 23
yet.)
Signed-off-by: Anders Kaseorg <anders@zulip.com>
608c787c52 fixed a bug where messages sent by the email gateway "as"
a user failed to properly attribute ownership of their attachments,
leaving the attachments orphaned and thus with nobody with permissions
to view them.
These orphaned attachments only remain longer than a few weeks if the
`delete_unclaimed_attachments` script has not been run reliably.
Since there is currently no shipped cron job for this, that is most
likely all deployments.
Add a migration to find such orphaned attachments, and re-attach them
to their original message. While theoretically the attachments
could have been later referenced in other messages -- which would be
very difficult to find and determine if they had access to the
attachment -- we only fix the original message.
In order to make this somewhat performant, we assume that the Message
rows associated with an Attachment made by the email gateway happened
within 5 minutes, since they must have been made during one HTTP
request.
This is complicated by the message potentially having been deleted; in
this case, the Attachment is moved into ArchivedAttachment, so it can
relate to the ArchivedMessage. The many-to-many
`zerver_archivedattachment_messages` relationship table cannot use its
own `id` sequence for the value, since the `id` is re-used when the
row is inserted into the `zerver_attachment_messages` table -- we
instead consume a value from the `id` sequence of the
`zerver_attachment_messages` table.
The intention was to continue the outer ‘for’ loop, not the inner one
(but Python doesn’t have labelled ‘continue’).
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This allows us to revoke MultiUseInvites by changing their .status
instead of deleting them (which has been deleting the helpful tracking
information on PreregistrationUsers about which MultiUseInvite they came
from).
This guarantees that the Realm is always non-None when we hit the
codepath is_static_or_current_realm_url via
do_change_stream_description, so that we can properly skip rewritting
some images.
Fixes#19405
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
We followed a same approach as in #22611 to mark migrations as noops. We
might eventually squash them.
Migration operations that only change the validators should be removed
as they are essentially noops that do not affect the database schema.
However, ./manage.py makemigration still generates a new migration for
validators change regardless. So we still have to keep one migration
that updates the validators to the latest state. We prefer to keep the
earliest one for this purpose.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This commit adds the OPTIONAL .realm attribute to Message
(and ArchivedMessage), with the server changes for making new Messages
have this set. Old Messages still have to be migrated to backfill this,
before it can be non-nullable.
Appropriate test changes to correctly set .realm for Messages the tests
manually create are included here as well.
zerver/migrations/0240_usermessage_migrate_bigint_id_into_id.py needs
to be updated to account for Django 4.1 creating AutoField as an
identity column rather than a serial column.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The sequence value should reflect the last id, not the next id, to
avoid leaving a gap of 1. Also, it should take ArchivedUserMessage.id
into account to avoid collisions during future archiving.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
SCIMClient is a type-unsafe workaround for django-scim2’s conflation
of SCIM users with Django users. Given that a SCIMClient is not a
UserProfile, it might as well not be a model at all, since it’s only
used to satisfy django-scim2’s request.user.is_authenticated queries.
This doesn’t solve the type safety issue with assigning a SCIMClient
to request.user, nor the performance issue with running the SCIM
middleware on non-SCIM requests. But it reduces the risk of potential
consequences worse than crashing, since there’s no longer a
request.user.id for Django to confuse with the ID of an actual
UserProfile.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
We are no longer writing to or reading the UserStatus.status field,
so we delete that from the model.
Fifth step in making user status `away` a deprecated way to access
`presence_enabled` for clients supporting older servers.
Part of transitioning from 'unavailable' user status feature to
'invisible mode' user presence feature.
Now that user status updates with `away=True|False` also update the
user's presence_enabled setting, we do a migration so that users with
`UserStatus.status=AWAY` also have the presence_enabled setting as
False (`away=!presence_enabled`).
Second step in making user status away a deprecated way to access
presence_enabled for clients supporting older servers.
Part of transitioning from 'unavailable' user status feature to
'invisible mode' user presence feature.
To allow `custom_profile_field` to display in user profile popover,
added new boolean field "display_in_profile_summary" in its model class.
In `custom_profile_fields.py`, functions are edited as per conditions,
like currently we can display max 2 `custom_profile_fields` except
`LONG_TEXT` and `USER` type fields.
Default external account custom profile fields made updatable for only
this new field, as previous they were not updatable.
Fixes part of: #21215
Updates the two UserProfile foreign key fields to have a backward
relation in the MutedUser model by changing the `related_name`
property.
This is a prep commit for removing users with a muted relationship
to the current user from read receipts.
This commit udpates can_remove_subscribers_group to be not null.
We already added a migration to set the value of this field for
existing streams and also added a commit to set this field to
admins system group for now while creating streams.
This migration sets can_remove_subscribers_group value to admins system
group for all the existing streams. In further commit we would change
can_remove_subscribers_group to be not null and thus we add this migration
to ensure all existing streams have this setting value set.
Added a user_list_style personal user setting to the bottom of
Settings > Display settings > Theme section which controls the look
of the right sidebar user list.
The radio button UI includes a preview of what the styles look like.
The setting is intended to eventually have 3 possible values: COMPACT,
WITH_STATUS and WITH_AVATAR; the final value is not yet implemented.
Co-authored-by: Tim Abbott <tabbott@zulip.com>
This also allows us to remove some assertions as we now know that
AVATAR_SALT will never be None.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This commit changes the code to consider zero as an invalid value for
message_content_edit_time_limit_seconds. Now to represent the setting that
user can edit the message anytime, the setting value will be "None" in
database and "unlimited" will be passed to API from clients.
Since validators do not affect the database, this migration is a noop.
Removing the migration fails check-database-compatibility. We might
eventually delete it when Django supports a cleaner method for deletion.
TODO:
Remove the type annotation when django-stubs is integrated.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
There may be some internal realms which were created after applying
"0382_create_role_based_system_groups.py" migration and this migration
is used to create system groups for those realms.
Because Django's ContentType objects are, by default, created lazily
when an actual object is created that will use them, this migration
would fail on any server that actually had RealmReactivationStatus
objects already, and had not yet created the ContentType for them.
ContentType objects are very simple:
zulip=> select * from django_content_type where model = 'realmreactivationstatus';
id | app_label | model
----+-----------+-------------------------
85 | zerver | realmreactivationstatus
So we can simply patch this by using get_or_create.
We now send a new user_topic event while muting and unmuting topics.
fetch_initial_state_data now returns an additional user_topics array to
the client that will maintain the user-topic relationship data.
This will support any future addition of new features to modify the
relationship between a user-topic pair.
This commit adds the relevent backend code and schema for the new
event.
Now that we can assume Python 3.6+, we can use the
email.headerregistry module to replace hacky manual email address
parsing.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
A user ran into an issue while upgrading where
ContentType.objects.get(model="realmreactivationstatus",
app_label="zerver") fails due to the object being missing. The reason
for that is to be yet figured out, but the immediate solution is clear
in the sense that the migration can just quit early
if not Confirmation.objects.filter(type=REALM_REACTIVATION).exists() and
that'll effectively skip it for almost all servers (because realm
reactivations links are something that's really only useful on Zulip
Cloud).
Fixes#21266.
We want to tie the prereg_user to the MultiUseInvite directly rather
than to the MultiUserInvite's confirmation object, because the latter is
not possible. This is because the flow is that after going through the
multiuse invite link, the PreregistrationUser is created together with a
Confirmation object, creating a confirmation link (via
create_confirmation_link) to which then the user is redirected to finish
account creation. This means that the PreregistrationUser is already
tied to a Confirmation, so that attribute is occupied.
PostgreSQL's `default_statistics_target` is used to track how many
"most common values" ("MCVs") for a column when performing an
`ANALYZE`. For `tsvector` columns, the number of values is actually
10x this number, because each row contains multiple values for the
column[1]. The `default_statistics_target` defaults to 100[2], and
Zulip does not adjust this at the server level.
This translates to 1000 entries in the MCV for tsvectors. For
large tables like `zerver_messages`, a too-small value can cause
mis-planned query plans. The query planner assumes that any
entry *not* found in the MCV list is *half* as likely as the
least-likely value in it. If the table is large, and the MCV list is
too short (as 1000 values is for large deployments), arbitrary
no-in-the-MCV words will often be estimated by the query planner to
occur comparatively quite frequently in the index. Based on this, the
planner will instead choose to scan all messages accessible by the
user, filtering by word in tsvector, instead of using the tsvector
index and filtering by being accessible to the user. This results in
degraded performance for word searching.
However, PostgreSQL allows adjustment of this value on a per-column
basis. Add a migration to adjust the value up to 10k for
`search_tsvector` on `zerver_message`, which results in 100k entries
in that MCV list.
PostgreSQL's documentation says[3]:
> Raising the limit might allow more accurate planner estimates to be
> made, particularly for columns with irregular data distributions, at
> the price of consuming more space in `pg_statistic` and slightly
> more time to compute the estimates.
These costs seem adequate for the utility of having better search.
In the event that the pgroonga backend is in use, these larger index
statistics are simply wasted space and `VACUUM` computational time,
but the costs are likely still reasonable -- even 100k values are
dwarfed by the size of the database needed to generate 100k unique
entries in tsvectors.
[1]: https://github.com/postgres/postgres/blob/REL_14_4/src/backend/utils/adt/array_typanalyze.c#L261-L267
[2]: https://www.postgresql.org/docs/14/runtime-config-query.html#GUC-DEFAULT-STATISTICS-TARGET
[3]: https://www.postgresql.org/docs/14/planner-stats.html#id-1.5.13.5.3
This commit removes "role" field from subscription
objects since we are not moving forward with stream
administrator concept and instead working on new
permssions model as per #19525.
This commit removes WILDCARD_MENTION_POLICY_STREAM_ADMINS
option of wildcard_mention_policy since we are not moving
forward with stream administrator concept and instead working
on new permssions model as per #19525.
We also add a migration to change wildcard_mention_policy of
existing realms to WILDCARD_MENTION_POLICY_ADMINS. This change
is fine since we were already treating both the setting values
as same as stream admin concept was not implemented completely.
This was removed in Django 4.0 except in historical migrations. We
might as well replace it everywhere.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Adds `want_advertise_in_communities_directory` to the realm model
to track organizations that give permission to be listed on such
a site / directory on zulip.com.
Adds a checkbox to the organization profile admin for
organizations to give permission to be advertised in the
Zulip communities directory.
Adds a help center article about the Zulip communities directory
and uses a shared intro documentation file to create sections in
the articles on creating an organization profile and moderating
open organizations.
Co-authored-by: Alya Abbott <alya@zulip.com>
The default of a Stream is to be public - having
history_public_to_subscribers default to False is inconsistent with
that. The defaults on the model should generally be consistent.
history_public_to_subscribers wasn't explicitly set when creating
streams via build_stream, thus relying on the model's default of False.
This lead to public streams being created with that value set to False,
which doesn't make sense.
We can solve this by inferring the correct value based on invite_only in
the build_stream funtion itself - rather than needing to add a flag
argument to it.
This commit also includes a migration to fix public stream with the
wrong history_public_to_subscribers value.
Fixes#21784.
Added a setting to the bottom of Settings > Display settings > Theme section
to display the reacting users on a message when numnber of reactions are
small.
This is a preparatory commit for #20980.
This is necessary for the migration 0386_fix_attachment_caches to run,
and likely makes more convenient any future parallel code interacting
with both Attachment and ArchivedAttachment.
This migration needs to be run after the previous commit is deployed
to a given Zulip installation, to fix any stale values of
is_realm_public and is_web_public.
Previously, Attachment.is_realm_public and its cousin,
Attachment.is_web_public, were properties that began as False and
transitioned to True only when a message containing a link to the
attachment was sent to the appropriate class of stream, or such a link
was added as part of editing a message.
This pattern meant that neither field was updated in situations where
the access permissions for a message changed:
* Moving the message to a different stream.
* Changing the permissions for a stream containing links to the message.
This correctness issue has limited security impact, because uploaded
files are secured both by a random URL and by these access checks.
To fix this, we reformulate these fields as a cache, with code paths
that change the permissions affecting an attachment responsible for
setting these values to the `None` (uncached) state. We prefer setting
this `None` state over computing the correct permissions, because the
correct post-edit permissions are a function of all messages
containing the attachment, and we don't want to be responsible for
fetching all of those messages in the edit code paths.
* Don't print the empty list for the vast majority of realms where
this is a noop.
* Make output a little more clear that this isn't revoking all
Confirmations, just those associated with deactivated users.
Mateusz Mandera
Alex Vandiver
Tim Abbott
Zixuan James Li
Joelute
Aman Agrawal
Sahil Batra
Prakhar Pratyush
Anders Kaseorg
Abhijeet Prasad Bodas
PIG208
evykassirer
Lauryn Menard
yogesh sirsat
Raghav Luthra
Kartik Srivastava
Zixuan James Li